Pen and Pal Are Nucleotide-Sugar Dehydratases That Convert UDP-GlcNAc to UDP-6-Deoxy-d-GlcNAc-5,6-ene and Then to UDP-4-Keto-6-deoxy-l-AltNAc for CMP-Pseudaminic Acid Synthesis in Bacillus thuringiensis*♦

Background: CMP-pseudaminic acid formation and role in Bacillus thuringiensis flagellin glycosylation are unknown. Results: Seven enzymes are required to convert UDP-N-acetylglucosamine to CMP-pseudaminic acid; UDP-N-acetylglucosamine is converted to UDP-6-deoxy-d-N-acetylglucosamine-5,6-ene and then to UDP-4-keto-6-deoxy-l-N-acetylaltrosamine. Conclusion: Two previously undescribed enzymes initiate the CMP-pseudaminic acid pathway. Significance: Identifying a unique CMP-pseudaminic acid pathway in a Gram-positive bacterium may provide new opportunities to control bacterial flagellin glycosylation and pathogenicity. CMP-pseudaminic acid is a precursor required for the O-glycosylation of flagellin in some pathogenic Gram-negative bacteria, a process known to be critical in bacterial motility and infection. However, little is known about flagellin glycosylation in Gram-positive bacteria. Here, we identified and functionally characterized an operon, named Bti_pse, in Bacillus thuringiensis israelensis ATCC 35646, which encodes seven different enzymes that together convert UDP-GlcNAc to CMP-pseudaminic acid. In contrast, Gram-negative bacteria complete this reaction with six enzymes. The first enzyme, which we named Pen, converts UDP-d-GlcNAc to an uncommon UDP-sugar, UDP-6-deoxy-d-GlcNAc-5,6-ene. Pen contains strongly bound NADP+ and has distinct UDP-GlcNAc 4-oxidase, 5,6-dehydratase, and 4-reductase activities. The second enzyme, which we named Pal, converts UDP-6-deoxy-d-GlcNAc-5,6-ene to UDP-4-keto-6-deoxy-l-AltNAc. Pal is NAD+-dependent and has distinct UDP-6-deoxy-d-GlcNAc-5,6-ene 4-oxidase, 5,6-reductase, and 5-epimerase activities. We also show here using NMR spectroscopy and mass spectrometry that in B. thuringiensis, the enzymatic product of Pen and Pal, UDP-4-keto-6-deoxy-l-AltNAc, is converted to CMP-pseudaminic acid by the sequential activities of a C4″-transaminase (Pam), a 4-N-acetyltransferase (Pdi), a UDP-hydrolase (Phy), an enzyme (Ppa) that adds phosphoenolpyruvate to form pseudaminic acid, and finally a cytidylyltransferase that condenses CTP to generate CMP-pseudaminic acid. Knowledge of the distinct dehydratase-like enzymes Pen and Pal and their role in CMP-pseudaminic acid biosynthesis in Gram-positive bacteria provides a foundation to investigate the role of pseudaminic acid and flagellin glycosylation in Bacillus and their involvement in bacterial motility and pathogenicity.


investigate the role of pseudaminic acid and flagellin glycosylation in Bacillus and their involvement in bacterial motility and pathogenicity.
Bacillus thuringiensis is a Gram-positive bacterium that has been isolated from diverse habitats, including soil, water, dust, plants, cadavers, and the intestines of insects (1)(2)(3)(4)(5). B. thuringiensis belongs to the Bacillus cereus group, which includes the two most notable pathogens B. cereus and Bacillus anthracis, the causal agents of food poisoning and anthrax, respectively (6). Members of this group have similar genetic backgrounds but are distinguished by their host specificities and their pathogenicity (7). For example, B. thuringiensis is characterized by its ability to form crystalline proteins (Bt toxin) that are lethal to many insects (8). Indeed, B. thuringiensis is used widely to control agricultural pests, including Lepidoptera, Diptera, and Coleoptera sp. (9), and may also have a role in controlling Anopheles gambiae, the principal vector of malaria (10).
B. thuringiensis as well as other Bacillus species are known to exist in the gut microflora of numerous insects (11,12). In this environment, some of these bacteria lose their flagella and become attached to the intestinal epithelium of insects (13). Recent cell biology studies have identified B. thuringiensis israelensis flagellum structures (7), and genomic analyses revealed that the bacterium has all of the protein components required to make flagellum and promote motility (14 -16). Nevertheless, the role of flagellum structures in Bacillus and its involvement in delivering the Bt toxin to the insect gut, or in pathogenicity, remains to be determined.
Flagellin, one of the proteins of the flagellum apparatus, is a globular protein monomer that stacks helically in the form of a hollow cylinder to build the filament of the flagellum (17)(18)(19). Studies of Bacillus sp. PS3 flagellin have shown that it is O-glycosylated, although the sugar involved in this modification was not identified (20). The results of preliminary studies in our laboratory 2 have indicated that in B. thuringiensis israelensis the flagellin is glycosylated with a sugar residue having a mass consistent with a diacetylaminotetradeoxy-nonulosonic acid.
Here, we report the identification of the B. thuringiensis pse operon and the functional characterization of the enzyme cluster that converts UDP-GlcNAc to CMP-Pse. Seven enzymes are required for the conversion of UDP-GlcNAc to CMP-Pse in B. thuringiensis (Fig. 1B), unlike Gram-negative bacteria that complete this reaction with six enzymes. Two dehydratase-like activities, which we named Pen and Pal, initiate the Pse pathway in B. thuringiensis. We used a combination of NMR spectroscopy and mass spectrometry to show that Pen converts UDP-D-GlcNAc to UDP-2-acetamido-6-deoxy-␣-D-xylo-hexopyranose-5,6-ene (herein abbreviated UDP-6-deoxy-D-GlcNAc-5,6-ene) and that Pal converts UDP-6-deoxy-D-GlcNAc-5,6-ene to UDP-2-acetamido-2,6-dideoxy-␤-L-arabino-hex-4-ulose (herein abbreviated UDP-4-keto-6-deoxy-L-AltNAc). Only one enzyme, PseB, is required to convert UDP-D-GlcNAc to UDP-4-keto-6-deoxy-L-AltNAc in Gram-negative bacteria. The identification of Pen and Pal and the CMP-Pse biosynthetic pathway in B. thuringiensis provides a basis for determining the biological role of Pse in flagellin glycosylation in Gram-positive bacterium.

EXPERIMENTAL PROCEDURES
Strains and Culture Conditions-The B. thuringiensis israelensis ATCC 35646 used in this study was stored in 16% glycerol at Ϫ80°C, streaked onto agar medium, and grown for 18 h at 30°C. The media (agar or liquid) used were Luria Bertani (LB) (10 g/liter tryptone, 5 g/liter yeast extract, 10 g/liter NaCl) and LB-Lennox (10 g/liter peptone, 10 g/liter NaCl, 5 g/liter yeast extract). Escherichia coli DH10B cells were used for cloning, and Rosetta2(DE3)pLysS (Novagen) cells were used to express protein.
His 6 -tagged Protein Expression and Purification-Rosetta2-(DE3)pLysS strains harboring expression plasmids were grown at 37°C and 250 rpm in 250 ml of LB supplemented with chloram- In the B. thuringiensis pathway, two enzymes are needed. First, Pen converts UDP-GlcNAc with enzyme-bound NAD ϩ to UDP-6-deoxy-D-GlcNAc-5,6-ene, and then Pal converts it to UDP-4-keto-6-deoxy-L-AltNAc. From this point, both pathways carry out similar enzymatic reactions (although the amino acid sequence for each specific enzyme is not conserved between species, see Table 1) leading to the final product of CMP-Pse. B, organization of the seven-genes pse operon and flanking regions in B. thuringiensis israelensis ATCC 35646. The locus number for each enzyme-encoding gene in the operon is shown.

TABLE 1
Amino acid sequence similarity between various proteins involved in the synthesis of CMP-pseudaminic acid phenicol (35 g/ml) and kanamycin (50 g/ml) for the pET28a and pET28b vectors or with chloramphenicol (35 g/ml) and spectinomycin (50 g/ml) for the pCDF vector. Gene expression was induced when cell A 600 reached 0.6 by adding 0.5 mM isopropyl 1-thio-␤-D-galactopyranoside. After induction, cells were grown for 18 h at 18°C and 250 rpm and then harvested by centrifugation (6,000 ϫ g) for 10 min at 4°C. The cell pellets were washed with water and then suspended in 10 ml of lysis buffer (50 mM Tris-HCl, pH 7.4, 10% (v/v) glycerol, 1 mM EDTA, 2 mM DTT, and 0.5 mM PMSF). Cells were lysed by sonication (31), and after centrifugation (6,000 ϫ g for 15 min at 4°C), the supernatant was supplemented with 1 mM DTT and 0.5 mM PMSF and re-centrifuged at 20,000 ϫ g for 30 min at 4°C. An aliquot (5 ml) of the supernatant was applied to a nickel-Sepharose fast-flow column (GE Healthcare; 2 ml of resin packed in a polypropylene column; inner diameter 1 ϫ 15 cm). Each column was pre-equilibrated with buffer A (50 mM Tris-HCl, pH 8, 10% (v/v) glycerol, 100 mM NaCl). The column was washed with 30 ml of buffer A containing 20 mM imidazole and then with 10 ml of buffer A containing 40 mM imidazole. His-tagged proteins were eluted with 5 ml of buffer A containing 250 mM imidazole. The eluates containing these proteins were divided into small aliquots, flash-frozen in liquid nitrogen, and kept at Ϫ80°C. The concentration of each protein was determined (32) from the A 280 nm (with ⑀ ϭ 14,200 cm Ϫ1 M Ϫ1 for His 6 RBTH_04253, ⑀ ϭ 40,120 cm Ϫ1 M Ϫ1 for RBTH_04254His 6 ). Proteins were separated by SDS-12.5% PAGE and visualized by staining with Coomassie Blue.
Enzyme Reactions-The activity of recombinant His 6 RBTH_ 04253 (herein referred to as Pen) was examined by HILIC-HPLC with UV or electrospray ionization-mass spectrometry (ESI-MS) detection and by time-resolved 1 H NMR spectroscopy. For HPLC-based assays, the total reaction volume was 50 l and included 50 mM Tris-HCl, pH 7.6, 1 mM UDP-GlcNAc, and about 1.5 g of purified Pen. Reactions proceeded for 1 h at 30°C, followed by enzyme inactivation (95°C for 2 min) and extraction with 50 l of chloroform (30). An aliquot (20 l) of the upper aqueous layer was removed and mixed with acetonitrile (40 l) and 0.5 M ammonium acetate, pH 5.3 (2 l). A portion (20 l) was analyzed using HILIC coupled to a ESI-MS/MS mass spectrometer or a UV diode array detector. ESI-MS/MS was performed using a Shimadzu LC-MS/MS IT-TOF MS system operating in the negative ion mode with a Nexera UFPLC LC-30AD pump, autosampler (Sil30), and column heater (set to 37°C). HPLC with UV detection was carried out with an Agilent 1260 pump system equipped with an autosampler, column heater (set at 37°C), and diode array detector (A 261 nm ). Enzyme reactions were separated on an Accucore 150-amid HILIC column (150 ϫ 4.6 mm, 2.6 m particle size, ThermoScientific) using 40 mM ammonium acetate, pH 4.3 (solvent A), and acetonitrile (solvent B). The column was equilibrated at 0.4 ml/min with 25% solvent A and 75% solvent B prior to sample injection (20 l). Following injection, the HPLC conditions were 0 -1 min, 0.4 ml/min with 25% solvent A, 75% solvent B and then a gradient to 50% solvent A and 50% solvent B over 24 min. The flow rate was then increased to 0.6 ml/min with a gradient to 25% solvent A and 75% solvent B over 5 min. The column was then washed for 5 min with 25% solvent A, 75% solvent B before the next injection. HPLC peaks of enzymatic reaction products detected by their A 261 nm (maximum for UDP-sugars) were collected, lyophilized, and suspended in D 2 O (99.9%) for NMR analysis or in H 2 O for MS/MS analysis.
The activity of recombinant RBTH_04254His 6 (herein refer as Pal) was examined by HILIC-HPLC-UV, ESI-MS and time- resolved proton NMR. The 50 l of HPLC-based assays included 50 mM Tris-HCl, pH 7.6, about 1 mM of the product (purified by HILIC) formed by Pen and 8.5 g of purified Pal. The reaction proceeded for 1 h at 30°C, followed by inactivation (95°C for 2 min) and chloroform extraction.
NMR-based Pen assays in a total volume of 180 l were performed in a mixture of 90 l of D 2 O and 90 l of H 2 O containing 50 mM Tris-HCl, pH 7.6, 1 mM UDP-GlcNAc, 0.1 mM 2,2dimethyl-2-silapentane-5-sulfonate (DSS) (which served as NMR reference), and ϳ3 g of purified Pen. The reaction mixture was transferred to a 3-mm tube and time-resolved 1 H NMR spectra were then obtained for up to 2 h at 30°C using a Varian DirectDrive 600-MHz spectrometer equipped with a cryogenic probe. Data were acquired before the addition of the enzyme at time 0 (t 0 ). After adding the enzyme, data acquisition was started after ϳ5 min to provide sufficient time to optimize magnet shimming. Sequential one-dimensional proton spectra (32 scans each) with presaturation of the water resonance were acquired over the course of the enzymatic reaction. All NMR spectra were referenced to the resonance of DSS set at 0.00 ppm.
NMR-based Pal assays in a total volume of 180 l were performed in 90 l of D 2 O and 90 l of H 2 O containing 50 mM Tris-HCl, pH 7.6, ϳ1 mM of the product formed by Pen, 0.1 mM DSS, and about 17 g of purified Pal. Time-resolved 1 H NMR acquisition was as described above.
Kinetic Studies-The linear dependence of enzyme concentration with respect to initial velocity was established by changing the protein concentration and maintaining the substrates at constant concentration. K m values were determined using regression analysis (nonlinear) with Prism software from plots of initial velocities versus varied substrate concentrations (10, 20, 40, 60, 80, 100, 150, 300, and 500 M UDP-GlcNAc).
Gel Filtration Assays-Purified Pen or Pal protein (in volume of 300 l) was mixed with 50 M UDP-GlcNAc and chromatographed on a Superdex-75 column (1 cm inner diameter ϫ 30 cm, GE Healthcare) at a flow rate of 0.5 ml/min, using 50 mM Tris-HCl, pH 8, 150 mM NaCl buffer. Peaks observed at 215 nm were collected and assayed for enzyme activity. The column was calibrated using a molecular mass standard (Bio-Rad).
Determination of Pen and Pal Enzyme-bound Co-factors-Purified Pen (600 l of 0.63 mg/ml) and purified Pal (1.2 ml of 1.62 mg/ml) were heated for 5 min at 95°C and then centrifuged at 14,000 ϫ g for 1 min. The supernatant was collected and concentrated to 100 l using a SpeedVac (Savant SC110). A 30-l aliquot was analyzed by HILIC-ESI-MS/MS for the detection of NAD ϩ or NADP ϩ . Both positive mode and negative mode were used with 25% collision-induced dissociation energy for MS/MS fragmentation.
NMR Spectroscopy Used to Characterize Product Structures-The HPLC peak corresponding to the enzymatic product of Pen was collected, lyophilized, dissolved in D 2 O (99.9%), and analyzed by two-dimensional NMR. Proton chemical shifts were assigned by a correlation spectroscopy (COSY) experiment and verified by a total correlation spectroscopy (TOCSY) experiment. Protonated carbon chemical shifts as well as multiplicities were determined by a multiplicity-edited heteronuclear single quantum coherence (HSQC) experiment. The chemical shift of nonprotonated carbon C5Љ was assigned by a heteronuclear multiple bond correlation (HMBC) experiment.
To characterize the structure of the products formed by Pal , the reaction assay of Pal was extracted with chloroform, and the aqueous phase was chromatographed on a Q15 anion-exchange column as described (33). The peak corresponding to the product was collected, lyophilized, suspended in 99.9% D 2 O, and analyzed by COSY, TOCSY, HSQC, and HMBC experiments.
LC-MS/MS Analysis-Reaction products were chromatographed on a HILIC column using a Shimadzu CBM-20A HPLC system equipped with an autosampler, coupled to a Shimadzu LCMS-IT-TOF ESI mass spectrometer operated in the negative mode. The HILIC conditions were the same as the HPLC conditions described under "Enzyme Reactions." Enzyme products were identified based on their retention time, the mass of their parent ion, and their mass spectral fragmentation pattern.

CMP-Pseudaminic Acid Operon in B. thuringiensis Consists
of Seven Genes-We first performed a BLAST search using amino acid sequences of known bacterial enzymes involved in CMP-Pse formation to identify B. thuringiensis genes that may have a role in CMP-Pse formation. The BLAST approach, however, identified several misleading gene targets. For example, the first enzyme in the pseudaminic acid pathway in H. pylori is a bi-functional UDP-GlcNAc 4,6-dehydratase/5-epimerase (PseB) that forms UDP-4-keto-6-deoxy-L-AltNAc. PseB has 46% amino acid sequence similarity (e-value 3e Ϫ84 ) to RBTH_05809 of B. thuringiensis israelensis ATCC 35646 and to Bc3750 of B. cereus ATCC 14579. However, Bc3750 was found not to encode a 5-inverting 4,6-dehydratase activity. 4 RBTH_04253 (i.e. Pen) has a 40% amino acid sequence similar-ity to PseB (e-value 7e Ϫ72 , see Table 1). Adjacent to the Pen is RBTH_04254 (Pal), annotated as a dehydratase (Fig. 1B), and has a low sequence identity to PseB (24%, see Table 1). RBTH_04255 (Pam), which has low (35%; e-value of 8e Ϫ71 ) amino acid identity with the functional aminotransferase PseC from H. pylori, follows the two annotated dehydratases. RBTH_04258 (Pdi) has low amino acid identity (30%; e-value of 4e Ϫ11 ) to the functional N-acetyltransferase PseH, whereas RBTH_04257 (Phy) has only 26% amino acid identity (e-value of 3e Ϫ9 ) to H. pylori PseG; RBTH_04259 (Ppa) shares sequence identity (41%; e-value of 1e Ϫ79 ) to H. pylori pseudaminic acid synthase PseI, and RBTH_04256 (Pcp) protein that shares very low sequence identity (29%; e-value 0.001) to functional pseudaminic acid cytidylyltransferase PseF from Campylobacter jejuni, although no similarity was detected compared with H. pylori PseF (see Table 1). Although the encoded proteins in this B. thuringiensis operon share an overall low amino acid sequence identity to known functional genes involved in CMP-Pse synthesis, we decided to clone and characterize these proteins because sugar (Pse) modification of flagella proteins has not been reported previously in Bacillus. Below, we describe that seven enzymes are required to form CMP-Pse from UDP-GlcNAc in B. thuringiensis (Fig. 1, A and B), unlike Gramnegative bacteria that complete this reaction with six enzymes (27). Biochemical Characterization of CMP-Pseudaminic Acid Operon in B. thuringiensis-Each of the seven predicted Pse pathway genes were cloned from B. thuringiensis and expressed in E. coli as recombinant His 6 -tagged proteins. Each protein was purified over a nickel-affinity column. Individual protein bands for recombinant Pen (40 kDa), Pal (37.1 kDa), Pam (47.6 kDa), Pcp (39.5 kDa), Phy (42.8 kDa), Pdi (24 kDa), and Ppa (40.7 kDa) were detected by SDS-PAGE (Fig. 2A). The ability of the recombinant enzymes to catalyze the various steps in the formation of CMP-Pse was then tested. As UDP-GlcNAc is the only commercially available substrate, we used the product formed by one enzyme as the substrate for the next enzyme. For example, recombinant Pen and Pal proteins were reacted with UDP-GlcNAc, and the product was then used as the substrate for Pam, the putative L-Glu:C4Љ-transaminase. The UDP-4amino-sugar formed together with acetyl-CoA was the substrate for Pdi, an acetyl-CoA:C4Љ-N-acetyltransferase. The resulting product (UDP-2,4,6-trideoxy-2,4-diNAc-L-altrose) then served as the substrate for the UDP-sugar hydrolase Phy, whose reaction product, along with phosphoenolpyruvate and Ppa, led to Pse. In the final reaction, Pse was reacted with CTP and recombinant Pcp to generate CMP-Pse (structure shown in Fig. 2B). The progression of each reaction was monitored by LC-MS-based assays.
HILIC-ESI-MS analysis of the products of the final enzymatic step (Fig. 2C) gave two peaks with retention times of 8.2 and 11.2 min and [M Ϫ H] Ϫ at m/z 333.12 and 638.16, respectively. These two values suggest a mass for a 9-carbon sugar and its nucleotide derivative CMP-diacetylaminotetradeoxy-nonulosonic acid, respectively. NMR analysis (Table 2) established that the peak eluting at 11.2 min is indeed CMP-5,7-diacetamido-3,5,7,9-tetradeoxy-L-glycero-␣-L-manno-2-nonulopyranosonic acid (CMP-Pse). The enzymatic product is not CMPlegionaminic acid (CMP-Leg) or other CMP-9-carbon sugars such as CMP-Leg 4-and 8-epimers, as the coupling constant between H5Љ and H6Љ (1.5 Hz) is small (expected for CMP-Pse) and not large (10.3 Hz) (expected for CMP-Leg). Fragmentation of the ion at m/z 638.16 gave two ions with m/z 322.04 and 413.08, (Fig. 2D) corresponding to CMP and pseudaminic acid 1-phosphate, respectively, and is consistent with the presence of CMP-Pse. Likewise, the peak eluting at 8.2 min (Fig. 2C)  We conclude that the seven-gene cluster between RBTH_ 04253 and 04259 in B. thuringiensis israelensis ATCC 35646 encodes all the enzymes required to convert UDP-GlcNAc to CMP-Pse (see Fig. 1A). Thus, we call this operon the Bti_Pse operon. Because the initial steps of the reaction involve two uncharacterized dehydratase-like enzyme activities, we fully characterized the activities of Pen and Pal.
Characterization of a Unique UDP-Sugar, UDP-6-Deoxy-D-GlcNAc-5,6-ene Formed by Pen, a UDP-GlcNAc 4-Oxidase/ 5,6-Dehydratase/4-Reductase-Our early analysis indicated that in B. thuringiensis two dehydratase-like proteins are required to initiate conversion of UDP-GlcNAc to a product that can be used for CMP-Pse formation. We therefore studied these two enzymes in more detail. Purified recombinant Pen (His 6 RBTH_04253) was reacted with UDP-GlcNAc, and the products formed were analyzed by HILIC HPLC with UV detection. A new peak with a reten-tion time of 14.5 min (Fig. 3A) was detected; a negative control reaction gave no new product. The peak was collected and examined by ESI-MS and MS/MS. The negative ion mode mass spectrum contained an ion [M Ϫ H] Ϫ at m/z 588.06 (Fig. 3B). MS/MS analysis of this ion gave two fragments at m/z 323 and 403 consistent with UMP and UDP, respectively, suggesting that the product was a UDP-sugar. The neutral loss of 18 atomic mass units suggests the UDP-GlcNAc product lacks the mass of water, implying that Pen is a dehydratase. However, further analysis of the enzyme product demonstrated that the enzyme is not a "regular" 4,6-dehydratase because no UDP-4-keto-sugar product was detected. Time-resolved 1 H NMR of the reaction showed the conversion of UDP-GlcNAc to a new product (labeled C) with a quartet of signals at 5.56 ppm (Fig. 3C and see zoom below). We also observed that this product lacks signals for a C6-methyl and a C4-keto sugar. Most known 4,6-dehydratases form a 4-keto sugar with C6-methyl group (i.e. UDP-4-keto-6-deoxy-sugar). Moreover, there was a new signal with a chemical shift of 4.09 ppm, which likely arose from the proton linked to C4Љ of the UDP-sugar. This suggests that Pen is a 4-oxidase/5,6-dehydratase/4-reductase and forms a product that has a double bond between C5Љ and C6Љ (i.e. 5,6-ene). A detailed explanation of the elucidation of the structure of this product is provided below.
NMR Characterization of UDP-6-Deoxy-D-GlcNAc-5,6-ene-Because the structure of the product from the Pen reaction could not be determined from the MS or time-resolved NMR experiments, the peak that eluted from the HILIC column at 14.5 min was collected and fully characterized by one-and two-dimensional NMR spectroscopy. The one-dimensional proton NMR spectrum (Fig. 4) indicated that the product was UDP-6-deoxy-D-GlcNAc-5,6-ene. Proton and carbon chemical shifts and proton coupling constants of the sugar moiety are listed in Table 2. The coupling constant between H1Љ and H2Љ was small (3.1 Hz), indicating an ␣-linkage to the phosphate of UDP. The coupling constant between H1Љ and the phosphorus of the diphosphate was 7.6 Hz, which is consistent with a D-sugar. The coupling constants between H2Љ and H3Љ (9.9 Hz), along with H3Љ and H4Љ (9.5 Hz), confirmed that the sugar had the gluco configuration. The H2Љ chemical shift at 4.18 ppm and the methyl proton resonance of the N-acetyl group (2ЉNAc) at 1.96 ppm are consistent with an acetamido moiety at the C2Љ. Two very diagnostic signal peaks for the C6Љ-methylene protons of product UDP-6-deoxy-D-GlcNAc-5,6-ene were identified at 4.87 and 4.89 ppm. These peaks were assigned to H6aЉ and H6bЉ of the 5,6-ene moiety of the product. A COSY experiment showed the correlation of protons that were two and three bonds apart (Fig. 5A). Surprisingly connectivity between H4Љ and H6Љ was also observed in this experiment, even though a correlation between protons that are four bonds apart is typically not observed (34). Proton chemical shifts were assigned and verified by a TOCSY experiment (Fig. 5B). Protonated carbon chemical shifts, as well as multiplicities, were established by an HSQC experiment (Fig. 5C). The signal of C6Љ gave reverse phase from C1Љ, C2Љ, C3Љ, and C4Љ. This suggests that protons at the 6Љ position are methylene protons (-CH 2 -), whereas other protons in the sugar moiety are methine protons (-CH-). The chemical shift of nonprotonated carbon C5Љ was assigned by an HMBC experiment (Fig. 5D), and the connectivities of C5Љ to H1Љ, H4Љ, and H6Љ were observed in the spectrum. These results, when taken together, provide evidence that Pen is a UDP-GlcNAc 4-oxidase/ 5,6-dehydratase/4-reductase that converts UDP-GlcNAc to UDP-6-deoxy-D-GlcNAc-5,6-ene.
dehydratase/4-reductase. We now describe the function of Pal, the second enzyme encoded by the pse operon of B. thuringiensis. The UDP-6-deoxy-D-GlcNAc-5,6-ene formed by recombinant Pen was purified by HPLC and used as a substrate for recombinant Pal. Two products (labeled K and H in Fig. 6A) eluted from the HILIC column as a broad UV 261 peak between 15 and 17 min and a retention time of ϳ16.5 min. A negative enzymatic reaction containing unrelated His 6 -tagged purified enzyme yielded no product. For initial characterization of the products (K and H), the broad peak was collected and analyzed by ESI-MS and MS/MS. Two ions at m/z 588.01 and 606.03 (Fig. 6B) were detected. Interestingly, the new UDP-sugar product (K) has a mass identical to UDP-6-deoxy-D-GlcNAc-5,6-ene. Their different elution times (16.5 versus 15.5 min, respectively) suggest that the product and substrate are chemically different. MS/MS analysis of each parent ion gave two ion fragments having m/z 323 and 403 values that are consistent with UMP and UDP, respectively. This indicated that the products were UDP-sugars.
To gain further insight into Pal-enzyme activity, we monitored the reaction by time-resolved 1 H NMR (Fig. 6C). Two products with a quartet of signals at 5.74 and 5.555 ppm in the anomeric region of the NMR spectrum appeared over the reaction time. The anomeric signal at 5.555 ppm (see 60-min time point) was clear but overlapped at early reaction time points (5-30 min) with the anomeric signal of the substrate, UDP-6-deoxy-D-GlcNAc-5,6-ene (labeled C, H-1Љ), at 5.56 ppm. The H-4Љ proton signal of the substrate (labeled C, H-4Љ) decreased over time while the amounts of product depicted by other proton signals (labeled H and K) increased. Two signals appeared at the 6-deoxy region around 1.55 and 1.37 ppm. These signals correspond to C6-methyl protons (H-6Љ) of product K (UDP-4-keto-6-deoxy-L-AltNAc) and H (hydrated form of K) sugar moiety, respectively. Proton and carbon chemical shifts and proton coupling constants of H are listed in Table 2.
Pen and Pal Together Produce UDP-4-Keto-6-deoxy-D-GlcNAc and UDP-4-Keto-6-deoxy-L-AltNAc-Another interesting aspect of these two Bacillus enzymes was the formation of an additional product when the enzymes were incubated together rather than in a sequential manner (see illustration in Fig. 8). When Pen and Pal were combined and incubated with UDP-GlcNAc, an additional product was formed besides K and H. We used 1 H NMR spectroscopy to monitor the reaction of recombinant enzymes with the substrate, UDP-GlcNAc, over 8 h to determine the nature of this new product (Fig. 7A, labeled  F). Two signals at 5.74 and 5.555 ppm corresponding to the anomeric protons of K and H appeared within 1 h and started to disappear after 3 h. Similarly, two signals at 1.55 and 1.37 ppm corresponding to the 6-deoxy protons of K and H also disappeared after 3 h. A new signal at 5.47 ppm developed gradually over the 8-h period as did a signal at 1.24 ppm. We have assigned these signals to the anomeric proton and H6Љ of the sugar moiety of UDP-2-acetamido-2,6-dideoxy-␣-D-xylo-hex-4-ulose (UDP-4-keto-6-deoxy-D-GlcNAc, product F), which is the 5-epimer of UDP-4-keto-6-deoxy-L-AltNAc. UDP-4keto-6-deoxy-D-GlcNAc was formed only when both Pen and Pal were added to the reaction; it was not detected when the two enzymes were added sequentially (Fig. 7B). To verify that FIGURE 6. Analyses of UDP-4-keto-6-deoxy-L-AltNAc, the enzymatic product of Pal by UV-HPLC, mass spectrometry, and time-resolved 1 H NMR. A, purified Pal was reacted with UDP-6-deoxy-D-GlcNAc-5,6-ene ("C," the purified enzymatic product of Pen) to yield at least two new UV-broad HPLC peaks labeled "K," "H," and marked by arrows. The top HPLC trace in A corresponds to purified product C standard, and the bottom trace is the negative control. an assay of purified irrelevant protein with C. The K and H peaks (eluted from HILIC column at 16 min) were collected and characterized by NMR (see Table 2). B, MS analysis in negative mode gives two [M Ϫ H] Ϫ ions at m/z 588.1 and 606.0 corresponding to two forms of UDP-4-keto-6-deoxy-L-AltNAc, 4-keto and its hydrated forms, respectively. C, time-resolved 1 H NMR showing Pal enzymatic conversion of UDP-6-deoxy-D-GlcNAc-5,6-ene (C) to UDP-4-keto-6-deoxy-L-AltNAc. As the reaction proceeds, two molecular species are produced, a 4-keto form and a 4-keto-hydrated form of UDP-4-keto-6-deoxy-L-AltNAc. The selected chemical shift region of the proton NMR spectrum that corresponds to the H-1Љ anomeric proton of enzymatic reactant and product is shown between 5.80 and 5.40 ppm. Note the H-4Љ peak of the substrate ("C") shown at 4.09 ppm is converted to new product K and H (the 4-keto and its hydrated form). The far right panel (between 1.60 and 1.30 ppm) shows the methyl groups (H-6Љ) belonging to the hydrated (H) and the keto (K) forms of the product UDP-4-keto-6-deoxy-L-AltNAc.
Enzymatic Properties of Pen and Pal-The pH-and temperature-dependent activity of Pen and Pal and kinetic data for these enzymes are summarized in Table 3. Pen has its highest activity in phosphate buffer at pH 7.64; however, high activity was also observed in MOPS-NaOH, pH 7.6, Tris-HCl, pH 7.4, and with the slight acidic condition in MES buffer at pH 6.85. The optimal temperature for Pen activity was at 37°C; however, the activities of Pen ranging from 30 to 50°C were similar and varied by no more than 10%. As for Pal, we were unable to separate the products from the substrates through the HILIC column. Consequently, we only measured the optimal pH and temperature for Pal by comparing the peak reduction of the substrates.
Comparative analyses of the kinetics between PseB from C. jejuni and H. pylori (35,36) with Pen from B. thuringiensis showed that Pen K m for UDP-GlcNAc has a similar affinity for the substrate (Table 3). However, Pen turnover (k cat /K m ) was 209, which is much higher than PseB, suggesting that the B. thuringiensis Pen converts UDP-GlcNAc into a product more efficiently.
To determine whether the 4-oxidase activities of Pen and Pal require NAD ϩ or NADP ϩ , and to address whether Pen enzyme functions as a dimer or forms a functional complex with Pal, we performed a series of experiments ( Fig. 9) using size-exclusion chromatography on Superdex-75. The proteins were preincubated with and without substrate (UDP-GlcNAc). The proteins were either loaded on the column individually or were loaded in a mix together. Protein peaks were then collected and tested for activity and analyzed by SDS-PAGE. Pen eluted from the column in two peaks (peak a and b, Fig. 9A). One peak migrated as a protein of 31.9 kDa (presumably a monomer), and the second peak eluted in the region for a protein with a mass of about 192.8 kDa. The earlier peak (presumably a pentamer, tetramer, or dimer of dimers) but not the latter peak had enzymatic activity when supplied with UDP-GlcNAc (Fig. 9B). Conversely, Pal eluted from the column as a single peak (peak d, Fig. 9A), migrating 87.6 kDa presumably as a dimer. This peak was only enzymatically active when supplemented with NAD ϩ (Fig. 9B). No Pal activity was obtained by adding NADP ϩ . Taken together, we propose that Pal activity involves the co-factor NAD ϩ for oxidation/reduction cycle of the enzyme product. The nature of the oligomeric state of Pen and Pal, however, is speculative, and the chromatographic mobility could be influenced by the shape of the protein.
We tested whether Pen and Pal interacted with each other; however, gel filtration assays showed when both enzymes were mixed together and chromatographed (Fig. 9A, bottom panel), each enzyme had the same elution pattern as the individual protein. Each peak was collected from the column and resolved on SDS-PAGE (Fig. 9B). A single peptide band was observed, (compare lanes a, b, and d with aЈ, bЈ, and dЈ). No difference in elution profiles was obtained by adding UDP-GlcNAc or with UDP-GlcNAc ϩ NAD ϩ to the combined enzymes. In addition, when different ratios of Pen, Pal, and UDP-GlcNAc were chromatographed (Fig. 9C), the peak signal increased proportionally. Taking together, the results suggest that Pen and Pal do not form a complex under this experimental setup.
No data indicating that Pen requires a cofactor for activity was obtained by size-exclusion chromatography (Fig. 9B). Nevertheless, we suspected that this protein binds NAD ϩ or NADP ϩ so tightly that it would only be released when denatured. Therefore, we heat-inactivated purified Pen as well as Pal and examined the products released by LC-ESI-MS/MS. NADP ϩ with [M ϩ H] ϩ ion m/z at 743.9, and diagnostic MS/MS ion fragments (603.9, 621.9, 489.9, and 409.9) were detected when Pen was denatured (Fig. 10A), whereas NAD ϩ with m/z at 664.0 (MS/MS 523.9, 541.9, 232.0, and 427.9) was detected when Pal was denatured (Fig.  10B). The inability of the size-exclusion column to separate bound-NADP ϩ from Pen suggests a strong interaction between FIGURE 7. Co-incubation of Pen and Pal with the UDP-GlcNAc (G) yields four molecular species as determined by time-resolved 1 H NMR. A, when co-incubation reaction was carried out up to 8 h, in addition to the keto and hydrated forms of UDP-4-keto-6-deoxy-L-AltNAc, another predominant hydrated form of UDP-4-keto-6-deoxy-D-GlcNAc was observed. The selected chemical shift region for the anomeric protons of reactant and product is shown between 5.75 and 5.35 ppm, and the diagnostic H-6Љ peak of the dual enzyme final products "F" is shown between 1.70 and 1.20 ppm. The peaks annotated * belong to a signal of Pal's hydrated product ("H") as an intermediate, and the peaks annotated ** belong to signal peaks of Pal's keto product ("K"). B, 2-h reaction of the single enzyme Pal incubated with UDP-6deoxy-D-GlcNAc-5,6-ene, showing only UDP-4-keto-6-deoxy-L-AltNAc was produced but no UDP-4-keto-6-deoxy-D-GlcNAc. The peak labeled with # belongs to a contaminant. the protein and its co-factor. By comparison with Pal, it is likely that within the Pen protein more amino acid residues are involved in the hydrogen bond connection between the co-factor and the enzyme. Together, these experiments provide evidence that NAD ϩ is a co-factor required for the oxidase/reductase activity of the Pal enzyme.

DISCUSSION
We have identified an operon in B. thuringiensis israelensis ATCC 35646 (Fig. 1, A and B) containing seven genes that encode enzymes involved in the conversion of UDP-GlcNAc to CMP-Pse. The product of each enzymatic reaction, UDP-6-deoxy-D-GlcNAc-5,6-ene, UDP-4-keto-6-deoxy-L-AltNAc, UDP-4-amino-4,6-dideoxy-N-acetyl-␤-L-altrosamine (UDP-4-amino-6-deoxy-L-AltNAc), UDP-2,4-diacetamido-2,4,6-trideoxy-␤-L-altropyranose (UDP-2,4,6-trideoxy-2,4-diNAc-L-altrose), 2,4-diacetamido-2,4,6-trideoxy-␤-Laltropyranose (2,4,6-trideoxy-2,4-diNAc-L-altrose), Pse, and CMP-Pse were characterized by LC-MS/MS and NMR spectroscopy. B. thuringiensis is unusual as it uses two dehydrataselike enzymes to initiate the conversion of UDP-GlcNAc to UDP-4-keto-6-deoxy-L-AltNAc. The known functional PseB from the Gram-negative bacteria uses only one enzyme to catalyze this reaction (27,37). In a sequential reaction pathway, Pen was incubated with UDP-GlcNAc, and Pal was then added. In the co-incubation pathway, both enzymes Pen and Pal were incubated together with UDP-GlcNAc. In the sequential reaction, UDP-GlcNAc was converted to UDP-6-deoxy-D-GlcNAc-5,6-ene (labeled as "C") by Pen, followed by Pal converting C to keto and hydrated forms of L-configured UDP-4-keto-6-deoxy-L-AltNAc (labeled as "K" and "H"). Note that products K and H were able to undergo C4Љ transamination reaction by Pam leading to product A, UDP-4-amino-6-deoxy-L-AltNAc. In the co-incubation pathway, UDP-GlcNAc was converted to a mixture of four products, including C, K, H, and a novel product F, a D-configured UDP-4-keto-6-deoxy-D-GlcNAc. After a long period of incubation, the product would be predominantly F, as the other products, including C, K, and H, would be consumed. Note that product F, UDP-4-keto-6-deoxy-D-GlcNAc, was not a substrate for Pam; hence Pam was unable to C4Љ transaminate this D-configured UDP-sugar. a Kinetic data of pseB in C. jejuni was from Ref. 35. b Kinetic data of pseB in H. pylori was from Ref. 36. c For Pen, optimal pH assays were determined using phosphate buffer in which Pen yielded the highest activity compared with MOPS-NaOH, Tris-HCl, and MES buffer. For Pal, optimal pH assays were determined using Tris-HCl buffer. d Optimal temperature assays were determined using Tris-HCl for both Pen and Pal. e The reaction was determined by HPLC-UV after a 5-min 30°C incubation for Pen. f The active forms of proteins eluted from the size-exclusion column were 192.8 kDa for Pen and 87.6 kDa for Pal, presumably suspected to be a tetramer and a dimer, respectively. mediate 5Љ-epimerization between UDP-4-keto-6-deoxy-L-AltNAc and UDP-4-keto-6-deoxy-D-GlcNAc. Pal shares low amino acid sequence identity to the dTDP-Glc 4,6-dehydratase from B. anthracis strain Ames. However, Pal is not acting on various NDP-sugars that were tested, including UDP-GlcNAc, UDP-GalNAc, and UDP-Glc. Pal shares even lower amino acid sequence identity to PseB and is unable to act on UDP-GlcNAc. Insight into the amino acid residues involved in the catalytic activities of Pen and Pal will require the use of site-directed mutagenesis in combination with a description of the three-dimensional structures of the enzymes.
In B. thuringiensis, two enzymes are required to generate the UDP-4-keto-6-deoxy-L-AltNAc. This reaction also generates UDP-6-deoxy-D-GlcNAc-5,6-ene. It is not known whether the UDP-GlcNAc derivative participates in other metabolic pathways. Interestingly, UDP-6-deoxy-D-GlcNAc-5,6-ene is an intermediate formed by TunA during the synthesis of tunicamycin by Streptomyces species (38). Whether or not the pathway of an analogue exists in B. thuringiensis remains unknown. Although the TunA enzyme structure was solved (38), very low amino acid sequence similarity (24% and e-value 0.004) exists between the Pen and the TunA protein, suggesting these genes are evolutionarily not related and perhaps arose from different ancestral genes. With regard to Pal, we are currently unaware of another single activity identical to this enzyme.
Our kinetics data (k cat/ K m ) for Pen suggest that it converts UDP-GlcNAc into product more efficiently (Table 3) compared with TunA and PseB (27,38). However, the product itself inhibited the Pen forward reaction. This resulted in a maximum ϳ10% conversion of substrate to product. Pal, however, rapidly converts the 5,6-ene to the 4-keto derivative and thereby facil-itates almost full conversion of UDP-GlcNAc to UDP-4-keto-6-deoxy-L-AltNAc. The formation of UDP-4-keto-6-deoxy-D-GlcNAc when Pen and Pal were combined together and reacted with UDP-GlcNAc suggests that Pal can be a standalone C5Љ-epimerase, albeit of low efficiency. Incubating C. jejuni PseB with UDP-GlcNAc for periods up to 15 h also resulted in the C5Љ epimerization UDP-4-keto-6-deoxy-L-Alt-NAc to UDP-4-keto-6-deoxy-D-GlcNAc (37); the exact mechanism to explain this remains unknown.
Our characterization of the first two enzymes (Pen and Pal) in the Bti-pse operon provides insight into the metabolic pathway leading to the formation of Pse in B. thuringiensis. However, we cannot exclude the possibility that the products formed by these enzymes also participate in other metabolic pathways. Additional studies are now required to determine whether other members of the Bacillaceae also have these enzymes. Understanding the molecular factors that control the flux of UDP-GlcNAc to different metabolic pathways in bacteria is important, as this nucleotide-sugar is a precursor used for the formation of wall polysaccharides and peptidoglycan as well as other glycans.