Purification of Mlc and Analysis of Its Effects on thepts Expression in Escherichia coli *

Products of the pts operon ofEscherichia coli have multiple physiological roles such as sugar transport, and the operon is controlled by two promoters, P0 and P1. Expression of the pts P0 promoter that is increased during growth in the presence of glucose is also activated by cAMP receptor protein·cAMP. Based on the existence of a sequence that has a high similarity with the known Mlc binding site in the promoter, the effects of the Mlc protein on the pts P0 promoter expression were studied. In vivo transcription assays using wild type and mlc-negative E. coli strains grown in the presence and absence of glucose indicate that Mlc negatively regulates expression of the P0 promoter, and Mlc-dependent repression is relieved by glucose in the growth medium. In vitro transcription assay using purified recombinant Mlc showed that Mlc repressed transcription from the P0 but did not affect the activity of the P1. DNase I footprinting experiments revealed that a Mlc binding site was located around +1 to +25 of the promoter and that Mlc inhibited the binding of RNA polymerase to the P0 promoter. Cells overexpressing Mlc showed a very slow fermentation rate compared with the wild type when grown in the presence of various phosphoenolpyruvate-carbohydrate phosphotransferase system sugars but few differences in the presence of non-phosphoenolpyruvate-carbohydrate phosphotransferase system sugars except maltose. These results suggest that the pts operon is one of major targets for the negative regulation by Mlc, and thus Mlc regulates the utilization of various sugars as well as glucose inE. coli. The possibility that the inducer of Mlc may not be sugar or its derivative but an unknown factor is proposed to explain the Mlc induction mechanism by various sugars.

The phosphoenolpyruvate-carbohydrate phosphotransferase system (PTS) 1 catalyzes the phosphorylation and transportation of a large number of its sugar substrates (1). The phosphoryl group from phosphoenolpyruvate is sequentially transferred to enzyme I, to HPr, to the sugar-specific enzyme II complex, and finally to the substrate as it is translocated across the membrane. The ptsH, ptsI, and crr genes encoding HPr, enzyme I, and enzyme IIA Glc , respectively, constitute the pts operon. The pts operon is regulated in a complex fashion by P0 and P1, each of which has two different transcription start sites depending on the availability of CRP⅐cAMP and on the topology of DNA (2). Transcription of the pts promoters increases in the presence of CRP⅐cAMP as well as glucose in vivo (3,4). The glucose-and cAMP-mediated activations occur independently of each other (2,4,5), even though glucose lowers the level of CRP and cAMP in the cell (6). It has been known that the target promoter that is activated when cells are grown in the presence of glucose is the P0 promoter, but the mechanism by which glucose activates the P0 transcription is not known. Transcription from the P0 is fully activated only when cells grow in the presence of both glucose and exogenous cAMP (2,4,7). The possibility of the presence of a repressor that is sensitive to glucose has been suggested based on these observations (7).
Mlc (making large colonies) is a newly identified global regulator of carbohydrate metabolism (8 -11). The mlc gene was originally found to cause the reduction of acetate accumulation when the overexpressing cells grow in the presence of glucose (12). The mlc gene was also shown to be identical with the previously characterized dgsA gene (10,13,14). All genes that are so far known to be under the control of Mlc are also regulated by CRP⅐cAMP. We noticed a sequence that has a high homology with the known Mlc binding site between ϩ1 and ϩ25 of the pts P0 promoter region (Fig. 1). In this study, a procedure is described for purification of the active form of Mlc to homogeneity, and the effects of Mlc on the expression of the pts P0 promoter were studied to determine whether the in vivo glucose effect on pts operon expression is mediated by Mlc. Bacterial Strains-SR702 (MC4100, suhX1) is a derivative of MC4100 (araD139 ⌬argF-lacU169rpsL150 thiA1 relA1 flbB5301 deoC1 ptsF25 rbsR) that expresses a high level of GroEL because of an IS1 element inserted upstream of the groE gene. It was constructed from KY1603 (15) by P1 transduction. The chaperonin GroEL protects RNA from nucleases (16) so that the unstable mRNA from the P0 promoter (17) is stabilized in suhX1 strain background because of a high level of GroEL. SR703 (SR702, mlc) was constructed from CP1036 by P1 transduction. Glutamine at residue 369 of Mlc is replaced by a stop codon (UAG) in CP1036 (18).

Materials-Cyclic
Primer Extension-Cells were grown in tryptone broth (1% Bacto tryptone, 0.8% NaCl) with or without 0.2% glucose. Total Escherichia * Financial support was received from the Korea Research Foundation made in the program year 1998. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
§ Contributed equally to this work. coli RNA was purified using Trizol reagent (Life Technologies, Inc.). RNA was resuspended in sterile distilled water. Purified ␥-32 P-end labeled primer P11 (5Ј-GCCAGTTTTTAACAGACGCGACGCACGAAG-3Ј) (7) was coprecipitated with 30 g of total cell RNA, and the pellet was resuspended in 20 l of 250 mM KCl, 2 mM Tris-HCl, pH 7.9, and 0.2 mM EDTA. Primer extension reactions were done as described by Ryu and Garges (2).
Construction of Expression Vector pNS100 and Overexpression of Recombinant Mlc-The DNA sequence from base 194 to 1490 of the E. coli mlc gene (GenBank TM accession code D32222) was amplified by polymerase chain reaction. The forward polymerase chain reaction primer (5Ј-GCGAAAATATAGGGAGTATCATATGGTTGC-3Ј) contained an engineered NdeI site (underlined) containing the ATG start codon (boldface, the original GTG was changed into ATG for better expression). The reverse polymerase chain reaction primer (5Ј-CTT-TCCTGGCCCAAATTGGGATCCCGCAAA-3Ј) contained an engineered BamHI site (underlined) located 33 base pairs downstream from the TAA stop codon. The polymerase chain reaction product was cloned into the NdeI and BamHI cloning sites of the vector pRE1 (19), and the sequence of the recombinant plasmid, pNS100, was verified by a cycle sequencing kit. The pNS100 was electroporated with an E. coli pulser (Bio-Rad) into E. coli strain GI698. In plasmid pNS100, the mlc gene is under the control of the strong P L promoter-cII ribosome binding site combination. The cI repressor gene is under the control of the trp promoter in GI698 (20). E. coli GI698 transformed with plasmid pNS100 was cultured at 30°C in 200 ml of synthetic medium (20) supplemented with 0.1 mg/ml ampicillin. When the culture reached an A 600 ϭ 0.5, tryptophan (0.1 mg/ml) was added to induce the expression of Mlc. The cells harvested 20 h after induction were washed once with 20 mM Tris-HCl, pH 7.5, containing 50 mM NaCl and stored at Ϫ20°C.
Phenotype of E. coli Overexpressing Mlc-Fermentation properties of E. coli GI698 strains transformed with pNS100 or pRE1 were compared with that of the mlc-negative mutant to check the effect of Mlc overexpression or deletion on the PTS activity. These strains were grown in the synthetic medium supplemented with ampicillin and then spotted onto MacConkey indicator plates containing various sugars as carbon sources, and color development of those colonies was observed. Four PTS sugars including glucose, mannose, mannitol, and N-acetylglucosamine and four non-PTS sugars including lactose, maltose, melibiose, and sucrose were tested.
Purification of Mlc-The cells containing overexpressed Mlc were sonicated in glycine-NaOH buffer, pH 9.5, and the cell debris was removed by ultracentrifugation at 100,000 ϫ g for 90 min. The supernatant was loaded onto a fast protein liquid chromatography Mono-Q 5/5 column equilibrated with glycine-NaOH buffer (buffer G), pH 9.5, containing 1 mM DTT and 50 mM NaCl. The column was washed with 3 column volumes of buffer G containing 1 mM DTT and 50 mM NaCl, and then Mlc was eluted with a linear salt gradient (7 column volumes of buffer G containing 50 mM NaCl to 7 column volumes of buffer G containing 500 mM NaCl in the presence of 1 mM DTT). The elution of Mlc was monitored by SDS-polyacrylamide gel electrophoresis, and the fractions containing substantial amounts of Mlc were pooled and concentrated in a 3 K Macrosep centrifugal concentrator (Filtron). The concentrate was chromatographed through a HiLoad 16/60 Superdex 75 prep grade column (Amersham Pharmacia Biotech) using buffer G with 1 mM DTT and 50 mM NaCl as eluent. The fractions containing Mlc were pooled and concentrated as described above.
In Vitro Transcription Assay-Reactions were done as described by Ryu and Garges (2) in a 25-l volume containing the following: 20 mM Tris acetate, pH 8.0, 3 mM magnesium acetate, 200 mM potassium glutamate, 1 mM dithiothreitol, 1 mM ATP, 0.2 mM GTP, 0.2 mM CTP, 0.02 mM UTP, 10 Ci of [␣-32 P]UTP (800 Ci/mmol), 2 nM supercoiled DNA template, 20 nM RNA polymerase holoenzyme, 100 g/ml bovine serum albumin, and 5% glycerol. Additional regulators such as 40 nM CRP, 100 M cAMP, or 0 -200 ng of Mlc were added to the reaction as needed. All components except nucleotides were incubated at 37°C for 10 min. Transcriptions were started by the addition of nucleotides and terminated after 10 min by the addition of 25 l of formamide loading buffer (80% formamide, 89 mM Tris base, 89 mM boric acid, 2 mM EDTA, 0.05% bromphenol blue, 0.05% xylene cyanol). RNA was resolved by electrophoresis on an 8 M urea, 6% polyacrylamide gel. The amounts of transcripts were measured using a Molecular Imager system (Bio-Rad).
DNase I Footprinting-DNase I protection experiments were done as described by Ryu et al. (7). Two nM DNA, 40 nM CRP, 200 ng of Mlc, and 100 M cAMP were mixed in transcription buffer in the combinations as indicated in Fig. 5 and incubated at 37°C for 10 min. RNA polymerase was added and incubated for an additional 10 min at 37°C before DNase I treatment. The DNase I treatment was terminated by adding an equal volume of 20 mM EDTA and heating at 85°C for 3 min. The DNase I-treated DNA was cleaned with a Wizard DNA cleanup kit (Promega, Madison, WI), probed with 32 P-end labeled primer using Klenow polymerase, and analyzed on a 6% sequencing gel.

RESULTS
Mlc Is a Repressor of the pts P0 Promoter-We have suggested the presence of a repressor that is inducible by glucose in the pts P0 promoter (2). The presence of a sequence that has a high similarity with the known Mlc binding site (9, 10) at the pts P0 promoter region prompted us to study the effect of Mlc on the expression of the promoter (Fig. 1). We tested the effect of Mlc on transcription of the P0 promoter in vivo by primer extension assay of total RNA extracted from cells grown in the presence or absence of glucose. The P0 transcription was stimulated by growth on glucose in a wild type strain as expected (Fig. 2). In the absence of glucose, the mlc-negative mutant showed a much higher transcription level of the P0 promoter than the wild type strain, implying that Mlc negatively regulates the expression of the P0 promoter in the wild type (compare lanes 2 and 4 in Fig. 2). The level of transcription from the P0 promoter of a mlc-negative mutant grown without glucose (Fig. 2, lane 4) was even higher than that of a wild type strain grown in the presence of glucose (Fig. 2, lane 1). This is because the P0 is activated further by CRP⅐cAMP in the absence of glucose in a mlc-negative mutant. In glucose-grown cultures, however, the mlc-negative strain showed a similar expression level of P0 promoter with the wild type strain (compare lanes 1 and 3 of Fig. 2). These results suggest that Mlc is a repressor of the pts P0 promoter, and Mlc-dependent repression is relieved by glucose in the growth medium.
Purification of Mlc-Mlc protein was purified to characterize its effects on the pts P0 promoter further. Mlc could be overproduced in E. coli strain GI698 containing pNS100 to about 25% of the total cell protein after induction (Fig. 3). Even though a high level of expression of Mlc could be achieved using the system, the protein was found in the insoluble fraction of the lysate (lanes 3 and 4 of Fig. 3). Various methods as suggested by Rudolph and Lilie (21) were tried to increase the solubility of Mlc in the cell extract, and we found that Mlc was soluble in glycine-NaOH buffer, pH 9.5. Mlc was purified using a combination of anion-exchange and gel filtration chromatography. The overall purification was 3.8-fold from the crude extract to the final preparation (Table I). From 100 ml of culture (about 350 mg of wet weight cell), a yield of 5.5 mg of approximately 95% pure Mlc was obtained as judged by SDSpolyacrylamide gel electrophoresis (Fig. 3, lane 6). The final preparation was free of RNase activity and was directly used for in vitro transcription assay.
Effect of Mlc on the pts P0 Transcription in Vitro-The effect of Mlc on the P0 transcription was studied further by an in vitro transcription assay using purified proteins. The supercoiled pHX DNA containing both P0 and P1 promoters was used as a DNA template (2). Mlc inhibited transcription from the P0 promoter but did not affect transcription of the P1 promoter (Fig. 4A). The specificity of the Mlc action on the P0 promoter was further demonstrated by the consistent activity of the rep that is originated from the plasmid origin of replication regardless of the presence of Mlc (Fig. 4A). The inhibitory effect of Mlc on the P0 promoter was reduced in the presence of CRP⅐cAMP (compare lanes 1 and 3 in Fig. 4A), even though the influence of CRP⅐cAMP was not significant.
Varying amounts of purified Mlc were added to the in vitro transcription reactions to determine the kinetic behavior of the purified protein in the transcriptional regulation of the P0 promoter. Half repression by the recombinant Mlc, in the absence and presence of CRP⅐cAMP, was obtainable with about 6 and 12 ng (Fig. 4B), respectively, showing that the stimulatory effect of the CRP⅐cAMP complex on P0 in the presence of Mlc was not significant, and Mlc is a stronger regulator.
Binding of Mlc to the pts P0 Promoter Region-We tested the binding of Mlc on the P0 promoter region to determine the mechanism of Mlc action. Binding of Mlc, CRP⅐cAMP, and RNA polymerase on the P0 promoter region was studied using a DNase I footprinting experiment. Fig. 5 showed that Mlc bound to the P0 promoter region, and the binding of both Mlc and CRP⅐cAMP to the P0 promoter region was independent of each other (compare lanes 2, 3, and 7). DNase I footprinting results (Fig. 5) confirmed the Mlc binding site around ϩ1 to ϩ25 of the P0 promoter as indicated by sequence comparison of the known Mlc binding sites to the region. DNase I footprinting results also showed that the binding of RNA polymerase to the P0 promoter region was inhibited in the presence of Mlc (compare lanes 2, 4, and 5 in Fig. 5).
Phenotype of E. coli Overexpressing Mlc under Various   Growth Conditions-The phenotype of E. coli overexpressing Mlc was examined on MacConkey agar plates containing various PTS and non-PTS sugar substrates. The overall colony size of GI698 harboring pNS100 (containing the mlc gene) was larger than the strain harboring a control vector, pRE1, regardless of the sugar substrates, and colonies on plates containing readily fermentable sugars showed a reddish purple color 6 h after spotting at 37°C. When non-PTS sugars such as lactose, maltose, melibiose, and sucrose were used as substrate, there were no remarkable differences between GI698 harboring pNS100 and the strain containing pRE1 except for maltose (Table II). When PTS sugars were included in MacConkey plates, however, Mlc overexpression seriously retarded the fermentation rate of those PTS sugars regardless of the sugar substrates (Table II). Repressing activity of Mlc and derepression of Mlc activity by readily fermentable sugar substrates have been shown for several sugar-specific transporter genes and the pts operon (8 -10, 18). According to these observations, the wild type cells grown in the presence of these sugar substrates should show the similar fermentation rate compared with the mlc-negative cells. When we compared fermentation patterns of the mlc mutant and the wild type cells to the GI698 strain harboring pRE1 or pNS100, the mlc-negative mutant showed no remarkable difference in fermentation patterns with the wild type cells or the GI698 strain harboring pRE1, as expected. These results imply that Mlc, which mainly affects utilization of PTS sugars, is a global regulator of carbohydrate metabolism.

DISCUSSION
Because of the multiple roles exerted by the gene products of the pts operon (22, 23), the intricate regulation of expression of the operon is crucial for the survival of the organism. The expression level of pts gene products in E. coli was reported to change 2-to 3-fold in response to the environmental changes such as the availability of the sugar substrates and oxygen (1). It has been known that the pts P0 promoter is activated when cells are grown in the presence of glucose. Examination of the pts P0 promoter region revealed a site that has a high similarity to the sequence of the known Mlc binding sites. The mlc encodes a 44-kDa protein, Mlc, that has been shown to regulate the expression of ptsG encoding IICB Glc , manXYZ encoding enzyme II of the mannose PTS, malT encoding the activator of maltose regulon, and mlc itself (8 -10). Mlc is proposed to be a global regulator of carbohydrate metabolism (8,9).
Comparison of transcription from the pts P0 promoter in wild type and mlc-negative mutant suggests that Mlc is a repressor of the P0 promoter. The P0 promoter was activated in the wild type, whereas it was repressed in a mlc-negative mutant when cells were grown in the presence of glucose (Fig. 2). The probable reason for the reduction of the activity of the P0 promoter when the mlc-negative strain grows in the presence of glucose is that the concentration of an activator of the P0 promoter, CRP⅐cAMP, was lowered in that condition (1,6). These results also indicated that the action of Mlc in the regulation of the pts operon is dominant over that of CRP⅐cAMP as proposed for the regulation of ptsG expression by Mlc recently (9).
It has been suggested that the concentration of Mlc in E. coli is limiting (9). Expression of the gene is autoregulated (8), and the translation efficiency of the mlc is low because the translation initiation codon of the mlc is GTG (12). We cloned the mlc under the control of P L and changed the GTG to ATG in our expression vector pNS100 for better expression of the gene. We could get more than 95% pure Mlc using a combination of ion-exchange and gel filtration chromatography. This is the first report describing the purification of Mlc. In vitro transcription assays with purified Mlc clearly showed that Mlc specifically inhibited transcription from the P0 promoter but had no effect on the P1 promoter (Fig. 4A). CRP⅐cAMP reduced the action of Mlc a little but could not prevent Mlc from inhibiting the P0 promoter. It is, however, possible that the modest influence of CRP⅐cAMP on Mlc action revealed by in vitro transcription assay can be significant in vivo because Mlc concentration is limiting in E. coli (9). In vitro transcription assay results together with the in vivo Mlc effects on the pts P0 promoter activity strongly demonstrated that Mlc was a repressor of the pts P0 promoter that can be induced by glucose. We identified one Mlc binding site centered at ϩ13 of the promoter. The sequence has high homology with the tentative consensus sequence of the Mlc binding site as proposed by Kimata et al.  Ϫ Ϫ (9) (Fig. 1). The binding sites of CRP⅐cAMP and Mlc do not overlap and their binding to the P0 promoter region was independent of each other as expected from the previous finding that the glucose-and cAMP-mediated activations of the pts promoter were independent of each other (2). Mlc inhibited the P0 promoter activity by interfering with the binding of RNA polymerase to the promoter (Fig. 5).
The common feature of five operons (manXYZ, malT, mlc, ptsG, and pts) identified as the Mlc regulon so far is that all of them have at least one CRP⅐cAMP binding site as well as a Mlc binding site, and thus they are under the dual regulation. This seems to be necessary for the fine control of expression of these genes to respond to various environmental conditions. It is interesting that E. coli overexpressing Mlc showed the same phenotype (the cells having a very slow fermentation rate) on N-acetylglucosamine and mannitol indicator plates as on glucose and mannose plates (Table II). The nagE gene of N-acetylglucosamine PTS and the mtlA gene of mannitol PTS correspond to the ptsG gene of glucose PTS and the manXYZ gene of mannose PTS in that all of them encode the major sugarspecific transporters, enzyme II. It is not known if Mlc is involved in the regulation of the nagE and mtlA genes, even though positive regulation of nagE by the CRP⅐cAMP complex was reported recently (24). In any case, we can expect that Mlc regulates utilization of many PTS sugars because it modulates the expression of the pts that encodes general proteins for all PTSs, enzyme I, and HPr. To serve as a common regulator of metabolism of many sugars, Mlc should be induced by different kinds of sugar. The inducer of Mlc has not been identified despite repeated attempts to identify the inducer that affects the Mlc binding to its binding site by many researchers (8,9). We examined the possibility that the inducer relieves repression by Mlc without affecting the binding of Mlc to its binding site using in vitro transcription assay. However, none of the sugars or their derivatives tested (glucose, glucose-6-P, maltose, mannitol, mannose, xylose, melibiose, pyruvate, N-acetylglucosamine, methyl ␤-D-thiogalactoside) showed any effect on the repression by Mlc. These results suggest the possibility that the inducer of Mlc may not be the sugar or its derivative. It has been known that glucose-mediated activation of the pts operon is dependent on the enzyme IICB Glc , which may act as a sensor protein (25). In this view, Mlc can act as a response regulator whose activity is modulated by enzyme IICB Glc . However, the possibility that Mlc is phosphorylated by enzyme IICB Glc is low because we could not get any evidence of phosphorylation of Mlc, even though we tried several different ap-proaches employing [ 32 P]phosphoenolpyruvate or [␥-32 P]ATP mixed with purified Mlc in the presence and absence of E. coli cell-free extracts to phosphorylate Mlc (data not shown). These may imply that the activity of Mlc is modulated by interaction with an unknown factor probably through protein-protein interaction as in the case of anti-factors (26,27). Glucose may affect the interaction by changing the phosphorylation state of enzyme IICB Glc , and other sugars may also affect the interaction through their own enzyme II. The unknown factor could be any one of PTS proteins or an unknown novel factor yet to be identified.
We are testing this possibility, and further work is needed to discover the inducer of Mlc would be.