当前位置:首页 >> 药学 >>

Active site structure and stereospecificity of Escherichia coli pyridoxine-5'-phosphate oxidase.


doi:10.1006/jmbi.2001.5254 available online at http://www.idealibrary.com on

J. Mol. Biol. (2002) 315, 385±397

Active Site Structure and Stereospecificity of Escherichia coli Pyridoxine-5H-phosphate Oxidase
Martino L. di Salvo1, Tzu-Ping Ko2, Faik N. Musayev3, Samanta Raboni3 Verne Schirch3 and Martin K. Safo3*
Dipartimento di Scienze Biochimiche ``A. Rossi Fanelli'' and Centro di Biologia Molecolare del Consiglio Nazionale delle Ricerche ? La Sapienza, Rome Universita Italy
2 Institute of Biological Chemistry, Academia Sinica Taipei 11529, Taiwan 1

Institute for Structural Biology and Drug Discovery, Virginia Commonwealth University Richmond, VA 23219, USA

3

Pyridoxine-5H -phosphate oxidase catalyzes the oxidation of either the C4H alcohol group or amino group of the two substrates pyridoxine 5H -phosphate and pyridoxamine 5H -phosphate to an aldehyde, forming pyridoxal 5H -phosphate. A hydrogen atom is removed from C4H during the oxidation and a pair of electrons is transferred to tightly bound FMN. A new crystal form of the enzyme in complex with pyridoxal 5H -phosphate shows that the N-terminal segment of the protein folds over the active site to sequester the ligand from solvent during the catalytic cycle. Using (4H R)-[3H]PMP as substrate, nearly 100 % of the radiolabel appears in water after oxidation to pyridoxal 5H -phosphate. Thus, the enzyme is speci?c for removal of the proR hydrogen atom from the prochiral C4H carbon atom of pyridoxamine 5H -phosphate. Site mutants were made of all residues at the active site that interact with the oxygen atom or amine group on C4H of the substrates. Other residues that make interactions with the phosphate moiety of the substrate were mutated. The mutants showed a decrease in af?nity, but exhibited considerable catalytic activity, showing that these residues are important for binding, but play a lesser role in catalysis. The exception is Arg197, which is important for both binding and catalysis. The R197 M mutant enzyme catalyzed removal of the proS hydrogen atom from (4H R)-[3H]PMP, showing that the guanidinium side-chain plays an important role in determining stereospeci?city. The crystal structure and the stereospeci?city studies suggests that the pair of electrons on C4H of the substrate are transferred to FMN as a hydride ion.
# 2002 Academic Press

*Corresponding author

Keywords: pyridoxine-5 -phosphate oxidase; pyridoxal 5H -phosphate; pyridoxamine 5H -phosphate; X-ray crystallography; ?avin mononucleotide

H

Introduction
The terminal step in the biosynthesis of pyridoxal 5H -phosphate (PLP) in Escherichia coli is the oxidation of pyridoxine 5H -phosphate (PNP) to PLP, which is catalyzed by pyridoxine phosphate oxidase (PNPOx).1,2 PLP is used to activate a wide variety of apoenzymes involved in amino acid metabolism, as well as enzymes in several other metabolic pathways. In mammalian systems, which do not synthesize PLP, PNPOx is an importAbbreviations used: PLP, pyridoxal 5H -phosphate; PMP, pyridoxamine 5H -phosphate; PNP, pyridoxine 5H -phosphate; PNPOx, pyridoxine phosphate oxidase; SHMT, serine hydroxymethyltransferase. E-mail address of the corresponding author: msafo@hsc.vcu.edu
0022-2836/02/030385±13 $35.00/0

ant step in a PLP salvage pathway. In addition to using PNP as a substrate, the enzymes from both E. coli and eukaryotic sources catalyze the oxidation of pyridoxamine 5H -phosphate (PMP) to PLP.1,3 PNPOx has been puri?ed from sheep and pig brain, rabbit liver, and E. coli.1,3 ± 7 The most extensive studies have been done with the rabbit liver enzyme.3,4,8 ± 15 Recently, crystal structures for the E. coli enzyme have been published with and without PLP bound at the active site.16 ± 18 The structure ? of yeast PNPOx has been determined to 2.7 A (RCSB entry 1CI0). Currently, the protein data bank lists amino acid sequences for 15 PNPOx enzymes from various sources. There is considerable sequence homology between all of these enzymes, suggesting that they share a common fold and mechanism. The E. coli enzyme has been
# 2002 Academic Press

386 shown to exist in an open structure in the absence of bound PLP,17 and a partially closed structure with PLP bound.18 During the oxidation of either PNP or PMP, a pair of electrons and a hydrogen atom are removed from the C4H and the electrons transferred to a tightly bound FMN moiety, forming FMNH2. The electrons are then transferred to oxygen, forming H2O2 and regenerating the tightly bound FMN. Studies with the rabbit liver enzyme showed that it lacked stereospeci?city in the hydrogen atom removed from the prochiral C4H of PMP.15 There was nearly equal removal of either the proR or proS hydrogen atoms as judged by equal numbers of 3H atoms appearing in the H2O or PLP fractions when (4H R)-[3H]PMP was used as substrate. We have determined a new crystal structure form of PNPOx with bound PLP in which residues 5-19, which were not seen in previous structures, are shown to form a lid on the active site. In this new structure, we see that Arg14 and Tyr17 both interact with the bound PLP at the active site and could potentially be important either in binding af?nity or the mechanism of electron transfer. Kinetic constants were determined for several site mutants involved in PLP binding and the effect of these mutations on the stereochemistry of hydrogen removal from C4H of PMP was determined.

X-ray Structure of Pyridoxine-5 0 -phosphate Oxidase

Results
Structural properties of the monoclinic crystal form of PNPOx in complex with PLP A monoclinic crystal form of PNPOx in complex ? resolution, although with PLP diffracts to 2.07 A the crystals are small and have maximal size of about 0.03 mm ? 0.03 mm ? 0.8 mm. Suitable crystals for data collection are obtained after 6 to 12 months. The crystals belong to the space ?, group C2 with unit cell constants a ? 83.36 A  ? ? b ? 52.24 A, c ? 53.91 A, b ? 101.45 and with one monomer per asymmetric unit. We were unable to obtain a monoclinic crystal form without co-crystallizing with PLP. The secondary structure assignment and overall topology of the monoclinic crystal structure are identical with the trigonal structure, which has been determined with and without bound PLP.17,18 Two monomers, which constitute the functional dimer, interact extensively along one-half of each monomer, as discussed.17 Each monomeric subunit consists of two domains (1 and 2) folded into an eight-stranded b-sheet surrounded by ?ve a-helical structures. The larger domain 1 is formed by the b-sheets b1-b5 and two of the ?ve a-helices, a1 and a2. The smaller domain 2 is made up of the three remaining a-helices, a3, a4 and a5. To compare the PNPOx structures, dimeric models of the trigonal and monoclinic crystals are superimposed using the LSQ procedures of O.19 For the trigonal crystal, the model containing PLP is used, unless noted otherwise. The trigonal and

monoclinic crystal structures are matched with a ? for 396 Ca root-mean-square deviation of 0.51 A atoms, using residues 21-218 of the dimeric structure. If a more stringent criterion for LSQ-?tting of ? is used, the root-mean-square deviation 1.0 A ? for 364 Ca atoms. As between the models is 0.37 A shown in Figure 1, models from the trigonal and the monoclinic crystals ?t with minimal deviations, including the FMN, PLP and phosphate groups. These non-protein molecules interact with PNPOx by the same repertoire in both crystals, except some additional bonds involving the PLP and the N-terminal residues in the monoclinic crystal. Re?nement statistics for the monoclinic structure are recorded in Table 1. Asn208, the only residue found in the disallowed region of the Ramachandran plot, has glycine-like f and c dihedral angles of 67  and ?133  , respectively, for the second position in a type IIH turn.20 The monoclinic model shows a hitherto unseen structure of the N terminus (residues 5-19) (Figures 1 and 2). In the trigonal crystal, the ?rst residue that could be identi?ed was Gly20. Residues 8-13 constitute a short 310 helix with the carbonyl oxygen atoms of Gln8, Gln9 and Ile10 hydrogen bonded to the nitrogen atoms of Ala11, His12 and Leu13, respectively. Residues 14-18 are in extended conformation, stretching over the active site. The NE and NH2 atoms of Arg14, and the OH atom of Tyr17 form hydrogen bonds with two phosphate oxygen atoms and the aldehyde oxygen atom of PLP, respectively. There is no strong conservation of Arg14 and Tyr17 in other PNPOx sequences, but the other enzymes do show in this extended region of the N terminus complete conservation of residues that can form hydrogen bonds. There are a number of other hydrogen bonds between this N-terminal segment with the rest of the PNPOx dimer, including a bifurcated

Figure 1. Least-squares superposition of monoclinic PNPOx dimer structure (red) with that of the trigonal PNPOx dimer structure (purple). The FMN, PLP and phosphate molecules of the monoclinic PNPOx structure are shown in magenta, while those of the trigonal structure are shown in green. The structures were superimposed as described in the text. The Figure was drawn using MOLSCRIPT30 and Raster3D.31

X-ray Structure of Pyridoxine-5 0 -phosphate Oxidase Table 1. Data collection and re?nement statistics of the PNPOx crystal
A. Data collection statistics Space group Unit cell dimensions ?) a, b, c (A b (deg.) ?) Resolution (A No. measurements No. unique reflections I/sI Completeness (%) Rmerge (%)a B. Structure refinement ?) Resolution limit (A No. reflections R-factor for 95 % working data set Rfree for 5 % test data set rmsd from standard geometry ?) Bond (A Angles (deg.) Dihedral (deg.) Improper (deg.) ? 2) Average B-values (A All 2094 non-hydrogen atoms The 856 backbone atoms The 919 sidechain atoms The 31 FMN atoms The 16 PLP atoms The 5 phosphate atoms The 267 water molecules Ramachandran plot (%) Most favored region Additional allowed region Generously allowed Disallowed region ?) Estimated coordinate error (A By Luzzati plot By SigmaA plot C2 (b-unique) 83.36, 52.24, 53.91 101.45 40-2.07 (2.14-2.07) 32,505 13,510 8.3 (2.1) 96.6 (92.9) 7.7 (26.4) 40-2.07 (2.15-2.07) 13,510 (1403) 0.178 (0.293) 0.230 (0.349) 0.010 1.54 22.9 2.45 27.8 25.1 27.2 18.2 21.5 24.9 40.2 91.1 6.8 1.6 0.5 R-factor 0.22 0.30

387 ? 2 surface areas on the two dimers in conand 636 A ? 2 per tact. The total crystal contact area is 2550 A dimer, accounting for about 15 % of the dimer sur? 2). This is within the typical range of face (17,930 A 2 ? 1100-4400 A per molecule22 for crystal contact areas. In one of the dimers, only one subunit is involved (14 residues), while in the other both subunits are involved (11 and 2 residues). Table 2 lists speci?c bonds between residues in direct contact in the dimer. In addition, there are at least 31 interface water molecules. In the monoclinic crystal, each dimer is in contact with six neighbours, four related by the C-centering symmetry and two by unit translation along ?2 the c-axis. The ?rst type of interface buries 710 A 2 ? areas on the opposing dimers and and 694 A involves 17 and 16 residues, respectively. The ? 2 and involves 14 residues second type buries 494 A ? 2 per on each dimer. The total buried area is 3796 A dimer, or about 21 % of the dimer surface ? 2), indicating more involved molecular (18,293 A packing than the trigonal crystal form. Some speci?c bonds are listed in Table 2. There are at least 36 and 16 water molecules near the two interfaces. Among the 16 residues of the second dimer of the ?rst interface, ten are from one subunit and the remaining six are from another subunit. The latter encompass the N terminus, which is sandwiched between two dimers as shown in Figure 3. This explains why these residues are seen only in the monoclinic crystal. In addition, the second interface contains a hydrophobic patch involving the dyad-related side-chains of Leu27, Ala29 and Ile213. The peptide N and O atoms of Trp211 also form two water-bridged hydrogen bonds across the dyad axis.

Rfree 0.29 0.35

Numbers in parentheses refer to the outermost resolution bin. a Rmerge ? ?(hIi ? I)/?I.

salt-bridge between the side-chains of Arg15 and Asp49 of the other subunit. The side-chains of Ile10 and Leu13 in the 310 helix are involved in hydrophobic interaction with a non-polar patch formed by Leu39, Cys43, Pro50, Thr51, Leu71 and Tyr74 in the other subunit. Crystal packing The speci?c volumes, or Matthews coef?cients,21 of the trigonal and monoclinic crystals are 3.01 and ? 3/Da, respectively. This reduced volume in 2.38 A the monoclinic crystal cannot be explained by the extensive interactions between subunits within a dimer, because they are virtually identical for both ? 2 surface crystal forms, which involve about 2600 A 17 area on each subunit (see Safo et al. for a description of these contacts). Rather, the reduced speci?c volume in the monoclinic crystal is due to differences in dimer packing. In the trigonal crystal, each dimer is in contact with four neighbours, related by the 31 screw symmetry axis inside the unit cell body, via identical interfaces. Not includ?2 ing water molecules, the interface buries 639 A

Table 2. Speci?c bonds at lattice contacts of PNPOx crystals
Residue 1 Atom 1 Residue 2 Atom 2 ?) Distance (A A. Trigonal: Residues 1 and 2 belong to dimers related by x, y, z and ?y, x?y?1, z?1/3 Glu94 OE1 Arg23 NH1 2.97 Glu94 OE2 Arg23 NH2 2.70 Thr123 OG1 Ile213 O 2.59 Ser176 O Arg24 NE 3.00 Ser176 OG Leu27 O 2.64 B. Monoclinic 1: Residues 1 and 2 belong to dimers related by x, y, z and x?1/2, y?1/2, z Glu94 OE1 Leu7* N 2.87 Glu94 OE2 His12* NE2 2.71 Lys117 NZ Gln169 O 3.35 Glu119 OE1 Arg135 N 2.70 Ser176 OG Gln9* OE1 2.84 Arg206 O Gln169 NE2 3.00 Asn208 N Gln168 OE1 3.02 Asn208 ND2 Gln165 OE1 2.98 Arg206 O Gln168 NE2 3.24 C. Monoclinic 2: Residues 1 and 2 belong to dimers related by x, y, z and x, y, z?1 Leu27 O Ala29 N 2.63 Ala29 N Leu27 O 2.63 *Residues belonging to different subunits of the symmetryrelated dimer.

388

X-ray Structure of Pyridoxine-5 0 -phosphate Oxidase

Figure 2. Difference Fourier (2Fo ? Fc) maps of the monoclinic PNPOx crystal for the N-terminal region. The maps were calculated ? resolution using all re?ecat 2.07 A tions and phase angles of (a) the initial protein model after molecular replacement search and (b) the ?nal model re?ned by CNS. Maps were contoured at the 1.0 s level and superimposed with the re?ned model. The polypeptide sequence shown is DELQQIAHLRREYTKG. The associated FMN and PLP molecules are also shown. The Figures were produced using Bobscript32 and Raster3D.31

We were unable to obtain a monoclinic crystal form without co-crystallizing with PLP. The reason could be due to the fact that the monoclinic structure represents a completely closed form of PNPOx, in which the N-terminal segment makes a number of interactions with the substrate PLP. Additionally, the N terminus is extensively involved in lattice contacts. Without bound PLP, the N terminus would remain ?exible and disordered in solution. Therefore, it could not be packed properly into the monoclinic lattice due to lack of stabilizing inter-dimer contacts at the molecular interface. Most likely, it is the combination of these interactions involving the lattice and the bound PLP that have led to the stabilization of the N terminus and this explains why these residues are seen in the monoclinic crystal. In contrast, the N terminus in the trigonal crystal faces the bulk solvent and lacks the lattice contacts observed in the monoclinic crystal. This has therefore led to its disorder, even though the trigonal crystal also binds PLP. Active-site interactions We have previously discussed the active site in both open and partially closed forms.17,18 In the native trigonal PNPOx without any bound PLP, the active site is open, while in the trigonal crystal with PLP bound the C terminus of helix a3, helix a4, and the loop located between these two helices ? and rotated (residues 129 - 140) have shifted 1.4 A 2.7  toward the active-site cavity. This tertiary movement enables the residues Tyr129, Arg133 and Ser137 to come closer and make interactions with the PLP phosphate moiety. In addition, the

?exible turn located between the strands S6 and S7 (residues 193-199), has shifted and rotated about ? and 9.4  , toward the active site, and allows 2.4 A closer hydrogen bond interaction between PLP and the residues His199 and Arg197. In the monoclinic crystal, the FMN molecule and the disposition of protein side-chains about it remain identical as in the trigonal crystal. The substrate PLP molecule occupies the same position in the active site and has essentially unaltered conformation in both crystals. In addition the two crystals show similar interactions between the protein and the bound PLP, with only a few rearrangements of some active-site residues. However, there are additional N-terminal residues that contribute to interaction with the PLP in the monoclinic crystal. The speci?c interactions between the PLP and the monoclinic crystal protein residues at the active site are shown in Figure 4. Three hydrogen bonds are conserved, between the OP1, OP2, and O3H ? ), OH (2.5 A ? ), and atoms of PLP and the NZ (2.7 A ? ) atoms of Lys72, Tyr129 and His199*, NE2 (2.6 A respectively. Interestingly, the guanidinium groups of both Arg133 and Arg197* are ?ipped over, resulting in NE and NH1 making two proper hydrogen bonds with the PLP phosphate moiety of ? and 2.8 A ? for Arg133 and 2.9 A ? and 3.3 A ? 2.6 A for Arg197*. In the trigonal crystal structure, only the NH1 atoms from the two residues are in hydrogen bond contact with the phosphate group. The side-chain of Arg197* also moves closer to the phosphate group, probably assisted in part by the hydrogen bonding of its NH2 atom to the backbone oxygen atom of Arg15* (not shown in Figure 4).

X-ray Structure of Pyridoxine-5 0 -phosphate Oxidase

389

Figure 3. Stereo drawing of two neighbouring PNPOx dimers related by C-centering symmetry operation (x ? 1/2, y ? 1/2, z) in the monoclinic crystal. The polypeptide chains are coloured cyan and magenta in one dimer, and green and blue in the other. The sandwiched N-terminal segment between the two dimers is shown in red. The extensive interactions with the neighbouring molecule at the crystal contact interface rendered more stability and less mobility of the N terminus, and thus allows it to be observed in the monoclinic crystal. This Figure was produced using GRASP.33

As described previously, the newly identi?ed Nterminal segment (residues 5-19) in the monoclinic crystal stretches over the active-site cavity entrance of the adjoining monomer to completely sequester the PLP, with the NE and NH2 atoms of Arg14* forming two unique hydrogen bonds with the PLP ? and 3.3 A ? , respectively. phosphate group of 3.0 A In addition, the Tyr17* hydroxyl group also makes ? ) with the aldehyde a short hydrogen bond (2.8 A oxygen atom of PLP. Thus, while the trigonal complex has a partially closed active site, that of the monoclinic complex is completely closed. Kinetic properties of active-site residues As shown in Figure 4, several residues that could be potentially important in binding and catalysis can be identi?ed. These residues were mutated individually to test for their role in the PNPOx mechanism. We have made the following mutant PNPOx forms; H199A, H199N, D49A, Y17F, R14E, R14M, R197M and R197E. The Km and kcat values, for each of these mutants are shown in Table 3. Except for H199N, all mutations have increased Km values, with H199A, R197 M and R197E showing greatly decreased af?nity for PNP.

Only R197E shows a large decrease in catalytic activity. His199 forms a hydrogen bond to O3H of the bound product PLP (Figure 4). This residue is conserved in all 15 currently known PNPOx sequences. Replacing His199 with another hydrogen bond donor (Asn199) results in no change in Km, but about a fourfold decrease in kcat (Table 3). However, removing this hydrogen bond in the H199A mutant results in a 233-fold decrease in af?nity but with no effect on kcat. This suggests that a hydrogen bond is critical for binding, but

Table 3. Kinetic constants for active site mutants of E. coli PNPOx
Enzyme Wild-type H199A H199N D49A Y17F R14E R14M R197M R197E Km (mM) 0.30 70 0.30 1.0 1.0 2.0 2.6 90 2400 kcat (s?1) 0.13 0.14 0.03 0.06 0.60 0.16 0.14 0.03 0.008

390

X-ray Structure of Pyridoxine-5 0 -phosphate Oxidase

Figure 4. Stereoview of the active site of monoclinic PNPOx showing the FMN cofactor (yellow), PLP ligand (green) and associated protein residues. Hydrogen bond and/or salt-bridge interactions are shown between PLP and protein residues from monomer A (cyan) and monomer B (magenta). Atoms are shown in stick representation, with oxygen and nitrogen atoms coloured red and blue, respectively. The Figure was drawn using MOLSCRIPT30 and Raster3D.31

that the basic property of His199 is not important for catalytic activity. The His199 imidazole sidechain is hydrogen bonded to the carboxylate sidechain of Asp49 (not shown in Figure 4). Removing this hydrogen bonding capability in the D49A mutant increase the Km value by a factor of 3 and decreases kcat by a factor of 2, suggesting that Asp49 plays a minor role in binding and catalytic activity. Residue 49 is conserved as either Asp or Glu in 12 of the currently known 15 sequences for PNPOx. Tyr17 and Arg14 are a part of the N-terminal lid that covers the active site. The hydroxyl group of ? hydrogen bond to the C4H oxyTyr17 forms a 2.8 A gen atom of PLP (Figure 4). Removing the hydrogen bonding property of Tyr17 in the Y17F mutant results in a threefold increase in Km, but a 4.6-fold increase in kcat. Since the mechanism of PNPOx is not yet fully understood, we cannot explain the increase in the kcat value, but it might mean that lifting the amino-terminal lid allowing product release may be at least partially rate-determining. Tyr17 is conserved in only six of the 15 structures, but in eight of the remaining nine sequences it is replaced by a group that can form a hydrogen bond with PLP. The guanidinium group of Arg14 ? to two forms two hydrogen bonds of 3.0 and 3.3 A oxygen atoms of the phosphate moiety of PLP. Changing the positive charge to a negative charge

in the R14E mutant increases Km by about sixfold, but has no effect on kcat. Removing the positive charge in the R14 M mutant reduces af?nity by eightfold, but again has no effect on kcat. Arg14 is conserved in only six of the 15 members of this family. Arg197 lies above the plane of the PLP ring and its guanidinium group forms a bifurcated set of hydrogen bonds with two oxygen atoms of the ? and 3.3 A ? phosphate group of PLP of 2.9 A (Figure 4). However, the guanidinium group also ? above the aldehyde moiety of PLP, lies 3.6 A suggesting that this residue may be important for binding of the substrate PNP and involved in chemistry by organizing adjacent residues and bound water molecules. These conclusions are supported both by the effects of mutations on this residue and its conservation in all 15 members of the PNPOx family. Changing the positive charge of Arg197 to a neutral side-chain in R197 M results in a 300-fold increase in Km and a fourfold decrease in kcat. In the R197E mutant, Km is increased 8000fold and kcat is decreased about 16-fold. Stereochemistry of hydrogen removal from C4H of PNP Scheme 1 shows the strategy used to determine the stereochemistry of hydrogen removal from C4H

X-ray Structure of Pyridoxine-5 0 -phosphate Oxidase

391

Scheme 1. Reactions used to determine stereospeci?city of PNPOx. Upper line, reactions used to make (4H R)[3H]PMP. Bottom line, reactions used for determining stereospeci?city of PNPOx.

of PMP. This is based in large part on the same method used to determine the stereochemistry of rabbit liver PNPOx.15 The key to this procedure is the preparation of (4H R)-[3H]PMP using the known stereochemistry of aspartate aminotransferase in converting PLP to PMP.23 The aspartate aminotransferase forms PMP stereospeci?cally and it serves as a convenient puri?cation method allowing the separation of the bound PMP from other reactants by size-exclusion chromatography. The PMP is freed at neutral pH by ethanol-denaturation of aspartate aminotransferase. The oxidation of (4H R)-[3H]PMP transfers one of the C4H hydrogen atoms to H2O and leaves the other hydrogen atom on the product PLP (Scheme 1, bottom line). PLP binds to an excess of apoSHMT, which again provides a convenient puri?cation method because of the large size relative to unreacted (4H R)-[3H]PMP and H2O. We used a ?ltration device to separate the product PLP from the other two potentially radioactive components. A blank sample in which PNPOx was omitted showed that 99 % of the counts passed through the ?lter (Table 4). However, when the

sample was dried, 90 % of these counts remained, showing that the counts were attributable to the unreacted substrate (4H R)-[3H]PMP. The observation that less than 100 % of the counts remained after drying is most likely a result of experimental error due to small volume determinations or minor radioactive contaminants in the (4H R)-[3H]PMP sample. Using wild-type PNPOx and (4H R)-[3H]PMP as substrate, we found 99 % of the radioactivity in the ?ltrate. After drying, only 7 % of the counts remained in the ?ltrate, suggesting that about 93 % of the counts were originally present as 3H2O (Table 4). This means that in the oxidation of (4H R)[3H]PMP the proR hydrogen atom on C4H is removed and the proS hydrogen atom remains with PLP. The non-volatile counts in the Microcon30 ?ow-through fraction most likely are attributable to unreacted (4H R)-[3H]PMP. The stereospeci?city was determined for each of the mutant enzymes. Except for R197 M, all mutant PNPOx reactions result in transfer of the proR hydrogen atom of PMP to H2O. With the R197 M PNPOx, 64 % of the counts were in the ?l-

Table 4. Stereochemistry of PNPOx oxidation of (4R)-[3H]PMP to PLP
Enzyme Blank Wild-type H199N D49A Y17F R14E R14 M R197 M cpm in Micron-30 flow-through 16,600 (99)a 17,000 (99) 15,100 (96) 17,000 (96) 16,700 (99) 17,400 (99) 17,200 (99) 11,700 (64) cpm in the Micron-30 retained fraction 180 160 600 900 130 135 100 6650 (1) (1) (4) (5) (1) (1) (1) (36) cpm in the dried flow-through fraction 15,000 (90) 1400 (7) 650 (41) 600 (4) 1300 (8) 1000 (6) 1200 (7) 10,100 (86)

Each reported value is the average of two separate experiments with the variation in counts never being greater than ?1 %. The values in parentheses are percentage of total counts. a The total counts varied slightly between experiments, but was always in the range of 16,500 to 17,500 cpm.

392 trate and 36 % were located in the holoSHMT fraction containing the product PLP. Of the counts in the ?ow-through fraction, 86 % were not volatile, suggesting that there was less than 50 % conversion of the PMP to PLP in this overnight incubation. However, the 6650 counts in the SHMT fraction shows that as much as 80 % of the PMP that was converted to PLP came by removal of the proS hydrogen atom on (4H R)-[3H]PMP. This suggests that in this mutant there is signi?cant loss of stereospeci?city. R197E and H199A PNPOx have either high Km or low kcat values, resulting in insuf?cient PLP product being formed during the incubation to determine the location of the 3H accurately.

X-ray Structure of Pyridoxine-5 0 -phosphate Oxidase

Discussion
Studies on rabbit liver PNPOx showed that in the oxidation of PMP the enzyme lacks stereospeci?city on removing the hydrogen atom from C4H .15 As noted by the authors of the rabbit liver PNPOx study, there are two possible mechanisms for oxidizing the C4H alcohol or amine group to an aldehyde by a two-electron transfer. (There is no evidence for oxidation by two one-electron transfers to FMN). One mechanism is the transfer of a hydride ion from PNP to FMN, leaving an electron-de?cient C4H . The proposed hydride ion transfer would generate a buildup of positive charge at C4H , which could be resonance-stabilized by the ; electron pairs on O3H as shown in Scheme 2A. The

Scheme 2. Description of two possible mechanisms for the transfer of electrons from the substrate C4H to FMN. The FMN ring is in front of the PMP ring to correspond to the structure observed in the monoclinic crystal. (a) Possible mode of resonance stabilization during a hydride ion transfer of C4H HR of PMP to FMN. (b) Possible mode of resonance stabilization during removal of a proton from C4H HR of PMP to generate a C4H carbanion. The carbanion would then form a covalent adduct with FMN as a part of the electron transfer process.

other proposed mechanism for transferring a pair of electrons from the substrate to FMN is removing a proton from C4H of PNP (PMP), resulting in a C4H carbanion, which then forms a covalent adduct with FMN. A resonance-stabilized structure for the

putative carbanion can also be proposed, as shown in Scheme 2B. The generation of the resonancestabilized carbanion at C4H requires an active-site base to accept the proton from C4H . With either of these two proposed mechanisms it is dif?cult to envision a reaction without stereospeci?city. With the E. coli PNPOx, we now see that the removal of the hydrogen atom from C4H is stereospeci?c. With the monoclinic crystal structure of PNPOx in complex with PLP, as shown in Figure 4, we now have the structure of the closed form of the enzyme with all active-site residues in contact with the product PLP. With the folding of the N-terminal residues over the active site, two additional hydrogen bonds are formed between enzyme and substrate, and solvent is excluded during the conversion of PNP to PLP. Knowing the stereospeci?city of the enzyme should provide a clear picture of the mechanistic pathway for PNPOx. In the substrate complex (PNP or PMP), C4H is tetrahedral and the lowest energy for the hydrogen atom that is removed will be perpendicular to the plain of the pyridine ring. In the hydride ion transfer mechanism, the hydrogen atom will be between the PNP and FMN rings (in Scheme 2A and B, FMN is in front of PMP). In the carbanion mechanism, the proton that is removed from C4H will be on the opposite side of the pyridine ring and away from the FMN ring. Our results show that the labeled hydrogen atom of (4H R)-[3H]PMP is removed during oxidation by PNPOx (Table 4). This ?nding supports the hydride ion transfer mechanism (Scheme 2A). The structure shown in Figure 4 has C4H of PLP located directly above N5 of FMN. The distance of ? between C4H of PLP and N5 of FMN is opti3.3 A mal for hydride transfer. The mechanism shown in Scheme 2B requires a base to be located near HR of C4H , which would be pointing on the opposite side of the PMP ring as discussed above. There are only two residues near C4H of the bound PLP that could be the putative base. These are His199* and Tyr17* (Figure 4). When these two residues are changed to non-basic residues, the enzyme retains considerable catalytic activity, showing that neither can be the proton acceptor required by the mechanism shown in Scheme 2B (Table 3). Also, neither are in the correct position to accept the proR hydrogen atom from C4H of PNP. There is a conserved water molecule located near the C4H hydroxyl group and the guanidinium group of Arg197*. It makes hydrogen bonds with four atoms; Arg197* NH1 ? ), Arg15* N (3.0 A ? ), Tyr17* OH (2.7 A ? ), and (3.2 A ? ). Even though this another water molecule (3.0 A ? 2, water molecule has a very low B-factor (21 A 2 ? compared to 40 A for the average water molecule) ? from HR of PMP when it is and is also 2.9 A oriented as shown in Scheme 2B, it is unlikely to be the putative base. There are several reasons for this conclusion. First, there is no additional valence electron available for bonding to C4H HR. Also, changing Try17* to a non-hydrogen bonding Phe17* increases kcat by 4.6-fold. Removing a

X-ray Structure of Pyridoxine-5 0 -phosphate Oxidase

393 hydrogen bond distance of the OH or NH2 group ? and 2.8 A ?, on C4H to Tyr17* and Arg197* are 2.9 A respectively. The conserved water molecule is ? distant. If you rotate the bond between about 3.6 A C4 and C4H so that 4H HS is now poised for transfer ? and to FMN, the C4H OH or NH2 are now 3.6 A ? ? 2.8 A from Tyr17* and Arg197* and 2.7 A from the water molecule (Figure 5(b)). Since this is a model, the distances of the C4H OH to the water molecule, Tyr17* and Arg197* are not so different that we can eliminate the possibility that the substrates PNP and PMP may be able to bind in either conformation, giving non-speci?c transfer of the pro chiral hydrogen atoms to FMN. In the conformation shown in Figure 5(b), the interaction of the OH or NH2 group attached to C4H with O3H of the substrate ring is lost, which would argue against this conformation. This juxtaposition of the OH group of PNP or NH2 group of PMP and the guanidinium group of Arg197* may determine why in E. coli PNPOx it is the structure corresponding to Figure 5(a) that is bound at the active site. In human and rat PNPOx sequences, Tyr17 is replaced by Asp. This change from a neutral to a negatively charged side-group may signi?cantly alter how this residue interacts with the substrate and the location of Arg197*. The sequence of rabbit PNPOx is not known but may have a substitution for Tyr17* giving a slightly different set of interactions at the active site. This might permit both orientations shown in Figure 5 to be populated, thus resulting in a loss of stereochemistry in the hydrogen atom removed from C4H of PMP.

hydrogen bond from this water molecule should make it a weaker base and result in a decrease in kcat. In the R197M mutant, kcat is reduced only fourfold. A greater effect on removing a hydrogen bond donor and positive charge would be expected if this water molecule was the catalytic base. In none of our structures with PLP bound at the active site do we see any covalent bond between C4H of the PLP and N5 of FMN, which would be an intermediate formed in a carbanion mechanism. But most important of all in eliminating the carbanion mechanism is the location of the aldehyde group of the product PLP. As shown in Scheme 2B, the amino group lies in a position that would put the product aldehyde oxygen atom on the side of the ring toward the phosphate residue and not O3H . Both the monoclinic and trigonal crystal structures show that in the PLP complex the aldehyde oxygen atom lies in the plane on the O3H side of the ring. This leaves unanswered why the rabbit PNPOx was found to lack stereospeci?city in removing the hydrogen atom from C4H of PMP.15 The structure shown in Figure 4 is for the product PLP bound at the active site. C4H is a planar trigonal sp2 hybrid carbon atom in PLP. As noted above for the substrates PNP and PMP, C4H has tetrahedral sp3 hybridization. In Figure 5(a) is shown the activesite structure modeled for PNP or PMP binding at the active site and with 4CH HR both perpendicular and between the pyridine ring of the substrate(s) and the FMN ring system for optimum geometry of hydride ion transfer. In this model, the closest

Figure 5. Model of PNP in the active site of monoclinic PNPOx showing interactions with C4H as a tetrahedral carbon atom with either 4H HR or HS pointed perpendicular and lying between the pyridine ring of PNP and N5 of FMN. (a) Model with C4H HR in position for a hydride transfer to FMN. (b) Model with C4H HS in position for a hydride transfer to FMN. The Figures were drawn using MOLSCRIPT30 and Raster3D.31

394 We mutated Arg197 to Glu or Met to test its role in binding and stereochemistry. As recorded in Table 3, the af?nity of PNP for both mutants has been decreased greatly. We attempted to determine the stereospeci?city of both the R197E and R197M mutant enzymes, but this was not possible for the R197E mutant because of its greatly reduced af?nity and catalytic activity. However, for the R197M mutant PNPOx, about 50 % of the original 17,000 counts have been converted to the products PLP attached to holoSHMT and H2O in the ?ltrate (6650 cpm in the retained fraction ? 1600 cpm lost during drying of the ?ltrate, Table 4). Of these counts in the two products, about 80 % appears in holoSHMT, showing that the R197 M mutant has lost signi?cant stereochemical control and transfers the proS hydrogen atom on C4H about 80 % of the time. This suggests that Arg197* plays a signi?cant role in binding of the substrate for optimum transfer of electrons to FMN.

X-ray Structure of Pyridoxine-5 0 -phosphate Oxidase About 100 mg of each pure mutant protein was obtained per liter of culture. Preparation of apoaspartate aminotransferase Holoaspartate aminotransferase (150 mg; PLP form) was incubated at room temperature for a few minutes in 5 ml of buffer A (100 mM potassium phosphate (pH 6.0), 0.2 mM EDTA, 1 mM dithiothreitol and 50 mM L-glutamate). After addition of solid ammonium sulfate to 25 % saturation, the solution was loaded onto a phenyl-Sepharose column (1 cm ? 10 cm) equilibrated with buffer A containing ammonium sulfate at 25 % saturation. The protein binds to the column as a sharp yellow band. The glutamate converts the enzyme-bound PLP to PMP, which dissociates under the high-salt condition. The column is washed with buffer A until the yellow band disappears and all pyridoxamine 5H -phosphate is eluted, as determined by the lack of an absorbance band at 325 nm in the eluate. The apoaspartate aminotransferase is eluted by ?rst adding 5 ml of buffer A lacking the glutamate. This is followed by washing the column with 20 mM potassium phosphate (pH 6.0), 0.2 mM EDTA, 1 mM dithiothreitol containing 15 % propylene glycol. Fractions containing apoaspartate aminotransferase absorbing at 280 nm are pooled and dialyzed against this buffer (without propylene glycol) after which the protein is concentrated by means of a Centriplus-30 ?lter device (Amicon-Millipore). The apoaspartate aminotransferase is stored at ?20  C at a concentration of about 18 mg/ml. Preparation of (4H R)-[3H]PMP The upper line of Scheme 1 shows the strategy for preparing (4H R)-[3H]PMP. A solution containing 50 nmol of apoaspartate aminotransferase, 1 nmol of PNPOx, 35 nmol of 4H -[3H]PNP in 500 ml of 20 mM potassium phosphate (pH 7.3), 20 mM L-glutamate, 1 mM dithiothreitol is incubated at 37  C for one hour. During this incubation the 325 nm absorbance of PNP shifts to the 355 nm and 428 nm absorbance peaks of holoaspartate aminotransferase. The reaction mixture is concentrated, by centrifugation in a Centricon-30 ?lter, to 250 ml and loaded onto a P6-DG column (1 cm ? 13 cm) previously equilibrated with 20 mM potassium phosphate (pH 7.2), 20 mM L-glutamate, 1 mM dithiothreitol and eluted with the same buffer. The glutamate shifts the equilibrium of holoaspartate aminotransferase from the PLP form to the PMP form with the formation of 2-oxoglutarate. As the enzyme progresses down the sizing column, the 2-oxoglutarate is separated from the enzyme blocking the reverse reaction, resulting in complete conversion from the PLP form to the PMP form of holoaspartate aminotransferase. Fractions of 0.5 ml are collected and those containing aspartate aminotransferase are pooled and concentrated by centrifugation in a Centricon-30 ?lter. The concentrated protein in the Centricon-30 ?lter is diluted with 1 ml of 20 mM potassium phosphate (pH 7.2), 1 mM dithiothreitol buffer and again concentrated. This washing is repeated three times to remove all of the L-glutamate. The holoaspartate aminotransferase solution is then made 30 % (v/v) in ethanol and incubated at 60  C for 30 minutes to denature the protein and to release (4H R)-[3H]PMP. After cooling to 4  C, the precipitated protein is removed by centrifugation at 14,000 rpm for ?ve minutes in a microcentrifuge. A spectrum of the supernatant (total volume 400 ml) is taken

Materials and Methods
Materials PMP, PLP, PN, and PM were obtained from Sigma Chemical Co., St. Louis, MO. All buffer solutions were made of the purest components available. E. coli aspartate aminotransferase was puri?ed from cell extracts of an over-expressing clone, kindly provided by Dr Roberto Contestabile, University of Rome.24 E. coli PNPOx and rabbit liver cytosolic SHMT were puri?ed from E. coli clones as described.7,25 4H -[3H]PNP and 4H -[3H]PLP were synthesized from unlabeled PLP and NaB[3H]4 (Amersham Life Science Inc., Arlington Heights, IL) as described.26 The apoSHMT was prepared by using L-cysteine to trap the PLP as a thiazolidine complex. L-Cysteine (10 mM) was added to 20 mg of puri?ed SHMT in 100 mM potassium phosphate (pH 6.8), 25 % saturated with ammonium sulfate. This solution was added to a 1 cm ? 5 cm phenyl-Sepharose column equilibrated with the same L-cysteine/ammonium sulfate-containing buffer. The enzyme binds tightly and continued washing with this buffer results in removal of the bound PLP, leaving apoSHMT bound to the column. After all PLP has been removed, as evidenced by the lack of 330 nm absorbing material in the eluate, the apoSHMT is eluted with 10 mM potassium phosphate (pH 7.3), 1 mM dithiothreitol. The apoSHMT is concentrated and dialyzed against the eluting buffer. Preparation of PNPOx site mutants All mutant forms of PNPOx were made on the wildtype construct pET22::PNPOx7 using the QuickChangeTM Site-Directed Mutagenesis Kit from Stratagene (La Jolla, CA). Oligonucleotides were synthesized by Integrated DNA Technologies, Inc., (Coralville, IA). E. coli MDS00 (W3110 lacI169 tna2 sup0 ?pdxH::(Cmr) (DE3)), used for expression of the mutant forms of PNPOx, were generated from E. coli strain TX2768 through a lDE3 Lysogenization Kit (Novagen, Madison, WI). For protein expression, bacterial cells were grown for 15 hours at 37  C in 1.5 ? LB medium without isopropyl-b-thiogalactoside induction. Protein puri?cation was as described.7

X-ray Structure of Pyridoxine-5 0 -phosphate Oxidase and 10 ml counted for tritium content. With this procedure, an average speci?c activity of about 14,000 cpm per nmol for (4H R)-[3H]PMP is obtained, as determined by using a molar absorbtivity coef?cient of 8300 cm?1 M?1 for PMP at 324 nm. Crystallization and data collection

395

Determination of stereospecificity of PNPOx The strategy for determining the stereospeci?city of PNPOx is given in the second line of Scheme 1. The substrate (4H R)-[3H]PMP is oxidized by PNPOx to PLP and the 3H will be released either as H2O or retained in the product PLP depending on the hydrogen removed from C4H . PLP that is generated during the PNPOx reaction is trapped by rabbit cytosolic apoSHMT forming the holo enzyme. The H2O and holoSHMT fractions are then separated by ?ltration. A solution containing 0.6 nmol of (4H R)-[3H]PMP, 6 nmol of apoSHMT, 0.9 nmol of PNPOx in 600 ml of 20 mM potassium phosphate (pH 7.3), 1 mM dithiothreitol is incubated at 37  C for two hours. Then another 6 nmol of apoSHMT is added and the reaction incubated for an additional 30 minutes, or until the 325 nm absorbance peak of PMP has shifted completely to the 428 nm absorbance peak of holoSHMT (an overnight incubation was performed for some mutant forms of PNPOx). The reaction mixture is concentrated by centrifugation in a Microcon-30 ?lter at 14,000 rpm for 15 minutes. After two 150 ml water rinses of the retained fraction, the ?ltrate solutions are combined and 200 ml counted for tritium content. The ?ltrate will contain the water, buffer, and any unreacted (4H R)-[3H]PMP, while the fraction retained by the Microcon-30 membrane will contain the holoSHMT with the bound PLP. To determine if the counts in the ?ltrate are H2O or residual (4H R)-[3H]PMP, a 200 ml aliquot is dried by vacuum centrifugation and dissolved in 500 ml. This process is repeated three times. After the ?nal drying, the remaining residue is dissolved in 200 ml of H2O and counted to determine how much of the original radioactivity in the ?ltrate is not volatile, which is a measure of the unreacted (4H R)-[3H]PMP. The retentate of the Microcon-30 ?lter containing the PLP bound to SHMT is dissolved in 300 ml of 20 mM potassium phosphate (pH 7.3), 1 mM dithiothreitol and 70 ml counted for tritium content.

Crystallization and structure determination of PNPOx in a trigonal unit cell has been described.16 ± 18 A monoclinic crystal form, with a needle morphology, was grown with a 9.3 mg/ml SeMet derivative of the native protein solution in 100 mM potassium phosphate (pH 7.5), 5 mM 2-mercaptoethanol by the hanging-drop, vapordiffusion method. The drop contained 3.0 ml of protein, 0.2 ml of 13 mM PLP and 3 ml of reservoir solution (1 M (NH4)2SO4, 1.5 % (v/v) dioxane, 0.1 M 2-[N-morpholino]ethanesulfonate/NaOH buffer, pH 7.0). The ?nal molar ratio of protein to PLP is approximately 1:5. Crystals appear after three months and grow slowly to sizes suitable for X-ray data collection after six months. Diffraction data sets were collected at 100 K using a Molecular Structure Corporation (MSC) X-Stream Cryogenic Crystal Cooler System (Molecular Structure Corporation, The Woodlands, TX, USA), an R-Axis II image plate detector equipped with OSMIC confocal mirrors and a Rigaku RU-200 X-ray generator operating at 50 kV and 100 mA. Prior to use in X-ray diffraction, the monoclinic crystals are ?rst washed in a cryoprotectant solution containing 2.1 M (NH4)2SO4, 4.2 mM PLP, 7.7 % (v/v) glycerol and then transferred to a similar solution containing 14.2 % glycerol. All data are processed using the BIOTEX software of MSC and the CCP4 suite.27 Statistics for the data set used in the re?nements are listed in Table 1. Structure determination and refinement The monoclinic crystal form of E. coli PNPOx was solved by molecular replacement methods using the program AMoRe of the CCP4 suite.27 The monomeric trigonal crystal structure,17,18 omitting all FMN, PLP, phosphate and water molecules was used to determine the structure of the monoclinic crystal form. Based on the solvent content of the unit cell (43 %, v/v), we expected one monomer in the asymmetric unit. The cross-rotation function was calculated using normalized structure factors with data in the resolution range of 8.0 ? . One unique solution was observed with a correto 4.0 A lation coef?cient of 26.1. The translation function, using the space group C2 resulted in a correlation coef?cient and R-factor of 54.5 and 39.4 %, respectively. Structure re?nements were performed with the CNS program,28 with overall anisotropic B-factor correction and a bulk solvent correction. A statistically random selection of 5 % of the total re?ection data was excluded from the re?nement and used to calculate the free R-factor (Rfree) as a monitor of model bias.29 Rigid-body re?nement of the molecular replacement model gave an ? . In the R-value of 0.364 for the entire data set at 2.07 A initial 2Fo ? Fc map, shown in Figure 2(a), there are densities for both FMN and PLP molecules at the active site. The N-terminal Gly20 did not have corresponding density and was deleted. However, densities in the nearby regions indicated a different N-terminal extension. All of these were ?rst modelled as 78 water molecules. After one round of positional and simulated annealing re?nement, a second 2Fo ? Fc map was calculated. It clearly showed densities for amino acid residues 8-20, which were missing from our previous structures of PNPOx.17,18 With all water molecules replaced by the N terminus residues 8-20, and addition of FMN and PLP, the model yielded R and Rfree values of 0.258 and 0.295, respectively. Further re?nement cycles included addition

Assays and determination of kinetic constants Determination of catalytic activity during puri?cation is performed in a 1 cm cuvette with 50 mM Tris ? HCl (pH 8.0), 1 mM dithiothreitol as the buffer at 37  C.26 (The kinetic constants do not change between pH 7.0, where the crystal structure was determined, and pH 8.0 for PNPOx.) The product PLP forms an aldimine with the Tris buffer that absorbs at 414 nm with a molar absorbtivity coef?cient of 5900 cm?1 M?1. When determining Km and kcat values, a 10 cm pathlength cell is used with 20 mM Tris (pH 8.5), 1 mM dithiothreitol. The PNP concentration is varied between 0.3 and two times the Km value. Enough PNPOx is added to give a maximum rate of about ?A414 nm of 0.1 per minute. Km and kcat values are determined from double reciprocal plots constructed with Sigma Plot. The concentration of PNPOx is determined from its absorbance at 278 nm as described.7

396
of 267 water molecules and a phosphate ion, as well as extension of the N terminus to Asp5. A portion of the ?nal 2Fo ? Fc map is shown in Figure 2(b). Omit 2Fo ? Fc and Fo ? Fc electron density maps were repeatedly used to inspect the structure for positional corrections. All model building and model corrections were carried out using the program O.19 Re?nement statistics are summarized in Table 1. Protein Data Bank accession numbers The atomic coordinate and structure factor sets for the monoclinic structure have been deposited in the RCSB Protein Data Bank with accession code 1jnw. 13.

X-ray Structure of Pyridoxine-5 0 -phosphate Oxidase oxidase for ?avin-phosphates. Biochim. Biophys. Acta, 359, 282-287. Horiike, K., Merrill, A. H., Jr & McCormick, D. B. (1979). Activation and inactivation of rabbit liver pyridoxamine (pyridoxine) 5H -phosphate oxidase activity by urea and other solutes. Arch. Biochem. Biophys. 195, 325-335. Horiike, K., Tsuge, H. & McCormick, D. B. (1979). Evidence for an essential hisdtidyl residue at the active site of pyridoxamine (pyridoxine)-5-phosphate oxidase from rabbit liver. J. Biol. Chem. 254, 66386643. Merrill, A. H., Jr, Korytnyk, W., Horiike, K. & McCormick, D. B. (1980). Spectroscopic studies of complexes between pyridoxamine (pyridoxine)-5phosphate oxidase and pyridoxyl 5H -phosphate compounds differing at position 4H . Biochim. Biophys. Acta, 626, 57-63. Choi, J.-D. & McCormick, D. B. (1981). Roles of arginyl residues in pyridoxamine-5H -phosphate oxidase from rabbit liver. Biochemistry, 20, 5722-5728. Bowers-Komro, D. & McCormick, D. B. (1985). Pyridoxamine-5H -phosphate oxidase exhibits no speci?city in prochiral hydrogen abstraction from substrate. J. Biol. Chem. 260, 9580-9582. Musayev, F. N., Safo, M. K., di Salvo, M. L., Schirch, V. & Abraham, D. J. (1999). Crystallization and preliminary X-ray crystallographic analysis of pyridoxine 5H -phosphate oxidase complexed with ?avin mononucleotide. J. Struct. Biol. 137, 88-91. Safo, M. K., Mathews, I., Musayev, F. N., di Salvo, M. L., Thiel, M. L., Abraham, D. J. & Schirch, V. (2000). X-ray structure of Escherichia coli pyridoxine ? 5H -phosphate oxidase complexed with FMN at 1.8 A resolution. Structure, 8, 751-762. Safo, M. K., Musayev, F. N., di Salvo, M. L. & Schirch, V. (2001). X-ray structure of Escherichia coli pyridoxine 5H -phosphate oxidase complexed with pyridoxal ? resolution. J. Mol. Biol. 310, 5H -phosphate at 2.0 A 817-826. Jones, T. A., Zou, J.-Y., Cowan, S. W. & Kjeldgaard, M. (1991). Improved methods for building protein models in electron density maps and the location of errors in models. Acta Crystallog. sect. A, 47, 392-400. Richardson, J. S. (1981). The anatomy and taxonomy of protein structure. Advan. Protein Chem. 34, 167339. Matthews, B. W. (1968). Solvent content of protein crystals. J. Mol. Biol. 33, 491-497. Janin, J. & Rodier, F. (1995). Protein-protein interaction at crystal contacts. Proteins: Struct. Funct. Genet. 23, 580-587. ? ger, J., Moser, M., Sauder, U. & Jansonius, J. N. Ja (1994). Crystal structures of Escherichia coli aspartate aminotransferase in two conformations. J. Mol. Biol. 239, 285-305. Goldberg, J. M., Swanson, R. V., Goodman, H. S. & Kirsch, J. F. (1991). The tyrosine-225 to phenylalanine mutation of Escherichia coli aspartate aminotransferase results in an alkaline transition in the spectrophotometric and kinetic pKa values and reduced values of both kcat and Km. Biochemistry, 30, 305-312. di Salvo, M. L., Delle Fratte, S., De Biase, D., Bossa, F. & Schirch, V. (1998). Puri?cation and characterization of recombinant rabbit cytosolic serine hydroxymethyltransferase. Protein Express. Purif. 13, 177-183.

11.

12.

Acknowledgments
This work was supported by grants DK 55648 (to V.S.) and HL04367 (to M.K.S.) from the National Institutes of Health. M.L.D.S. was partly supported by a ? e Ricerca grant from Ministero Istruzione Universita (MIUR).

14. 15.

References
1. Zhao, G. & Winkler, M. (2000). Kinetic limitation and cellular amount of pyridoxine (pyridoxamine) 5H -phosphate oxidase of Escherichia coli K-12. J. Bacteriol. 177, 883-891. 2. Hill, R. E., Himmeldirk, K., Kennedy, I. A., Pauloski, R. M., Sayer, B. G., Wolf, E. & Spenser, I. D. (1996). The biogenetic anatomy of vitamin B6. J. Biol. Chem. 271, 30426-30435. 3. Choi, J-D., Bowers-Komro, D. M., Davis, M. D., Edmondson, D. E. & McCormick, D. B. (1983). Kinetic properties of pyridoxamine (pyridoxine)-5H phosphate oxidase from rabbit liver. J. Biol. Chem. 258, 840-845. 4. Kazarinoff, M. N. & McCormick, D. D. (1975). Rabbit liver pyridoxamine (pyridoxine) 5H -phosphate oxidase: puri?cation and properties. J. Biol. Chem. 250, 3436-3443. 5. Choi, S. Y., Churchich, J. E., Zaiden, E. & Kwok, F. (1987). Brain pyridoxine-5H -phosphate oxidase: modulation of its catalytic activity by reaction with pyridoxal 5H -phosphate and analogs. J. Biol. Chem. 262, 12013-12017. 6. Churchich, J. E. (1984). Brain pyridoxine-5-phosphate oxidase: a dimeric enzyme containing one FMN site. Eur. J. Biochem. 138, 327-332. 7. Di Salvo, M., Yang, E., Zhao, G., Winkler, M. E. & Schirch, V. (1998). Expression, puri?cation, and characterization of recombinant Escherichia coli pyridoxine 5H -phosphate oxidase. Protein Express. Purif. 13, 349-356. 8. Kazarinoff, M. N. & McCormick, D. B. (1973). N-(5H phospho-4H -pyridoxyl) amines as substrates for pyridoxine (pyridoxamine) 5H -phophate oxidase. Biochem. Biophys. Res. Commun. 52, 440-446. 9. Merrill, A. H., Horiike, K. & McCormick, D. G. (1978). Evidence for the regulation of pyridoxal 5H -phosphate formation in liver by pyridoxamine (pyridoxine) 5H -phosphate oxidase. Biochem. Biophys. Res. Commun. 83, 984-990. 10. Kazarinoff, M. N. & McCormick, D. B. (1974). Speci?city of pyridoxine (pyridoxamine) 5H -phosphate

16.

17.

18.

19.

20. 21. 22. 23.

24.

25.

X-ray Structure of Pyridoxine-5 0 -phosphate Oxidase 26. Yang, E. S. & Schirch, V. (2000). Tight binding of pyridoxal 5H -phosphate to recombinant Escherichia coli pyridoxine 5H -phosphate oxidase. Arch. Biochem. Biophys. 377, 109-114. 27. Collaborative Computing Project No 4. (1994). The CCP4 suite: programs for protein crystallography. Acta Crystallog. sect. D, 50, 760-763. 28. Brunger, A. T., Adams, P. D., Clore, G. M., DeLano, W. L., Gros, P., Grosse-Kunstleve, R. W. et al. (1998). Crystallography & NMR System: a new software suite for macromolecular structure determination. Acta Crystallog. sect. D, 54, 905-921. 29. Brunger, A. T. (1992). Free R-value: a novel statistical quantity for assessing the accuracy of crystal structures. Nature, 35, 42-475.

397
30. Kraulis, P. J. (1991). MOLSCRIPT: a program to produce both detailed and schematic plots of protein structures. J. Appl. Crystallog. 24, 946-950. 31. Merrit, E. A. & Murphy, M. E. P. (1994). Raster3D version 2.0: a program for photorealistic molecular graphics. Acta Crystallog. sect. D, 50, 869-873. 32. Esnouf, R. M. (1997). An extensively modi?ed version of MolScript that includes greatly enhanced coloring capabilities. J. Mol. Graph. 15, 132-134. 33. Nicholls, A., Sharp, K. A. & Honig, B. (1991). Protein folding and association: insights from the interfacial and thermodynamic properties of hydrocarbons. Proteins: Struct. Funct. Genet. 11, 281-296.

Edited by R. Huben (Received 2 August 2001; received in revised form 2 November 2001; accepted 6 November 2001)


相关文章:
5年制第三章酶_图文.ppt
维生素B6(Vit B6) 包括吡哆醇(pyridoxine), 吡哆醛(...(active center) 定义:又称为活性部位(active site...立体异构专一性(stereospecificity) 3. 可调节性 酶...
四川大学生物化学课件 第3章酶_图文.ppt
维生素B6(Vit B6) 维生素B 包括吡哆醇(pyridoxine),...(active center) 定义:又称为活性部位(active site...specificity) 立体异构专一性(stereospecificity) 3. ...
...mechanism and stereospecificity of Escherichia coli L-....unkown
On the catalytic mechanism and stereospecificity of Escherichia coli L-...(L-TAs) represent a family of homologous pyridoxal 5′-phosphate-dependent ...
Stereospecificity of Tripeptide Utilization in a Methionine ....unkown
stereospecificity of peptide utilization in Escherichia coli K-12 4212,...are not recognized by the oligopeptide transport system (3, 5, 9)....
Stereospecificity of the Ferric Enterobactin Receptoorf ....unkown
( ' . S A Stereospecificity of the Ferric ...Escherichia coli K-12* (Received for publication,...indicates outer membrane receptor ture 1) (5). ...
...in the stereospecificity pocket and at the entrance of the....unkown
Mutations in the stereospecificity pocket and at the entrance of the active site of Candida antarctica lipase B enhancing enzyme enantioselectivity Zsuzsanna ...
STRUCTURE AND STEREOSPECIFICITY OF METAL COMPLEXES.unkown
庐Dn0\芦%. H (A vs 5F'j\ STRUCTURE AND STEREOSPECIFICITY OF METAL COMPLEXES A thesis submitted for the degree of Doctor of Philosophy by TREVOR ...
An Apparent Lack of Stereospecificity in the Reaction ....unkown
(1976) 157, 573-576 Printed in Great Britain 573 An Apparent Lack of Stereospecificity in the Reaction Catalysed by Deoxyribose 5-Phosphate Aldolase due ...
...active site residue directs oxygenation stereospecificity ....unkown
A single active site residue directs oxygenation stereospecificity in lipoxygenases: Stereocontrol is linked to the position of oxygenation Gianguido Coffa and...
Cancer.unkown
5.10 http://dx.doi.org/10.4172/1948-5956....Vitamin B6 (pyridoxine) has also been reported ...Both linoleic acid and its stereoisomer, ...
STEREOSPECIFICITY OF 2,4-DIAMINOBUTYRIC ACID WITH RESPECT TO ....unkown
(1977), 59, 218-219 SHORT COMMUNICATION STEREOSPECIFICITY OF 2,4-DI...stereo- specific, since R(-)-nipecotic acid is approximately 5 times more...
THE STEREOSPECIFICITY OF a-CHYMOTRYPSIN-CATALYSED HYDROLYSIS ....unkown
THE STEREOSPECIFICITY OF a-CHYMOTRYPSIN-CATALYSED HYDROLYSIS AND ALCOHOLYSIS OF SPECIFIC ESTER SUBSTRATES* By B. HALFERN,J?. RICKS,?and J. W. WESTLEY~ ...
...Editing Domain Variants Switch the Stereospecificity of ....unkown
Article pubs.acs.org/biochemistry Hyperactive Editing Domain Variants Switch the Stereospecificity of Tyrosyl-tRNA Synthetase Charles J. Richardson∥ and Eric ...
...Stereospecificity of Adenylate Kinase by Site-Directed ....unkown
stereospecificity of enzymes toward specific isomer(s) of nucleoside phosphorothioates, a property uncovered by mechanistic enzy- ' ADP, Adenosine 5'-...
STEREOSPECIFICITY OF HYDROGEN TRANSFER BY THE HUMAN PLACENTAL....unkown
Volume 72, number 2 FEBS LETTERS December 1976 STEREOSPECIFICITY OF HYDROGEN...[5,6-3H]PGEI (2.4 X 10' dpm) (the Radiochemical Centre Ltd, ...
SITE SELECTIVITY AND STEREOSPECIFICITY IN MASS SPECTRAL ....unkown
SITE SELECTIVITY AND STEREOSPECIFICITY IN MASS SPECTRAL ELIMINATION REACTIONS Mark M. Green and John Schwab Department of Chemistry, University of Michigan Ann...
...A Hydratase Activity and Its Stereospecificity Using High-....unkown
Oleo Sci. 60, (5) 221-228 (2011) Analysis of Enoyl-Coenzyme A Hydratase Activity and Its Stereospecificity Using High-Performance Liquid Chromatography ...
Stereospecificity of Fatty Acid 2-Hydroxylase and ....unkown
Stereospecificity of Fatty Acid 2-Hydroxylase and Differential Functions of 2-Hydroxy Fatty Acid Enantiomers Lin Guo1, Xu Zhang1, Dequan Zhou1, Adewole ...
Stereospecificityof the Microsomal Oxidation of Ethanol.unkown
Stereospecificity in the formation of acetaldehyde by the microsomal ethanol- oxidizing system The retention of 3Hby ethanol on oxidation to acetaldehyde is...
Distribution in Organisms and Stereospecificity of f3- ....unkown
//www.tandfonline.com/loi/tbbb19 Distribution in Organisms and Stereospecificity of β- Hydroxyisobutyrate Dehydrogenase Junzo Hasegawaa a Biochemical Research...
更多相关标签: