Purification and properties of the Escherichia coli K-12 NAD-dependent nucleotide diphosphosugar epimerase, ADP-L-glycero-D-mannoheptose 6-epimerase.

The Escherichia coli K-12 NAD-dependent nucleotide-diphosphosugar epimerase, ADP-L-glycero-D-mannoheptose 6-epimerase, catalyzes the conversion of ADP-D-glycero-D-mannoheptose to ADP-L-glycero-D-mannoheptose. ADP-L-glycero-D-mannoheptose is a key intermediate of lipopolysaccharide inner core biosynthesis in several genera of Gram-negative bacteria. Sedimentation equilibrium and sodium dodecyl sulfate-polyacrylamide gel electrophoresis of the purified epimerase revealed that the native enzyme has a molecular mass of 240 kDa and a subunit molecular weight of 37,000 +/- 3,000. Lectin binding studies of the purified epimerase indicated that the protein is glycosylated. There was 1 mol of tightly bound NAD+ per enzyme subunit. Variable but small fractions of purified preparations of epimerase are highly fluorescent and contain NADH. The native enzyme can be resolved into apoenzyme and NAD+ by acidic ammonium sulfate precipitation. The catalytic activity can be reconstituted with the addition of NAD+ to the apoenzyme. Optimum pH range for enzyme activity is broad, between 5.5 and 9.5. It exhibits a temperature optimum at 42 degrees C. The Km and Vmax for the substrate is 0.1 mM and 46 mumol 30 min-1 mg-1, respectively. The native enzyme displays UV and fluorescence spectra that are consistent with the presence of enzyme bound NAD+. CD spectra of the holoepimerase indicate 11% alpha-helical and 36% beta-sheet structures.

* The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked  and increased permeability to a large number of hydrophobic agents, most importantly, antibiotics. Wild type strains usually exclude these hydrophobic agents, rendering them refractory to antibiotic treatments. The common occurrence of the epimerase in several genera of Gram-negative bacteria provides a viable strategy for targeting this protein in antibiotic therapy (5, 6). We previously reported the cloning and sequencing of the rfaD gene from E. coli, the identification of the promoter region and the transcription start site (7). We also reported the preliminary purification and characterization of the gene product. The N terminus of each monomer has the fingerprint sequence Gly-X-Gly-X-X-Gly, which is characteristic of the ADP-binding Papfold of FAD-binding and NAD-binding proteins (8). In this study, we investigated the catalytic properties as well as the quaternary structure of the epimerase. These studies were facilitated by the development of a rapid and simple purification procedure to obtain large quantities of the homogeneous enzyme.

Materials
The substrate, ADP-n-glycero-n-mannoheptose, was extracted and purified from a rfaD mutant (CL515), which accumulates this nucleotide, as described previously by Coleman (2). Blue 2-Sepharose CL-GB resin, phenylmethylsulfonyl fluoride, DNase I, RNase, n-galactose, streptomycin sulfate, lactate dehydrogenase kit, and NADH were purchased from Sigma. Octyl-agarose resin was from ICN ImmunoBiologicals and Sigma. The Lectin-Link kit was from Genzyme. [35S1Methionine was purchased from DuPont. Nucleotide sugars were obtained from Calbiochem.

Assay of AflP-L-glycero-o-mannoheptose 6-Epimerase
The epimerase assay was described previously by Coleman et al. (6). A typical assay mixture contained 0.1 M Tris acetate, pH 8.5, 25 1.1~ NAD, 1.25 m M MgCl,, 5 nmol ADP-D-glycero-o-mannoheptose, and enzyme, in a final volume of 50 pl. The reaction mixture was incubated at 37 "C for 30 min and was terminated by boiling for 3 min. The enzyme activity was determined by monitoring the formation of ADP-L-glycero-D-mannoheptose by high performance liquid chromatography. One unit of enzyme activity is defined as the epimerase activity capable of producing 1 nmol of ADP-L-glycero-n-mannoheptose in 30 min at 37 "C in 0.05 ml reaction mixture.

Determination of NAD
Epimerase-bound NAD was dissociated by perchloric acid treatment (9). The purified enzyme (639.6 pg) in 10 m M glycine-NaOH, pH 8.5 was treated with 35% perchloric acid at a ratio of 9:1, viv, and the mixture (200 pl) was incubated at 0 "C for 20 min (9). The protein precipitate was separated from the supernatant by centrifugation at 6000 x g for 5 min. NAD content in the precipitate and supernatant was determined by three different methods. The precipitate fraction of the epimerase was resuspended in 50 pl of 0.1 M NaOH and the NAD+ content was determined by a specific assay for pyridinium compounds, the methyl ethyl ketone procedure (9, 10). NAD+ content was determined in the perchloric acid supernatant fractions spectrophotometrically. Estimate of the NAD+ content was based on the A,,, values (see under "Results"). The NAD+ content of supernatant fractions was also determined by The presentation of the biosynthetic pathway using chemical monitoring its reduction to NADH by lactate dehydrogenase. The supernatant was adjusted to pH 8.9 with NaOH prior to lactate dehydrogenase assay. NADH formation was monitored spectrophotometrically at 340 nm.
Preparation of Apoenzyme and Reconstitution with NAD ADP-L-glycero-D-mannoheptose 6-epimerase was resolved into apoenzyme and NAD+ by treatment with acidic ammonium sulfate (pH 2.7) a t 0 "C as described by Gomi et al. (12) and Porter and Boyd (13). In a typical experiment, the apoenzyme was reconstituted by incubation with 200 p NAD+ or NADH a t room temperature for 20 min. Unbound NAD+ or NADH were removed from the reconstituted enzyme by gel filtration (13) on PD-10 columns (Sephadex G-25).
Other Methods Glycosyl residues on the epimerase were detected using the Genzyme Lectin-Link kit (i.e. an avididbiotin system) and the Western blot and visualization protocols provided by the manufacturer. The cyanogen bromide procedure used to cleave the epimerase protein was described previously by Matsudaira (14). Neutral sugar analysis was performed as described previously by Coleman (3).
Cell Extraction Crude extracts were prepared as described previously (7) from French pressates of E. coli strain CL627 (a K38 strain containing a plasmid-borne E. coli K-12 rfaD gene (i.e. pCG6) (7) which can be thermally induced to exclusively expressed, the rfaD gene product (i.e. the epimerase protein). The in vivo expression system employed to exclusively expressed a cloned gene, following thermal induction, was described previously by Tabor and Richardson (15). Cells were grown in LB medium or in a defined medium (i.e. for preparation of radiolabeled protein) as described previously (2,7). For enzyme purification, extracts were prepared from unlabeled cells (50 g, wet weight) and [3KSlmethionine-radiolabeled cells (4 g, wet weight).
Purification of ADP-L-glycero-D-mannoheptose 6-Epimerase Hydrophobic Znteraction Chromatography-The crude extract (see Table I) was applied to an octyl-agarose column equilibrated with TEM buffer (10 mM Tris, 10 mM EDTA, 0.1 mM P-mercaptoethanol, 1 p~ pepstatin A, 57 PM phenylmethylsulfonyl fluoride, pH 8.0). Proteins were initially eluted with TEM buffer, pH 8.0, followed by a stepwise gradient of KC1 from 0.3 M to 0.6 M in TEM buffer. The enzyme was eluted in the 0.6 M KC1 fraction. The protein fractions containing enzyme activity were pooled, desalted, and concentrated with Amicon cells.
Afinity Chromatography-The pooled protein fractions were applied to a blue Sepharose CL-GB column equilibrated with TEM buffer, pH 8.0. Proteins were first eluted with the equilibration buffer, followed by

TMLE I Purification of ADP-L-glycero-D-mannoheptose 6-epimerase
Step Spectroscopic and Analytical Methods Absorption spectra were recorded in a Hewlett-Packard 8452 diode array spectrophotometer at 23-25 "C. Fluorescence spectra were recorded on a Perkin Elmer MPF-3 spectrofluorimeter. Circular dichroism spectra were recorded at 25 "C, in a Jasco J-5OOC spectropolarimeter, using a DP-500 data processor and 1 cm length quartz cuvettes in 0.04 M sodium phosphate buffer, pH 7.2 (16). For CD experiments. the protein concentration was 0.2 mg/ml for the holo-and apoenzymes.
SDS-polyacrylamide gel electrophoresis was performed in 12% gel under reducing conditions as described by Laemmli (17). The subunit M, of epimerase was determined from the relative mobilities of protein standards. Gel filtration on a n high performance liquid chromatography Zorbax GF-250 column (9.4 x 250 mm) was used to estimate the molecular weight of the native protein. The molecular mass of the native protein was determined by the sedimentation equilibrium procedure described by Attri and Minton (18). Protein concentrations were estimated using the Biuret and Bradford assay methods (19,201. Chromatographic data were acquired and analyzed by an automated data collection system described by Minton and Attri (21).

Effect of Substrate Concentration on Enzyme Activity
The standard enzyme assay mixture as described was used with various concentrations of ADP-D-glycero-D-mannoheptose. Purified epimerase (65 ng) was used to initiate the reaction.

Enzyme Inhibition
Purified epimerase (5.2 pg) was preincubated with each inhibitor in an enzyme assay mixture of 50 pl at 25 "C for 15 min. The reaction was initiated by adding 4.4 nmol of ADP-D-glycero-D-mannoheptose. The enzyme activity was determined under standard assay conditions described above. Each determination was based on the results of triplicates. The effect of pH on the activity of purified epimerase was investigated over a pH range of 3.5-9.5 with four different buffers (0.1 M): acetate (pH 3.5-5.01, MES' (pH 5.5-6.5), HEPES (pH 7.0-8.01, and Tris acetate (pH 8.5-9.5). In each case, the enzyme assay was carried out with 5.2 pg of enzyme in a total volume of 50 pl. Each data point represents the average of duplicate determinations.
Dmperature Stability-The effect of different incubation temperatures on stability and activity was examined by preincubating reaction mixes containing purified epimerase (5.2 pg) for 1 min at the desired temperature, adding substrate, and continuing incubation at that temperature for 30 min. The standard enzyme assay mixture (50 pl) was used. Each data point represents the average of two or three determinations.

RESULTS
Purification of UP-L-glycero-o-mannoheptose 6-Epimeruse-An E. coli strain, CL627, which overproduces the epimerase (7), was used for purification of the enzyme. Following thermal induction, 12-16% of the total protein of this strain is epimerase. Purification procedure is outlined in Table I. Substantial enrichment of the epimerase was achieved by using octyl-agarose (Fig. 2), resulting in greater than 7-fold increase Fraction Number in specific activity. Subsequent chromatography with blue 2-Sepharose CL-GB resin resulted in a homogeneous preparation of the enzyme (Fig. 31, with an overall yield of 45%. Molecular Weight-Gel filtration studies (Fig. 4A) suggest a native protein molecular weight in the range of 230,000-250,000. Sedimentation equilibrium studies with the native epimerase indicated a molecular mass of 240,000. The subunit molecular weight was estimated to be 37,000 2 3000 (Fig. 4B). Thus, the native enzyme is composed of six identical subunits.
Carbohydrate Content-The purified epimerase is glycosylated as demonstrated by binding to concanavalin A ( C o d ) , and to a lesser degree to Datura stramonium agglutinin and wheat germ agglutinin (Fig. 5 A ) . The lectins used in this study have different carbohydrate binding specificities (22). ConA binding is consistent with the presence of mannose containing glycans while the binding of D. stramonium agglutinin and wheat germ agglutinin suggests the presence of N-acetyl-Dgiucosamine. The differential binding specificities gave indication regarding possible carbohydrate moieties in the epimerase. Preliminary neutral sugar analyses of the epimerase yielded fucose, mannose and galactose. This is consistent with region(s) of the protein, we prepared CNBr fragments of the epimerase. A single 14,000 cyanogen bromide fragment of the epimerase showed binding with the ConA lectin (Fig. 5C, lane  2'). An analysis of the deduced amino acid sequence of the rfaD gene product, previously reported by our laboratory (71, showed a stretch between Met-81 and Met-303 that corresponds to a calculated molecular weight of 14,000. Potential glycosylation sites in this small region of the epimerase polypeptide chain include 8 asparagines, 5 serines, and 7 threonines. Catalytic Properties of the Enzyme-The epimerase activity increases with increasing concentration of ADP-D-glycero-Dmannoheptose in a typical hyperbolic fashion (Fig. 6). A Michaelis constant (K,) of 0.1 m M was calculated for ADP-Dglycero-D-mannoheptose (Fig. 6, inset ). The corresponding maximum velocity (V,,,,,) of 46 pmol30 min-', mg" was determined from the same plot. ADP, ADP-glucose, ATP, and NADH were found to inhibit the enzyme activity (Table 11). Enzyme activity was completely inhibited by 0.1 mM of ADP and ADP-glucose. Inhibition was observed even at 0.02 mM, albeit to a lesser degree. ATP and NADH were less inhibitory and enzyme activity was only partially inhibited a t 1 m M and 0.1 m M levels. The optimum pH range for the enzyme (Fig. 7) is quite broad, ranging between 5.5 and 9.5. The result suggests that the enzyme activity can withstand wide fluctuation of pH. The epimerase activity exhibits a temperature optimum a t 42 "C ( Fig. 8), although the curve is not steep, suggesting a range of temperature stability.
Enzyme-bound NAD-Purified ADP-L-glycero-D-mannoheptose 6-epimerase was tested for the presence of NAD by several methods . The A,,dA,,, and A,,,,/A,,, ratios of the perchloric supernatants were 0.83 and 0.28, respectively, while authentic NAD under identical condition yielded similar ratios of 0.85 and 0.26. Specifically, following perchloric acid dissociation of 17 nmol of epimerase subunit, 16.8 nmol of NAD+ was recovered in the supernatant fraction. Thus, a 1:l correlation was found for epimerase subunit and NAD. A second determination of the enzyme-bound NAD+ content was performed by reduction to NADH with lactate dehydrogenase. Using this method, 31 pmol of NADH was detected in 32 pmol of epimerase subunit. As a control, treatment of the supernatant with Neurospora

Effect of various compounds on the ADP-t-glycero-n-mannoheptose 6-epimerase activity from E. coli K-12
Compound Enzyme activity" a t concentration (mM): Spectral Properties of Epimerase-The absorption spectrum of purified epimerase (Fig. 9A) displays two major maxima centered at 272 nm and 350 nm (see inset). An absorption maximum at 272 nm instead of the usual 278 nm absorption due to protein suggests that there is absorption due to nonprotein moiety bound to the enzyme. Free NADH absorbs at 340 nm, and may shift to higher wavelengths when it is bound to a protein. The absorbance at 350 nm (Fig. SA) may be due to NADH as well as 1 mol of NAD+ bound per subunit of the enzyme as determined by chemical analysis . The A,,,/A has previously been observed when another NAD+ containing enzyme, S-adenosylhomocysteine hydrolase, was reduced by adenosine or NaBH, (13). Fluorescence emission spectra of the epimerase and NADH are shown in Fig. 1OA. The emission maximum of the epimerase (curve 1 ) is 450 nm. The emission maximum of unbound NADH is 470 nm (curue 2). The fluorescence intensity of the epimerase-bound NADH is approximately 2-fold that of equimolar concentration of unbound NADH. The emission spectrum of epimerase-bound NADH is not only increased in intensity but the wavelength of the maximum emission is also shifted to a shorter wavelength (470 to 450 nm) relative to unbound NADH. Increased intensity and shift in wavelength for enzyme bounded NADH has been observed for beef heart muscle lactic dehydrogenase and horse liver alcohol dehydrogenase (25,26). Excitation fluorescence spectra (Fig. 10B) were also obtained for the epimerase and  Table 111. Apoepimerase was inactive in the standard epimerase assay but activity was restored following incubation with 200 V M NAD+. The specific activity of the reconstituted enzyme was consistently greater than 100% of the untreated native enzyme. It was also observed that suboptimal concentration of NAD+ (<0.1 mM) resulted in partial reactivation (52%) of the inactive apoenzyme. In contrast, NADH reconstituted enzyme resulted in only 15% of the activity of untreated native enzyme. However, this meager activation following the addition of NADH is probably the results of adventitious oxidation of NADH in solution. Fluorescence analysis of the NAD reconstituted enzyme, unlike the NADH and untreated native epimerase, showed no fluorescence when exposed to UV light (302-345 nm).
Secondary Structure of Epimerase-Circular dichroism spectroscopy, which is sensitive to the contribution of various secondary structural elements, was used to evaluate the overall conformation of the epimerase. Fig. 11 shows the far-ultraviolet CD spectra of holoenzyme and apoenzyme. The holoepimerase has an intense spectrum (curve 1 ), with double minima a t 208 and 222 nm and a maxima around 190-195 nm. Analysis of the holoenzyme CD spectrum indicates a protein with 11% a-helical and 36% P-sheet structures. The CD spectrum of the apoepimerase (curve 2) was greatly reduced in intensity with a single minimum around 215 nm; analysis of curve 2 indicates a predominant P-sheet structure (i.e. 45% p-sheet).

DISCUSSION
The goal of these studies was to characterize the physicochemical structure of an epimerase that is required for lipopolysaccharide core biosynthesis in several genera of Gram-negative bacteria. The collective data suggest that ADP-L-glycero- FIG. 11. Ultraviolet circular dichroism spectra of holo-and apoepimerase. Curves I and 2 are, respectively, howere digitized, downloaded and analyzed, loepimerase and apoepimerase. Spectra as previously described (131, in terms of secondary structure by least-squares fits using the PC-Mlab computer program (Civilized Software, Inc., Bethesda, MD). Protein concentrations (100-200 pg/ml) were estimated by absorption at D-mannoheptose 6-epimerase is similar to a group of epimerases (27, 28) that involves a NAD+-dependent redox catalysis. The inhibition of the epimerase by nucleotide sugars or nucleotide diphosphates and sugar mixtures is reminiscent of the reductive inactivation of UDP-epimerase by NADH, by UDP-sugars (several aldohexoses or aldopentoses) or by free sugars in the presence of UMP (11,(29)(30)(31). The observed reductive inactivation of UDP-galactose 4-epimerase has been shown to be directly related to the reduction of the tightly bound cofactor NAD+ (11,29).
Lipopolysaccharide is reported to contribute to the pathogenicity of enteric and nonenteric Gram-negative bacteria. Previously, we have reported (6) that the epimerase from E. coli shares significant structural and functional similarities with the epimerase from Pseudomonas aeruginosa. This conclusion is based on a number of observations including enzymatic activities, electrophoretic mobility of partially purified epimerase from €? aeruginosa and its cross-reactivity to antibody raised against the purified E. coli enzyme. Kontrohr and Kocsis ( 5 ) has reported the partial purification of a similar activity in Shigella that is required for the synthesis of L-glycero-D-mannoheptose.
L-Glycero-D-mannoheptose (heptose) is a common lipopolysaccharide component of the inner core of several genera of Gram-negative bacteria. The presence of heptose in the lipopolysaccharide of Gram-negative bacteria requires ADP-L-glycero-D-mannoheptose 6-epimerase activity. Heptoseless strains are less effective pathogens than wild type counterparts. Therefore, structural and functional studies of the epimerase may lead to novel antibacterial agents based on inhibition of the epimerase activity.