The Phosphonopyruvate Decarboxylase from Bacteroides fragilis*

The Bacteroides fragilis capsular polysaccharide complex is the major virulence factor for abscess formation in human hosts. Polysaccharide B of this complex contains a 2-aminoethylphosphonate functional group. This functional group is synthesized in three steps, one of which is catalyzed by phosphonopyruvate decarboxylase. In this paper, we report the cloning and overexpression of the B. fragilis phosphonopyruvate decarboxylase gene (aepY), purification of the phosphonopyruvate decarboxylase recombinant protein, and the extensive characterization of the reaction that it catalyzes. The homotrimeric (41,184-Da subunit) phosphonopyruvate decarboxylase catalyzes (kcat = 10.2 ± 0.3 s–1) the decarboxylation of phosphonopyruvate (Km = 3.2 ± 0.2 μm) to phosphonoacetaldehyde (Ki = 15 ± 2 μm) and carbon dioxide at an optimal pH range of 7.0–7.5. Thiamine pyrophosphate (Km = 13 ± 2 μm) and certain divalent metal ions (Mg(II) Km = 82 ± 8 μm; Mn(II) Km = 13 ± 1 μm; Ca(II) Km = 78 ± 6 μm) serve as cofactors. Phosphonopyruvate decarboxylase is a member of the α-ketodecarboxylase family that includes sulfopyruvate decarboxylase, acetohydroxy acid synthase/acetolactate synthase, benzoylformate decarboxylase, glyoxylate carboligase, indole pyruvate decarboxylase, pyruvate decarboxylase, the acetyl phosphate-producing pyruvate oxidase, and the acetate-producing pyruvate oxidase. The Mg(II) binding residue Asp-260, which is located within the thiamine pyrophosphate binding motif of the α-ketodecarboxylase family, was shown by site-directed mutagenesis to play an important role in catalysis. Pyruvate (kcat = 0.05 s–1, Km = 25 mm) and sulfopyruvate (kcat ∼ 0.05 s–1; Ki = 200 ± 20 μm) are slow substrates for the phosphonopyruvate decarboxylase, indicating that this enzyme is promiscuous.

Bacteroides fragilis is a human pathogen that causes intraabdominal abscess formation in its host (1,2). The bacterial capsular polysaccharide complex is the major virulence factor for abscess formation. The capsular polysaccharide complex is composed of three distinct polysaccharides, polysaccharides A, B, and C (3)(4)(5)(6). These polysaccharides consist of repeating units that contain a zwitterionic motif of negative and positive charged groups. The zwitterionic charge motif plays an essential role in the induction of the host defense response, which leads to abscess formation. The 2-aminoethylphosphonate (AEP) 1 unit of polysaccharide B contributes the positive and negative charges that form the zwitterionic motif (see Fig. 1).
AEP is the phosphonate counterpart to phosphoethanol amine, a common lipid polar head-group. The P-C bond of AEP is resistant to both chemical and enzymatic hydrolysis. The AEP unit is found in proteins (7), lipids (8, 9 -12), and polysaccharides (4) located at the cell surfaces in certain parasitic organisms. These AEP conjugates either participate in host infection, as in the case of the B. fragilis polysaccharide B, or they are responsible for the persistence of the parasite within the host.
The presence of AEP in the B. fragilis polysaccharide B was first demonstrated by NMR structural analysis (3). More recently, the polysaccharide B biosynthetic pathway gene locus was sequenced (13). Three genes, aepX, aepY, and aepZ, which encode proteins that share significant sequence identity with the three enzymes of the AEP biosynthetic pathway, are included within this locus (Fig. 2). To our knowledge, this is the first known example of the AEP biosynthetic pathway gene cluster in a bacterium. Moreover, the opportunity now exists for the isolation of the three pathway enzymes for mechanistic study and inhibitor design. Because the AEP pathway enzymes are not present in humans, they are excellent candidates for drug targeting.
What is presently known about the AEP biosynthetic pathway and the three enzymes that catalyze it has resulted from a "patch-work" effort. The AEP biosynthetic pathway was first discovered in Tetrahymena pyriformis (14 -17). In this organism, AEP is incorporated into phosphonolipids, which form the plasma membrane. The AEP pathway was shown to consist of the three steps depicted in Fig. 3. In the first step of the reaction, P-enolpyruvate is converted to phosphonopyruvate (Ppyr). P-enolpyruvate mutase, the enzyme that catalyzes this step, has been isolated from several different organisms and characterized (18 -20). The conversion of P-enolpyruvate to Ppyr is thermodynamically unfavorable (K eq ϳ 1 ϫ 10 Ϫ3 ), and thus, the ensuing decarboxylation step catalyzed by Ppyr decarboxylase is required to drive the Ppyr-forming reaction forward. The T. pyriformis Ppyr decarboxylase is membranebound and difficult to isolate for characterization (21). What we do know about this enzyme derives from the study of the bacterial enzyme which functions in biosynthetic pathways leading to bialaphos, fosfomycin, and phosphinothricin tripeptide in Streptomyces hygroscopicus (22), Streptomyces wendmorensis (23,24), and Streptomyces viridomogenes (25), respectively. In these pathways, P-enolpyruvate mutase and Ppyr decarboxylase collaborate to form the common precursor phosphonoacetaldehyde (Pald). The S. hygroscopis Ppyr decarboxylase has been isolated and its native size and its cofactor requirement (viz. thiamine pyrophosphate and Mg(II)) have been defined (26).
As with the Ppyr decarboxylase of Tetrahymena pyriformis, the third enzyme of the pathway, Pald transaminase proved to be membrane-bound and difficult to isolate. Using partially purified enzyme, Kim (21) was able to demonstrate catalysis of pyridoxal phosphate-dependent transamination of Pald with L-alanine functioning as the ammonium group donor. A related transaminase can be found in bacteria adapted for the use of AEP as an alternate source of carbon, nitrogen, and phosphorous. Because the physiological reaction is catalyzed in the direction of Pald formation, this enzyme has become known as AEP transaminase (27)(28)(29)(30)(31). It, too, is dependent on pyridoxal phosphate; however, the ammonium group acceptor is not pyruvate but rather L-glutamate (32). Nevertheless, the bacterial AEP transaminase shares sufficient sequence identity with the B. fragilis Pald transaminase to indicate that the two enzymes share common ancestry.
The AEP pathway enzymes of B. fragilis have not been isolated for characterization. Our goal was to express the B. fragilis AEP pathway genes in Escherichia coli for purification and study of the three pathway enzymes. In this paper, we report the cloning and overexpression of the B. fragilis Ppyr decarboxylase gene (aepY), purification of the protein, and for the first time, an in-depth study of the Ppyr decarboxylase.

EXPERIMENTAL PROCEDURES
Materials-Thiamine pyrophosphate chloride (TPP), dihydro-␤-nicotinamide adenine dinucleotide (␤-NADH), yeast alcohol dehydrogenase, and the buffers used in protein purification and kinetic assays were purchased from Sigma and used without further purification. Phospho-nopyruvate and sulfopyruvate were synthesized according to the published methods (33,34). Recombinant Bacillus cereus phosphonoacetaldehyde hydrolase was purified as described previously (35). B. fragilis genomic DNA was purchased from American Type Culture Collection (ATCC 25285D). The primers used in PCR-based DNA amplification were custom synthesized at Invitrogen. For the cloning of the phosphonopyruvate decarboxylase gene, the sequence of the 5Ј to 3Ј primer was ATTCAGACGCATATGGTAAGTGTA and that of the 3Ј to 5Ј primer was TCTTTCTTTGGATCCTCATGAATGCGT, with the introduced restriction sites underlined. The enzymes used in DNA manipulation were purchased from Invitrogen and used with the buffers provided.
PCR-based Cloning of Phosphonopyruvate Decarboxylase Gene-The genomic DNA template was denatured (30 min at 94°C) before adding Pfu DNA polymerase. The target gene was amplified by 20 cycles of 94°C denaturation for 1 min, 55°C annealing for 50 s, and 73°C elongation for 3.5 min. The PCR product was purified by electrophoresis and digested using NdeI and BamHI restriction enzymes. The digest was ligated to an NdeI-and BamHI-cut pET 3a vector. The resulting clone, named bf-Pyrdecarb-pET 3a, was used to transform E. coli BL21(DE3) competent cells. The gene sequence was verified by DNA sequencing carried at the Center for Genetics in Medicine, University of New Mexico School of Medicine, Albuquerque, NM.
Protein Purification-E. coli BL21(DE3) cells, transformed with the named bf-Pyrcarb-pET 3a clone, were grown to 1 A 600 nm at 22°C in 1.2 liters ϫ 4-LB media containing 100 g/ml ampicillin. Following an 8-h induction period with 0.2 mM isopropyl-␤-D-thiogalactopyranoside, the cells were harvested by centrifugation at 6500 rpm, 4°C (all purification steps were carried out at 4°C except where noted). The cell pellet was suspended in 100 ml of 50 mM K ϩ HEPES containing 5 mM MgCl 2 and 1 mM dithiothreitol, pH 7.5 (referred to as buffer A, hereafter). Cells were lysed at 1000 p.s.i. in a French pressure cell and then centrifuged at 20,000 rpm for 30 min. The supernatant was subjected to ammonium sulfate precipitation. The 40 -85% fraction was collected by centrifugation and dissolved in buffer A for overnight dialysis against buffer A. The dialysate was chromatographed on a DEAE-Sepharose column (3.0 ϫ 60 cm) (equilibrated with buffer A) using a 1.4-liter linear gradient of KCl (0.15-0.60 M) in buffer A as eluant. The Ppyr decarboxylase-containing fractions (eluted at ϳ0.35-0.40 M KCl) were identified using the spectrophotometric activity assay (described in the following section) and analyzed by SDS-PAGE. The desired Ppyr decarboxylasecontaining fractions were combined and chromatographed on a hydroxylapatite column (3.0 ϫ 40 cm) equilibrated with buffer A, using a 1.4-liter linear gradient of phosphate (0 -0.25 M) in buffer A as eluant. Column fractions containing the Ppyr decarboxylase (eluted at ϳ0.08 -0.12 M phosphate) in Ͼ95% purity (as judged by SDS-PAGE analysis) were combined and concentrated in 50 mM K ϩ HEPES buffer containing 5 mM MgCl 2 , 1 mM MnCl 2 , and 1 mM dithiothreitol, pH 7.5 (buffer B) for storage at Ϫ80°C. Yield: 3.7 mg/g wet cell (or 22 mg of cell culture/liter).
Site-directed Mutants-The site-directed mutants E213A, D258A, and D260A were prepared by PCR and commercial primers using the clone bf-Pyrdecarb-pET 3a as a template. The PCR product was purified by electrophoresis and digested using NdeI and BamHI restriction enzymes. The digest was ligated to an NdeI-and BamHI-cut pET 3a vector and then used to transform competent E. coli BL21(DE3) cells. The gene sequence was verified by DNA sequencing carried out at the Ppyr Decarboxylase Molecular Size Determination-The molecular mass was calculated from the amino acid composition, derived from the gene sequence, by using the EXPASY Molecular Biology Server program Compute pI/MW (36). The molecular mass was measured by mass spectrometry (Biopolymer Mass Spectrometry Core Facility, Weill Medical College of Cornell University) and by SDS-PAGE (4% stacking gel and 12% separating gel). Commercial protein molecular weight standards were used to generate a plot of log M r versus distance traveled on the gel. The size of native Ppyr decarboxylase was estimated by gel filtration column chromatography (1.6 ϫ 60 cm, Amersham Biosciences Biotech Superdex 200-column, eluted at 4°C with 25 mM K ϩ HEPES, 0.15 M KCl, pH 7.5). Commercial protein molecular weight standards were used to generate a plot of log M r versus elution volume from the column.
Metal Ion Activation-The metal ion-free protein (3 mg/ml) was prepared by exhaustive dialysis against 50 mM K ϩ HEPES (pH 7.5, 4°C) containing 1 mM dithiothreitol and 20 mM EDTA. The dialyzed protein was then concentrated to 10 mg/ml for kinetic study.
where v is the initial velocity, V max is the maximum velocity, [S] is the substrate concentration (here it is 10 mM pyruvate), K m is the Michaelis-Menten constant of pyruvate, [I] is the inhibitor concentration and K i is the inhibition constant.

RESULTS AND DISCUSSION
Protein Purification and Size Determination-The DNA sequence of the cloned gene agreed with the published sequence (GenBank TM accession number AF285774_6). The recombinant Ppyr decarboxylase was purified to homogeneity (see Fig. 4) by using the 4-step protocol summarized in Table I in an overall yield of 3.7 mg/g wet cells. The steady-state kinetic constants for catalyzed decarboxylation of phosphonopyruvate are:k cat ϭ 10. The theoretical molecular weight of the phosphonopyruvate decarboxylase calculated from the amino acid sequence was 41,184, which agrees with the experimental molecular weight of 41,199, measured by matrix-assisted laser desorption ionization time-of-flight mass spectrometry. The subunit size estimated by SDS-PAGE analysis was 40 kDa compared with the native protein size of 120 kDa determined by gel filtration chromatography. Thus, the quaternary structure of Ppyr decarboxylase appears to be homotrimeric. (The S. hygroscopicus native Ppyr decarboxylase molecular size was reported as 135 kDa (26)).
Sequence Homologs-At the time of this writing, there are a total of seven Ppyr decarboxylase sequences listed in the Protein Data Bank. The Ppyr decarboxylase-encoding genes in B. fragilis (NCBI protein data bank ID code AAG26466) (378 amino acids), Bacteroides thetaiotaomicron (NCBI protein data bank ID code NP_810632) (374 amino acids), Amycolatopsis orientalis (NCBI protein data bank ID code CAB45023) (371 amino acids), and Clostridium tetani E88 (NCBI protein data bank ID code NP_782297) (376 amino acids) are positioned between the genes encoding homologs of P-enolpyruvate mutase and AEP transaminase. Pairwise sequence alignments made with the B. fragilis Ppyr decarboxylase (which activity has been demonstrated in this work) demonstrated 76, 35, and 43% sequence identity, respectively. The other three Ppyr decarboxylases (401, 384, and 397 amino acids long, respectively) function in the bialaphos, fosfomycin, and phosphinothricin tripeptide biosynthetic pathways of S. hygroscopicus (22), S. wendmorensis (23,24) and S. viridomogenes (25), respectively. A pairwise sequence alignment of these three Ppyr decarboxylases with the Ppyr decarboxylase from B. fragilis identified 34, 50, and 32% sequence identities, respectively.
A ClustalW-based sequence alignment of the seven Ppyr decarboxylase sequences identified 61 stringently conserved residues (16%). A total of 26 of the 61 stringently conserved residues are polar and, thus, are potential candidates for catalytic residues and for substrate-or cofactor-binding residues.
The B. fragilis Ppyr decarboxylase sequence was used as the query in a BLAST (Basic Local Alignment Search Tool) search of the gene data bank for protein homologs. The sulfopyruvate decarboxylase of the coenzyme M pathway, found in methaneforming bacteria (38), was identified as the closest homolog. Studies of the sulfopyruvate decarboxylase from Methanococcus jannaschii (39) have shown that the native enzyme is a dodecamer of 6 ␣-subunits (ComD 169 amino acids long; 34% sequence identity with the N-terminal half of the Ppyr decarboxylase) and 6 ␤-subunits (ComE 169 amino acids long; 39% sequence identity with the C-terminal half of Ppyr decarboxylase). It may be inferred that the ␣and ␤-subunits of the sulfopyruvate decarboxylase correspond to N-terminal and Cterminal domains of the Ppyr decarboxylase.
Sulfopyruvate decarboxylase (39) and Ppyr decarboxylase are more distant members of a family of TPP-and Mg(II)-dependent decarboxylases that includes acetohydroxy acid synthase/acetolactate synthase (40), benzoylformate decar-  boxylase (41), glyoxylate carboligase, indole pyruvate decarboxylase, pyruvate decarboxylase, the acetyl phosphate-producing pyruvate oxidase, and the acetate-producing pyruvate oxidase (42) (sequence identities between B. fragilis Ppyr decarboxylase and these proteins with a sequence range of 23-29%). The chemical reactions catalyzed by these enzymes are shown in Fig. 5. The common chemistry catalyzed by the family of enzymes is the decarboxylation of an ␣-keto carboxylate using the TPP cofactor as an "electron sink." Metal Ion and TPP Activation-Metal-free Ppyr decarboxylase, prepared by exhaustive dialysis, is inactive. TPP-free Ppyr decarboxylase, also prepared by exhaustive dialysis, is inactive. The K m for TPP activation was determined by measuring the initial velocity of the Mg(II) (5 mM)/Mn(II) (1 mM)-activated Ppyr decarboxylase-catalyzed decarboxylation of Ppyr (saturating at 50 M). The initial velocity data defined the k cat ϭ 9.7 Ϯ 0.3 s Ϫ1 and TPP K m ϭ 13 Ϯ 2 M. Other TPP-activated enzymes are known to bind TPP with K d values in the nanomolar to micromolar range (43,44).  6. The ClustalW sequence alignment representing the Ppyr decarboxylase-containing enzyme family. The individual sequences shown were picked as representatives of the subfamilies. The stringently conserved residues across all the subfamilies are highlighted in yellow, and the TPP signature motif is boxed. The polar residues interacting directly with Mg(II) or TPP are indicated by *, and the polar residues conserved across Ppyr decarboxylases and sulfopyruvate decarboxylases are highlighted in gray and marked by #. The residues are numbered for the B. fragilis Ppyr decarboxylase. PPyr_Bfra, B. fragilis Ppyr decarboxylase (NCBI protein data bank ID code AAG26466); SPyr_Meth, sulfopyruvate decarboxylase from Methanosarcina acetivorans str. C2A (NCBI protein data bank ID code NP_618188); Asyn_Ecol, the acetolactate synthase isozyme I large subunit from E. coli CFT073 (NCBI protein data bank ID code NP_756456); POxi_Ecol, pyruvate oxidase from E. coli O157 (NCBI protein data bank ID code NP_286643); BFdc_Pseu, benzoylformate decarboxylase from Pseudomonas putida (NCBI protein data bank ID code P20906).
FIG. 7. pH rate profiles of Ppyr decarboxylase catalysis. See "Experimental Procedures" for details. The apparent pK a of 6.1 Ϯ 0.1 and pK a of 8.5 Ϯ 0.1 were obtained by computer fitting the log k cat profile to Equation 2, and pK a of 6.7 Ϯ 0.2 and pK a of 7.7 Ϯ 0.2 were obtained by computer fitting the log (k cat /K m ) profile.
Mg(II) and TPP-binding Residues-The results described