Metabolism of Sucrose and Its Five Linkage-isomeric -D-Glucosyl-D- fructoses by Klebsiella pneumoniae PARTICIPATION AND PROPERTIES OF SUCROSE-6-PHOSPHATE HYDROLASE AND PHOSPHO- -GLUCOSIDASE*

From the ‡Microbial Biochemistry and Genetics Unit, Oral Infection and Immunity Branch, NIDCR, National Institutes of Health, Bethesda, Maryland 20892, the ¶Institut für Organische Chemie, Technische Universität Darmstadt, D-64287 Darmstadt, Germany, the Biology Department, University of Rochester, Rochester, New York 14627-0211, the **Vaccine and Therapeutic Development Section, Oral and Infection and Immunity Branch, NIDCR, National Institutes of Health, Bethesda, Maryland 20892, and the ‡‡Department of Infectious Diseases, Children’s National Medical Center, Washington, D. C. 20010-2970

The discovery in 1964 of the phosphoenolpyruvate-depend-ent sugar:phosphotransferase system (PEP:PTS) 1 by Roseman and colleagues (1) is a landmark in our understanding of carbohydrate dissimilation by microorganisms. Since the initial description of this multi-component system in Escherichia coli, the PEP:PTS has been established as the primary route for the transport and concomitant phosphorylation of a wide variety of sugars by bacteria from both Gram-negative (2,3) and Grampositive genera (4,5). In many species, including Bacillus subtilis, Lactococcus lactis, Streptococcus mutans, Escherichia coli, and Klebsiella pneumoniae (6,7), sucrose is accumulated via the PTS simultaneously with phosphorylation at C-6 of the glucopyranosyl moiety of the disaccharide. Intracellularly, sucrose-6-phosphate (sucrose-6-P) is hydrolyzed by sucrose-6phosphate hydrolase (8,9) to glucose-6-phosphate and fructose, which are then fermented via the glycolytic pathway to yield primarily lactic acid.
The structures of sucrose, its five isomeric ␣-D-glucosyl-Dfructoses (trivially designated trehalulose, turanose, maltulose, leucrose, and palatinose), and some related ␣-linked disaccharides are depicted in Fig. 1. In contrast to the many reports of sucrose fermentation, there are few references to the utilization of the isomeric glucosyl-fructoses by microorganisms (12). This fact is of particular relevance to oral biology in light of the associative role(s) of sucrose and streptococcal species in the etiology of dental caries (13,14). Sucrose is the precursor for glucan synthesis that facilitates attachment of S. mutans to the tooth surface; subsequent fermentation of the disaccharide to lactic acid initiates the demineralization of tooth enamel. In this context, isomers of sucrose attract attention as potential substitutes for dietary sucrose (15)(16)(17) because they are about half as sweet as sucrose, are not metabolized (noncariogenic), and, in the case of palatinose and leucrose, are produced on an industrial scale (18,19). From the limited information available, one might reasonably conclude that the isomers cannot be translocated by the membranelocalized transporter EII(CB) of the sucrose-PTS or that intracellular sucrose-6-P hydrolase is unable to hydrolyze the phos-* 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.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s) AF337811.
§ To whom correspondence and reprint requests should be addressed: NIDCR, National Institutes of Health, Bldg. 30 Our interest in these issues stemmed from a survey of disaccharide utilization by K. pneumoniae that revealed excellent (and unexpected) growth of this organism on all five isomers of sucrose (12). Furthermore, although organisms grown previously on a particular isomer readily metabolized sucrose and all other isomers, cells of K. pneumoniae grown previously on sucrose fermented only sucrose (12). Comparative analyses of proteins in various cell extracts (by two-dimensional PAGE) revealed high level expression of a specific polypeptide (molecular mass ϳ 50 kDa) during growth on the isomers, but this protein was not induced by growth of the organism on sucrose. These observations provided the first indication that for K. pneumoniae, the initial steps in metabolism of sucrose, and those of its analogs, might be separable and distinct. In the present study we have identified two adjacent genes (aglA and aglB) in K. pneumoniae that encode a membrane-localized transport protein of the PTS (EIICB, or AglA) and a nucleotide (NAD ϩ ) plus metal-dependent phospho-␣-glucosidase (AglB), respectively. Together, these proteins facilitate the phosphorylative translocation and subsequent hydrolysis of the five ␣-linked isomers of sucrose.
To facilitate the comparison of the properties of sucrose-6-P hydrolase with those of AglB, the genes encoding the two proteins (scrB (7) and aglB, respectively) have been cloned, and both enzymes have been purified from high expression systems. Recently, we prepared trehalulose-6Ј-P, turanose-6Ј-P, maltulose-6Ј-P, leucrose-6Ј-P, and palatinose-6Ј-P in substrate quantity (12), and the availability of these novel compounds permitted the determination of the substrate specificities of highly purified AglB and sucrose-6-P hydrolase. Remarkably, sucrose-6-P hydrolase, which by sequence-based alignment is assigned to Family 32 of glycosyl hydrolases, hydrolyzed only sucrose-6-P. In contrast AglB, which belongs to Family 4, catalyzed the cleavage of the five isomeric 6Ј-phosphoglucosylfructoses. In this paper, a comparative assessment of conformational, overall shape and polarity features of sucrose-6-P and its isomeric disaccharide-6Ј-phosphates is given, providing insight into the molecular basis for substrate discrimination by the two phosphoglucosyl hydrolases.
Growth of K. pneumoniae ATCC 23357-The organism was grown at 37°C in 1-liter bottles, each containing 800 ml of the medium defined by Sapico et al. (21) supplemented with 0.4% (w/v) of the appropriate sugar. After growth to stationary phase, the cells were harvested by centrifugation (13,000 ϫ g for 10 min at 5°C) and washed twice in 25 mM Tris-HCl buffer (pH 7.5) containing 1 mM MgCl 2 . The yield was ϳ2 g wet weight of cells/liter.
Electrophoresis Procedures-SDS-PAGE was carried out in the Novex XCell Mini-Cell system (Invitrogen). Novex NuPage (4 -12%) Bis-Tris gels and MES-SDS running buffer (pH 7.3) were used together with Novex Mark 12 TM protein standards, and proteins were stained with Coomassie Brilliant Blue R-250. For Western blots, proteins were transferred to nitrocellulose membranes using NuPage transfer buffer and SeeBlue TM prestained standards. The Amersham Pharmacia Biotech Multiphor flat-bed electrophoresis unit, precast Ampholine PAG plates (pH range, 3.5-9.5) and broad range standards were used for electrofocusing experiments.
Analytical Methods-During purification, the activity of AglB in column fractions was detected by hydrolysis of the chromogenic substrate, pNP␣Glc6P. The specific activity of the enzyme was determined in a discontinuous assay that contained in 2-ml: 0.1 M Tris-HCl buffer (pH 7.5), 1 mM MnCl 2 , 0.5 mM NAD ϩ , and 1 mM pNP␣Glc6P. After the addition of the enzyme preparation, samples of 0.25 ml were removed at 20-s intervals (over a 2-min period) and immediately injected into 0.75 ml of 0.5 M Na 2 CO 3 . The A 400 nm of the yellow solution was measured, and rates of pNP formation were calculated by assuming a molar extinction for the p-nitrophenoxide anion ⑀ ϭ18,300 M Ϫ1 cm Ϫ1 . One unit of AglB activity is the amount of enzyme that catalyzes the formation of 1 mol of pNP min Ϫ1 . Two-dimensional polyacrylamide gel electrophoresis (PAGE) and protein microsequencing were carried out by Kendrick Laboratories, Inc. and by the Protein Chemistry Core Facility, Columbia University, NY, respectively. The mass of AglB was determined by electrospray in an HP1100 mass spectrometer, and the sequence of N-terminal amino acids was determined with an ABI 477A protein sequencer (Applied Biosystems Inc.) with an on-line ABI 120A phenylthiohydantoin analyzer. Protein concentrations were determined by the BCA protein assay kit (Pierce). The procedure for immunodetection of AglB with polyclonal antibody to MalH from F. mortiferum has been described previously (20).
The components of the amplification mixtures (100 l) were: 5 units of Pfu DNA polymerase (Stratagene, La Jolla, CA), 1ϫ buffer provided by the manufacturer, 20 mM each of the four DNTPs, 100 ng of DNA, and 250 ng of each primer. Amplifications were carried out in a thermal cycler (PerkinElmer 9600, PerkinElmer Life Sciences). After an initial 2-min denaturation at 95°C, the mixtures were subjected to 30 cycles of amplification. Each cycle consisted of 1 min denaturation at 95°C, 1 min annealing at 58°C, and extension at 72°C for 2 min/kilobase of insert. These were followed by a 10-min runoff at 72°C. The PCR products were purified (QIAquick PCR purification kit, Qiagen) and ligated into pCR-Blunt vector (Invitrogen, Carlsbad, CA). After transformation into E. coli TOP 10 competent cells, colonies were selected on LB agar plates containing 50 g/ml kanamycin.
Cloning of the K. pneumoniae ATCC 23357 aglB Gene in E. coli-For amplification of the gene aglB, two primers were synthesized from the nucleotide sequence shown in Fig. 4: forward primer KPBF, 5Ј-CCCAC-CATGGGAGGCAGTATCATG-3Ј (the aglB sequence, base pairs 2001-2015, is in bold face, and the NcoI site is underlined); reverse primer KPBR, 5Ј-CCCAGAATTCTTAATGCAGCTCAGG-3Ј (the sequence complementary to the downstream region of aglB, base pairs 3321-3335, is in bold face, and the EcoRI site is underlined). PCR amplification was performed using high fidelity Pfu DNA polymerase. The amplified 1.3-kilobase DNA fragment was digested with restriction endonucleases (NcoI and EcoRI), electrophoresed through 1% agarose gel, and purified (QIAquick gel extraction kit). The fragment was ligated into the similarly digested (NcoI-EcoRI) high expression vector pSE380 (Invitrogen) to form pAP-16. (In this construct, the aglB gene is under control of the powerful trc hybrid promoter, which is also regulated by the lacO operator and the product of the lacI q gene. Because the plasmid also carries lacI, expression of aglB is strongly repressed in the absence but is fully induced in the presence of isopropyl-␤-D-thiogalactopyranoside (IPTG)). Plasmid pAP-16 was transformed into competent cells of E. coli TOP 10 (Invitrogen), and transformants were selected on LB agar containing 150 g/ml ampicillin.
DNA Sequence Analysis-DNA fragments cloned in pCR-Blunt vector were sequenced by the dideoxynucleotide chain termination method using the Sequenase 7-deaza-dGTP sequencing kit (U. S. Biochemicals, Cleveland, OH), and [␣-35 S]deoxyadenosine triphosphate was used for labeling. For all clones, both strands of DNA inserts were sequenced. The MacVector sequence analysis package (Version 7.0, Genetics Computer Group, Madison, WI) was used to compile, edit, and analyze the results.
Growth of Cells and Preparation of Extract Containing AglB-E. coli TOP 10 (pAP-16) was grown at 37°C in LB medium containing ampicillin (150 g/ml) to a density A 600 nm ϳ 0.4 units. IPTG (0.5 mM) was then added to the culture, and growth was continued for 3 h. The culture was harvested by centrifugation (13,000 ϫ g for 10 min at 5°C), and the cells (ϳ 2.1 g wet weight/liter) were washed by resuspension and centrifugation from 25 mM Tris-HCl buffer (pH 7.5) containing 1 mM MnCl 2 and 0.1 mM NAD ϩ (designated TMN buffer). Washed cells (ϳ38 g) were resuspended in 80 ml of TMN buffer, and the organisms were disrupted (at 0°C) by 2 ϫ 1.5-min periods of sonic oscillation in a Branson instrument (model 350) operating at ϳ75% of maximum power. The extract was clarified by centrifugation (180,000 ϫ g for 2 h at 5°C), and the high-speed supernatant was transferred to sacs and dialyzed overnight against 4 liters of TMN buffer.

Purification of AglB
The enzyme was purified by low pressure chromatography, and all procedures were performed in a cold room.
Step 1: TrisAcryl M-DEAE (Anion Exchange) Chromatography-Dialyzed high-speed supernatant (ϳ85 ml) was transferred at a flow rate of 0.8 ml/min to a column of TrisAcryl M-DEAE (2.6 ϫ 14 cm) previously equilibrated with TMN buffer. Nonadsorbed material was removed by washing with TMN buffer, and AglB was eluted with 800 ml of a linear, increasing concentration gradient of NaCl (0 -0.3 M) in TMN buffer. Fractions (8 ml) were collected, and AglB activity was revealed by the intense yellow color formed upon addition of fraction samples (4 l) to microtiter wells containing 100 l of pNP␣Glc6P assay solution. Fractions with the highest activity (22)(23)(24)(25)(26) were pooled and concentrated to 19 ml by pressure filtration (Amicon PM-10 membrane, 40 psi). Ammonium sulfate crystals (1.9 g) were added slowly with stirring to a concentration of 0.75 M.
Step 2: Phenyl-Sepharose CL-4B (Hydrophobic) Chromatography-The ϳ 20 ml solution from step 1 was transferred (flow rate 0.5 ml/min) to a column of phenyl-Sepharose CL-4B (2.6 ϫ 14 cm) equilibrated with TMN buffer containing 0.75 M (NH 4 ) 2 SO 4 . Nonadsorbed protein(s) were eluted, and then 500 ml of a decreasing, linear gradient of (NH 4 ) 2 SO 4 (0.3-0 M) in TMN buffer was passed through the column. Fractions of 5 ml were collected, and AglB was recovered primarily in fractions 45-60. These fractions were pooled and concentrated by Amicon filtration to ϳ9 ml.
Step 3: Ultrogel AcA-44 (Molecular Sieve) Chromatography-Approximately 3 ml of preparation from step 2 was applied at a flow rate of 0.15 ml/min to a column of Ultrogel AcA-44 (1.6 ϫ 94 cm) previously equilibrated with TMN buffer containing 0. 1 M NaCl. Fractions of 2.1 ml were collected, and those containing maximum AglB activity (50 -53) were pooled. (This procedure was repeated with the remaining 2 ϫ ϳ3-ml portions of concentrate from phenyl-Sepharose chromatography). AcA-44 chromatography yielded a total of ϳ24 ml of highly purified AglB (3 mg/ml; specific activity 4.15 units/mg).
Kinetic Analysis and Substrate Specificity of AglB-A continuous spectrophotometric assay was used for substrate specificity studies and determination of kinetic parameters for AglB. This indirect glucose-6-P dehydrogenase/NADP ϩ -coupled assay monitors formation of glucose-6-P during the AglB-catalyzed hydrolysis of substrates. The standard 1-ml assay contained: 0.1 M HEPES buffer (pH 7.5), 1 mM MgCl 2 , 1 mM MnCl 2 , 1 mM NAD ϩ , 1 mM NADP ϩ , 1 mM substrate (6Ј-P isomer of sucrose, or phospho-␣-glucoside), and 2 units of glucose-6-P dehydrogenase/hexokinase. Reactions were initiated by addition of 15 l (45 g) of AglB preparation, and the increase in A 340 nm was recorded in a Beckman DU 640 spectrophotometer. Initial rates were determined using the kinetics program of the instrument, and a molar extinction coefficient ⑀ ϭ 6, 220 M Ϫ1 cm Ϫ1 was assumed for calculation of NADPH formed (equivalent to glucose-6-P liberated). In kinetic analyses the concentration range of substrate was usually 0.2-4 mM, and kinetic parameters were determined from Hofstee plots with an Enzyme Kinetics program (dogStar software, Version 1.0c). The products of turanose-6Ј-P hydrolysis (glucose-6-P and fructose) were determined by inclusion of 5 mM ATP and 2 units of phosphoglucose isomerase in the assay.
Cloning of the Sucrose-6-P Hydrolase Gene (scrB) from K. pneumoniae The scrB gene was amplified from K. pneumoniae genomic DNA using the low-error-rate FailSafe TM polymerase (Epicentre). In the forward primer (5Ј-GGCCATGGCGCTCTCTCTGACGCTGAA-3Ј), base pairs 3119 -3137 of the K. pneumoniae scrYAB operon (GenBank TM accession number X57401) are bolded, and the NcoI site is underlined. In the reverse primer (5Ј-GGGGGTCGACTACGCGTTTGGTTT-TCATCA-3Ј), base pairs 4587-4606 of scrYAB are bolded, and the SalI site is underlined. The amplicon was digested with NcoI and SalI and ligated into similarly digested pProEX Hta (Life Technologies, Inc.), and the recombinant plasmid(s) was transformed into E. coli K12 strain DH5␣-E (Life Technologies, Inc.). Ampicillin-resistant transformants were selected, and the scrB genes of four plasmids containing inserts of approximately the correct size were sequenced. All shared the following differences from the published (7) scrB coding sequence: 991C3 T, 1006G3 T, 1249C3 T, 1270C3 T, 1549C3 T, 1552A3 G, 1675G3 A, 1699T3 C, 1738G3 A (numbering as in pScrBLong). All of these differences are silent, and one plasmid was chosen and designated pScrBLong.

Growth of E. coli DH5␣E (pScrBLong) and Expression of Sucrose-6-P Hydrolase
The organism was grown in LB medium containing 200 g/ml ampicillin. At A 600 nm ϭ 0.5, IPTG was added (1 mM) and growth was continued for ϳ 4 h. Cells were harvested and washed with 25 mM HEPES buffer (pH 7.5) as described earlier. The yield was ϳ2.9 g wet weight of cells/liter.

Purification of Sucrose-6-P Hydrolase
Briefly, the purification of sucrose-6-P hydrolase was as follows. A high-speed supernatant was prepared, after resuspension and sonication, of 10 g of E. coli DH5␣E (pScrBLong) resuspended with 20 ml of 25 mM HEPES buffer (pH 7.5). The dialyzed preparation was applied to a column of TrisAcryl M-DEAE, and after washing with the same buffer, sucrose-6-P hydrolase was eluted with an increasing gradient of NaCl (0 -0.5 M). Fractions with sucrose-6-P hydrolase activity were pooled, concentrated to 8 ml, and then mixed gently with 30 ml of 0.1 M MES buffer (pH 5). Precipitated material was removed by centrifugation, and the clarified solution was applied to a column of phosphocellulose P-11 (Whatman) previously equilibrated with 0.1 M MES buffer (pH 5). Nonadsorbed proteins were removed, and sucrose-6-P hydrolase was eluted with the same buffer containing an increasing concentration of potassium phosphate buffer (0 -0.1 M, pH 7). Active fractions (eluted at ϳ 50 mM P i ) were pooled and concentrated. sucrose-6-P hydrolase was purified to homogeneity by passage of this solution through an AcA-44 gel filtration column previously equilibrated with 50 mM HEPES buffer, pH 7.5, containing 0.1 M NaCl. Concentration of active fractions yielded about 22 mg of sucrose-6-P hydrolase of specific activity 12.5 units/mg (with 10 mM sucrose as substrate in the assay; see below).

Sucrose-6-P Hydrolase Assay
Sucrose-6-P is the natural substrate for sucrose-6-P hydrolase, but the enzyme also hydrolyzes sucrose when the disaccharide is present at high concentration. Because of the limited availability of sucrose-6-P, the parent sugar was used as substrate during purification of sucrose-6-P hydrolase, and the glucose-6-P dehydrogenase/hexokinase-NADP ϩcoupled assay measured glucose formed by sucrose hydrolysis. The 1-ml assay contained: 0.1 M HEPES buffer (pH 7.5), 1 mM MgCl 2 , 1 mM NADP ϩ , 1 mM ATP, 10 mM sucrose, 2 units of glucose-6-P dehydrogenase/hexokinase, and enzyme solution. One unit of sucrose-6-P hydrolase activity is the amount of enzyme that catalyzes the formation of 1 mol of glucose/min.

RESULTS
Growth of K. pneumoniae on Sucrose and Its Isomers-Recently (12) we reported the growth of K. pneumoniae on sucrose and its five isomeric ␣-D-glucosyl-D-fructoses (see Fig. 1 for structures). Additionally, we showed that organisms grown previously on a particular isomer readily fermented sucrose as well as each of the ␣-D-glucosyl-D-fructoses, whereas sucrosegrown cells, surprisingly, metabolized only sucrose (12). Examination of the protein composition of the various cell extracts by two-dimensional PAGE revealed high level expression of one specific polypeptide (molecular mass ϳ50 kDa) during growth on either of the five sucrose isomers (e.g. palatinose and maltulose, Fig. 2) and in fact on related disaccharides such as maltose, isomaltose, maltitol (for formulae, see Fig. 1), and even methyl-␣-D-glucopyranoside (data not shown). Significantly, the ϳ50-kDa protein was not detectable in an extract prepared from sucrose-grown cells (Fig. 2).
Identity of the Protein Induced during Growth on Sucrose-Isomeric Glucosyl-fructoses-Proteins from a duplicate two-dimensional PAGE gel of maltulose-grown cell extract were transferred by Western blot to a polyvinylidene difluoride membrane. Microsequence analysis provided the following sequence for the first 25 residues from the N terminus of the highly expressed ϳ50-kDa protein: MKKFSVVIAGGGSTFTP-GIVLMLLA. A BLAST (43) search of the nonredundant protein data bases with this sequence as probe revealed 91 and 82% identity, respectively, with the N termini of an unusual 6-phospho-␣-glucosidase (EC 3.2.1.122), previously purified from Fusobacterium mortiferum (MalH (44)) and B. subtilis (GlvA (45)).
Phospho-␣-glucosidase activity is readily detected by the intensely yellow p-nitrophenolate (pNP) anion released upon hydrolysis of pNP␣Glc6P. This chromogenic substrate was rapidly hydrolyzed by extracts of cells grown on the glucosylfructoses and other ␣-glucosides, but essentially no activity was detectable in the extract from sucrose-grown cells (Table I). Western blots performed with antibody raised against phospho-␣-glucosidase from F. mortiferum (20) revealed a striking correlation between the amount of induced immunoreactive protein of ϳ50 kDa (Fig. 3) and the hydrolytic activities of the various extracts (Table I). The protein induced during growth on the five ␣-D-glucosyl-D-fructoses was thus identified as phospho-␣-glucosidase.
Cloning and Sequence Analysis of the Agl Region of K. pneumoniae-Although suggestive, the available data (Figs. 2 and 3 and Table I) did not establish a functional role for phospho-␣glucosidase in dissimilation of the five ␣-D-glucosyl-D-fructoses by K. pneumoniae. Recently, we demonstrated the PEP-dependent phosphorylation of the five sucrose isomers via the PTS activity of palatinose-grown cells of K. pneumoniae, and trehalulose-6Ј-P, turanose-6Ј-P, maltulose-6Ј-P, leucrose-6Ј-P, and palatinose-6Ј-P were prepared in 20 -50-mg amounts (12). To determine whether these derivatives were hydrolyzed by AglB, it was first necessary to purify this enzyme. To this end, aglB, the gene encoding the phospho-␣-glucosidase, and an adjacent upstream gene, aglA, were cloned and sequenced.

FIG. 1. Chemical formulae and established abbreviations (10) of sucrose, its five linkage isomeric glucosyl-fructoses, and of some related disaccharides (R ‫؍‬ H) and their respective monophosphates (R ‫؍‬ PO 3 2؊ ), invariably carrying their phosphate ester groups attached to the glucosyl-C-6.
For the reducing disaccharides, only the tautomeric form predominating in solution (10, 11) is depicted. The nonreducing sucrose-6-P is the singular substrate for the sucrose-6-P hydrolase, whereas all others are hydrolyzed by the 6-phospho-␣-glucosidase described herein. polypeptide of 540 residues (calculated M r ϭ 58,373) that contains fused C and B domains characteristic of a membranelocalized EII(CB) transport protein of the PTS (46). The aglA gene is preceded by a potential ribosome-binding site (GAGGA) centered ϳ11 nucleotides from the start codon. Purification of Phospho-␣-glucosidase (AglB) from E. coli TOP(pAP-16)-Cells of E. coli TOP(pAP-16) produced high levels of an IPTG-inducible protein with an estimated M r ϳ 50 kDa as expected for the full-length polypeptide encoded by aglB (Fig. 6A, lane 1). This protein cross-reacted with phospho-␣glucosidase antibody (Fig. 6B, lane 1), and the cell extract catalyzed the immediate hydrolysis of pNP␣Glc6P. AglB was purified by conventional low-pressure chromatography, and to stabilize the enzyme, 0.1 mM NAD ϩ and 1 mM Mn 2ϩ ion were included in all buffers. Throughout the four-stage procedure, 3 On line at www.expasy.ch/cgi-bin/lists?glycosid.txt.

FIG. 2. Analysis by two-dimensional PAGE of proteins in extracts prepared from cells of K. pneumoniae
grown on various disaccharides. The white circles indicate the induced ϳ50-kDa phospho-␣-glucosidase (AglB) in organisms grown previously on either maltose, palatinose, or maltulose. This protein was not detectable (white arrow) in the extract prepared from sucrose-grown cells of K. pneumoniae. Approximately 50 g of protein was applied per gel, and polypeptides were visualized by silver staining. Prior to electrophoresis, tropomyosin (1 g) was added to each sample as an IEF internal standard. This protein (black arrowhead) migrates as a doublet with the lower polypeptide spot ϳ33 kDa and pI ϭ 5.2.  (20). Extracts were prepared from cells of K. pneumoniae grown on the indicated sugars, and approximately 15 g of protein was applied per lane. Note the absence of immunoreactive protein in sucrose-grown cells. Stds., standards. the purification of AglB was monitored by enzymatic assay (Table II), SDS-PAGE (Fig. 6A), and immunoblot methods (Fig.  6B). Approximately 70 mg of electrophoretically pure enzyme was obtained from ϳ38 g wet weight of cells. Although purified in reasonably active form, AglB was progressively inactivated throughout the purification, and the specific activity of the final preparation (4.2 units/mg) was only ϳ3-fold higher than that of the original dialyzed cell extract (1.2 units/mg).

FIG. 3. Western blot showing the sugar-specific induction and cross-reactivity of the ϳ50-kDa protein (AglB) with antibody raised against purified MalH (phospho-␣-glucosidase) from F. mortiferum
Properties of AglB-The molecular weight of AglB determined by electrospray/MS (M r 49,254) was within two units of the theoretical weight average M r of 49,256 deduced from the amino acid sequence encoded by aglB. However, in the final stage of purification, AglB emerged from the AcA-44 gel filtration column in a volume suggestive of a protein of molecular mass ϳ 100 kDa. Cross-linking studies also revealed the formation of a similarly sized product after incubation of the enzyme with various homo-bifunctional imidoesters (Fig. 6C,  lanes 2-4). It appears likely that in solution AglB exists as a catalytically active homodimer. Analytical electrofocusing revealed two species (Fig. 6D, lane 2) having estimated pI values of 5.4 and 5.6 that agreed fairly well with the theoretical pI (5.69) deduced from the amino acid composition of AglB. The homogeneity of the purified enzyme was confirmed by the unambiguous determination of the first 26 residues from the N terminus, MKKFSVVIAGGGSTFTPGIVLMLLAN. This sequence was precisely that deduced by translation of aglB and, importantly, was in perfect agreement with that of the polypeptide induced during growth of K. pneumoniae on the sucroseisomeric glucosyl-fructoses (Fig. 2).
Cofactor, Metal Ion Requirements, and Substrate Specificity of AglB-Phospho-␣-glucosidases MalH and GlvA from F. mortiferum (20,44) and B. subtilis (45), respectively, exhibit requirements for nucleotide (NAD ϩ ) and divalent metal ion (Mn 2ϩ , Co 2ϩ , or Ni 2ϩ ) for activity. AglB exhibited similar requirements and, in the absence of these cofactors, was unable to hydrolyze pNP␣Glc6P (Table III). Inclusion of NAD ϩ in the assay elicited substrate cleavage, but enzyme activity increased 3-6-fold upon further addition of Mn 2ϩ , Co 2ϩ , or Ni 2ϩ . Other divalent metal ions tested, including Mg 2ϩ , Ca 2ϩ , and Zn 2ϩ , were either without effect or were inhibitory. The activity FIG. 4. Nucleotide sequence of the Agl region of K. pneumoniae. This ϳ3.5-kilobase DNA fragment contains genes aglA and aglB that encode an EIICB transport protein of the PEP:PTS and an NAD ϩ plus metal-dependent 6-phospho-␣-glucosidase, respectively. A potential ribosomal binding site (RBS) preceding aglA is underlined. The deduced amino acid sequences are shown below the nucleotide sequence in single-letter code. The N-terminal amino acid sequence of AglB obtained by Edman degradation is boxed. The positions of primers KPBF and KPBR used for PCR amplification of aglB are indicated by arrows above the nucleotide sequence.
of AglB was optimal at ϳ36 C°in either 0.1 M Tris-HCl or HEPES buffers (pH 7.5) containing 0.1 mM NAD ϩ and 1 mM Mn 2ϩ ion. In the presence of requisite cofactors, AglB hydrolyzed all 6-phospho-␣-D-glucosides tested including all phosphorylated isomers of sucrose. The kinetic parameters for each substrate are presented in Table IV. There was no detectable cleavage of the corresponding nonphosphorylated compounds. Importantly, sucrose-6-P itself was not hydrolyzed by AglB nor was it an inhibitor of enzyme activity. Studies with turanose-6Ј-P (Table V) established that, as for the chromogenic analog (pNP␣Glc6P), the same cofactors were required for the hydrolysis of this PTS product. Throughout the time course of the experiment, the 1:1 stoichiometry between [glucose-6-P:fructose] confirmed these two metabolites as the only reaction products from AglB-catalyzed hydrolysis of turanose-6Ј-P.
Purification and Substrate Specificity of Sucrose-6-P Hydrolase-AglB readily hydrolyzed the five 6-phosphoglucosyl-fructoses, whereas sucrose-6-P, remarkably, was not a substrate for this enzyme. It was of interest, to determine whether sucrose-6-P hydrolase would exhibit the converse specificity with respect to potential substrates. sucrose-6-P hydrolase was purified from E. coli DH5␣E (pScrB Long) as described under "Experimental Procedures." The four-stage procedure (Fig. 7) provided 20 -30 mg of electrophoretically pure sucrose-6-P hydrolase with an estimated molecular mass of ϳ53 kDa by SDS-PAGE (Fig. 7, lane 4), which was in agreement with the molecular weight of 52,708 deduced by translation of the scrB gene (ref. 7 and Swiss Protein Database accession no. P27217). The mass of sucrose-6-P hydrolase determined experimentally by electrospray mass spectroscopy (M r 52,581) was about 127 mass units lower than the calculated mass av . Except for the absence of methionine at the N terminus, microsequence analysis confirmed exactly the predicted sequence of the first 28 residues of the polypeptide SLPSRLPAILQAVMQGQPQAL-ADSHYPQ. Sucrose-6-P hydrolase catalyzed the hydrolysis of sucrose and sucrose-6-P at comparable rates (V max . sucrose ϭ 31.2 Ϯ 1.1; V max . S6P ϭ 40.4 Ϯ 2.3 mol hydrolyzed min Ϫ1 mg Ϫ1 ). However, the affinity of the enzyme for the phosphorylated disaccharide (K m S6P ϭ 85.3 Ϯ 15.1 M) was Ͼ200-fold greater than for sucrose (K m sucrose ϭ 20.3 Ϯ 1.9 mM). There was no detectable hydrolysis (at 1 mM) of any of the phosphorylated isomers of sucrose, and sucrose-6-P hydrolase failed to hydrolyze other phospho-␣-glucosides including maltose-6Ј-P and trehalose-6-P.
Conformational Analysis of Sucrose-6-P and Disaccharide-6Ј-phosphates-Insight into the remarkable discrimination of sucrose-6-P hydrolase and phospho-␣-glucosidase for their substrates was provided by conformational analysis of these phosphorylated compounds using molecular dynamics simulations. Sucrose and sucrose-6-P differ from other disaccharides not only by the fact that they are nonreducing (the two sugar units are linked through their anomeric centers) but by the predetermined orientation of the glucose and fructose portions toward each other. In the solid state, the two sugars are conformationally fixed by two intramolecular hydrogen bonds ( Fig. 8 and Refs. 26,27,50,51). On dissolution of the disaccharide in water, these bonds are replaced by an H 2 O molecule bridging glucosyl-O-2 and fructosyl-O-1 through hydrogen bonding (Fig.  8, center, and Ref. 52), to yield an overall conformation close to that in the crystalline state. The molecular geometry of sucrose-6-P in water, which emerges from a nanosecond molecular dynamics simulation in a truncated octahedron box containing 641 water molecules, again is very similar to that of sucrose in the crystal and in aqueous solution (Fig. 8, right), so that a water bridge of the Glc-2-O⅐ ⅐ ⅐H 2 O⅐ ⅐ ⅐O-1-Fru is likewise to be inferred. A comparison of the molecular geometry of sucrose-6-P in water with the geometries of the nine disaccharide-6Јphosphates reveals their distinctly different molecular shapes. Unlike sucrose-6-P, which by virtue of the intramolecular water bridge between glucose and fructose assumes a remarkably compact conformation in solution, the nine disaccharide-6Јphosphates lack any interaction of this type and hence invariably adopt a more extended, longish molecular geometry. These differences in molecular shape are emphasized by juxtaposition of the solvent-accessible surface of sucrose-6-P (Fig. 9, top) with those of the nine disaccharide-6Ј-phosphates shown superimposed in Fig. 9 (bottom).

Transport and Hydrolysis of Sucrose and Its Isomers by K.
pneumoniae-Circumstantial evidence indicated that the transport and dissimilation of the five O-␣-linked isomers of sucrose by K. pneumoniae occurred by a route different from the PTS-sucrose-6-P hydrolase route used for sucrose itself (12). For example, sucrose-grown cells failed to metabolize any of the isomers, and the PEP:PTS activity of cells grown on a particular isomer (e.g. palatinose) catalyzed the phosphorylation of all other isomers. Importantly, growth of K. pneumoniae on the five ␣-D-glucosyl-D-fructoses induced a high level expression of a polypeptide (molecular mass ϳ 50 kDa) that was not present in organisms grown on sucrose. In this study, the gene (aglB) that encodes the induced protein has been cloned and sequenced, and the protein itself (AglB) has been identified as an NAD ϩ and metal-dependent phospho-␣-glucosidase. The gene aglB lies adjacent to a second gene, aglA, which encodes an EII(CB) component of the PEP:PTS (46). It is our contention that together, AglA and AglB facilitate the phosphorylative translocation and subsequent cleavage of phosphorylated isomers of sucrose (and related ␣-glucosides) by K. pneumoniae.
Properties of sucrose-6-P hydrolase and Phospho-␣-glucosi- dase (AglB)-In some of their properties, sucrose-6-P hydrolase and AglB show similarity. For example they are of comparable (monomer) size, they are exacting for the glucose-6-P moiety of their substrates, and both exhibit poor or no affinity for nonphosphorylated disaccharides. However, in their amino acid sequences, cofactor requirements, and assignments to different families of the glycosyl hydrolase superfamily, sucrose-6-P hydrolase and AglB are quite different. The amino acid sequence of sucrose-6-P hydrolase (deduced from the scrB gene (7)) has essentially no homology with that of AglB, and by the amino acid-based sequence classification of Henrissat (47), AglB and sucrose-6-P hydrolase are assigned to Families 4 and 32, respectively, of the glycosyl hydrolase superfamily (48). 3 Sucrose-6-P hydrolase has no cofactor requirements, whereas AglB is dependent upon both NAD ϩ and divalent metal ion (Mn 2ϩ , Ni 2ϩ , or Co 2ϩ ) for catalytic activity (Table III). Indeed, these cofactor requirements for AglB were predicted by virtue of the extraordinarily high sequence identity between the putative polypeptide encoded by aglB and those of the Family 4 phospho-␣-glucosidases shown in the multiple alignment in Fig. 5. The role(s) for NAD ϩ and metal ion have not been established,   a The 2-ml assay solution contained 50 mM Tris-HCl buffer (pH 7.5). When required, appropriate divalent metal ion (1 mM) and NAD ϩ (0.5 mM) were included. Phospho-␣-glucosidase (60 g) was added, and after 2 min of preincubation at 25°C, pNP␣Glc6P was added to a final concentration of 1 mM. Hydrolysis of substrate (i.e. formation of pNP) was followed as described under "Experimental Procedures." b Expressed as mol of pNP␣Glc6P hydrolyzed min Ϫ1 mg enzyme Ϫ1 . c ND, no detectable activity. and it is presently unclear whether these cofactors play catalytic and/or structural roles in AglB and related enzymes of Family 4 (44,45,53,54). Results obtained from site-directed mutagenesis of the phospho-␣-glucosidase (GlvA) in B. subtilis (45) suggest that residues close to the N terminus comprise the NAD ϩ -binding domain (see, Fig. 5). Glycosyltransferases comprise a superfamily of Mn 2ϩ -dependent enzymes (55) that use UDP-glucose, UDP-galactose, and related compounds as substrates for modification (via glycosylation) of a wide variety of biological molecules in both prokaryotes and eukaryotes. Most, if not all, members of this large family contain a conserved motif D(X)D that participates in the substrate recognition/ catalytic process by interaction of the aspartyl residues with the ribose moiety of the nucleotide or via coordination with Mn 2ϩ ion (56). Interestingly, this motif is also present in AglB and in other phospho-␣-glucosidases, and the conserved DND residues lie adjacent to the putative NAD ϩ -binding domain of these enzymes (Fig. 5). Furthermore, site-directed substitution at the first aspartic residue of this motif (D41G and D41E) in GlvA results in loss of hydrolytic activity (45). These findings plus the fact that the D(X)D motif is conserved in other members of Family 4 (see Fig. 4 in Ref. 45) may indicate a role for the two acidic residues in Me 2ϩ ion-binding in AglB and related glycosyl hydrolases. Substrate Discrimination by Sucrose-6-P Hydrolase and 6-Phospho-␣-glucosidase-Sucrose-6-P and its five phosphorylated linkage isomers have recently been prepared and characterized by 1 H and 13 C NMR spectroscopy (12). The availability of these derivatives in substrate amount permitted specificity and kinetic analyses to be carried out with highly purified sucrose-6-P hydrolase and AglB. These studies establish unequivocally that sucrose-6-P hydrolase hydrolyzes only sucrose-6-P to form glucose-6-P and fructose. The specificity of sucrose-6-P hydrolase for its single substrate (sucrose 6-phosphate) is noteworthy because it suggests that their reciprocal molecular recognition (as a prerequisite to fission of the intersaccharidic linkage to glucose-6-P and fructose) is unique, not even tolerating minor changes in the linkup of the two sugars, as for example those realized in the five isomeric glucosyl-fructoses. In contrast, the 6-phospho-␣-glucosidase (AglB), which is in-  (52). The conformation of sucrose-6-phosphate emerging from a 1000-ps MD simulation in a box containing 641 water molecules (right) is so similar to that of sucrose in water that a Glc-2-O⅐ ⅐ ⅐H 2 O⅐ ⅐ ⅐O-1-Fru water bridge is likewise inferred. The dotted contours refer to the solvent-accessible surface into which ball-and-stick models have been inserted; for easier comparison, the glucosyl moiety is kept in the same orientation.
FIG. 9. Solvent-accessible surface of sucrose-6-P in frontopened form with ball-and-stick model insert (top) as set against those of the nine other disaccharide-6-phosphates (bottom) superimposed on each other with the ␣-D-glucose-6-P portion (left half) kept in the same orientation. The slender form of the fructose moiety of sucrose-6-P (top, right half) renders the shape of the molecule different; notably it is more compact than that of the other disaccharide-6-phosphates. duced by growth of K. pneumoniae on the five glucosyl-fructoses, appears to be less specific and is tolerant of a considerable variation in both the structure and size of the O-linked aglycone. Indeed, the NAD ϩ and metal ion-dependent phospho-␣glucosidase hydrolyzed not only the 6Ј-phosphoglucosyl-fructoses but also the phosphorylated derivatives of related ␣-linked disaccharides such as maltose-6Ј-P, isomaltose-6Ј-P, and maltitol-6-P. Remarkably, AglB was unable to hydrolyze sucrose-6-P. Explanations for enzyme specificity and substrate discrimination must reside in the molecular geometries and polarities of the individual disaccharide phosphates and/or in the threedimensional structures of the two enzymes. Presently, only a structural model based on threading methods has been proposed for those enzymes (including sucrose-6-P hydrolase) that by sequence-based alignment are assigned to Family 32 of glycosyl hydrolases (57). Moreover, only a preliminary x-ray analysis has been reported for one enzyme member (GlvA from B. subtilis, (58)) of Family 4, to which AglB is assigned. Thus, we were led to probe the substrates with respect to structure, molecular shape, and polarity for clues to understanding the specificity of the two phosphoglucosyl hydrolases. From the markedly different molecular geometries of the phosphorylated disaccharides in solution (Figs. 8 -10), one might reasonably assume that shape recognition (by the respective binding domains) may be an important determinant of enzyme specificity. Another and conceivably more significant contribution to substrate discrimination may originate from differences in the distribution of hydrophobic and hydrophilic regions over the contact surfaces of the disaccharide phosphates. In eliciting the sweetness response, for example, sucrose is believed to dock onto the taste bud receptor protein via its hydrophobic region (59), which, on the basis of the calculated MLP profiles, encompasses the entire outer surface side of the fructose moiety (51,59). The same docking procedure is expected for sucrose-6phosphate at the active site of sucrose-6-P hydrolase, inasmuch as the MLP profile of sucrose and its 6-phosphate (Fig. 10, top center) are essentially the same, i.e. a pronounced hydrophilic 6-phosphoglucosyl part (blue areas) facing a distinctly hydrophobic (yellow) fructose portion. Fig. 10 shows the MLPs of sucrose-6-P and its five isomeric 6Ј-phosphoglucosyl-fructoses in the fully closed (upper portion) and in the front-side-opened form with ball-and-stick model inserts (lower portion). The MLP patterns of the five 6Ј-phosphoglucosyl-fructoses, albeit having essentially identical hydrophilic (blue) glucose-6-P halves, clearly differ from sucrose-6-P with respect to the shape, intensity, and distribution of their hydrophobic (yellow) surface domains (Fig. 10). These may perhaps be the major factors that prevent docking of the isomeric phosphates at the sucrose-6-P binding site of sucrose-6-P hydrolase and, conversely, that preclude binding of sucrose-6-P to the active site of phospho-␣-glucosidase.
Conclusion-This study and our earlier paper (12) are the first reports of bacterial growth on the five isomers of sucrose. However, genetic units similar to the Agl region of K. pneu- FIG. 10. Molecular lipophilicity patterns (MLPs) of sucrose-6-P (top center entry) and its five isomeric 6-phosphoglucosyl-fructoses in fully closed and front-side-opened form with ball-andstick model inserts. The relative hydrophobicity portraits were mapped in color-coded form onto their individual contact (solvent-accessible) surfaces with the colors ranging from dark blue (most hydrophilic areas) to yellow-brown (hydrophobic domains). moniae are present in the genomes of B. subtilis, F. mortiferum, and E. coli (Fig. 11). The phospho-␣-glucosidase(s) of these species are clearly homologous (Fig. 5), and the PTS transporter (AglA) has extensive homology with GlvC of B. subtilis, MalB of F. mortiferum, and Glv(CB) of E. coli. The gene organization is similar in the three Gram-negative species, but for B. subtilis (Gram-positive) the gene order is reversed and a gene glvR, which encodes a regulatory protein, separates the phospho-␣-glucosidase and PTS genes (60,61). Our recent finding that F. mortiferum can also grow on the sucrose isomers 4 suggests that genes homologous to aglA and aglB may be prerequisites for bacterial growth on these compounds. Parenthetically, it may be noted that neither of these genes has been found during sequencing of the S. mutans genome, and these deficiencies may explain the inability of this organism to metabolize the sucrose isomers.