Probing the Roles of Active Site Residues in D -Xylose Isomerase*

The roles of active site residues His 54 , Phe 94 , Lys 183 , and His 220 in the Streptomyces rubiginosus D -xylose isomerase were probed by site-directed mutagenesis. The kinetic properties and crystal structures of the mutant enzymes were characterized. The pH dependence of diethylpyrocarbonate modification of His 54 suggests that His 54 does not catalyze ring-opening as a general acid. His 54 appears to be involved in anomeric selection and stabilization of the acyclic transition state by hy- drogen bonding. Phe 94 stabilizes the acyclic-extended transition state directly by hydrophobic interactions and/or indirectly by interactions with Trp 137 and Phe 26 . Lys 183 and His 220 mutants have little or no activity and the structures of these mutants with D -xylose reveal cy- clic (cid:97) - D -xylopyranose. Lys 183 functions structurally by maintaining the position of Pro 187 and Glu 186 and catalytically by interacting with acyclic-extended sugars. His 220 provides structure for the M2-metal binding site with properties which are necessary for extension and isomerization of the substrate. A second M2 metal binding site (M2 (cid:42) ) is observed at a relatively lower occupancy when substrate is added consistent with the hypothesis that the metal moves as the hydride is shifted on the extended substrate DNA M dGTP S. m M 1.5 m M MgCl 2 0.05% Tween 20 (w/v), 0.1 mg/ml bovine albumin m M Tris-HCl 8.8). PCR a Hybaid thermocycler, following parameters: cycle 1, 90 °C for 10 min, 94 °C for 0.5 min, 65 °C for 2 min, 72 °C for 2 min; cycles 2–34, °C for 0.5 min, 65 °C for 2 min, 72 °C for 2 min; cycle 35, 94 °C for 0.5 min, 65 °C for 2 min, 72 °C for 5 min. After the first PCR, 1 (cid:109) used in a second PCR. The , -Ser/xylitol. alternative conformations Asp Glu 186 the -Ser/xylitol structure be equivalent (not and each of the conformations were at 0.5 occupancy. The average B factors for ALT1 and ALT2 of the Asp 255 side chain are and 8 Å 2 , and for the two positions of the Glu 186 side chain are 13 and 8 Å 2 . The new positions for the Glu 186 side chain the phenyl group of Phe 26* rotate 9° about (cid:120) 1 from its position in the xylitol-free His 220 -Ser structure. Significant movement of M1 (0.5 Asp 287 OD2 Å), and Asp 245 OD1 Å, not shown) is also observed, but M1 ligand distances (2.1 to 2.3 Å) do not differ significantly from the previously described structures in this study or others (15, 36).

Initially, a cis-enediol mechanism was proposed for D-xylose isomerase (14,22), similar to the mechanism of triose-phosphate isomerase. However, isotope exchange experiments (23) and crystallographic analyses (15,16) with various substrates and inhibitors suggests that the reaction proceeds via a metalmediated hydride shift (Fig. 1a). The currently accepted pathway for the reaction involves the preferential binding of ␣-Dxylopyranose (24,25) followed by ring opening (25), extension of the substrate, and then the hydride shift (15,16,26). Recently, Meng et al. (27) have proposed that the hydride shift occurs on the cyclic form of sugar (Fig. 1b).
Site-directed mutagenesis has been used to probe the functions of specific active site residues in D-xylose isomerase (27)(28)(29)(30)(31)(32), however, only a few structures of mutant enzymes have been reported (19,(33)(34)(35). Kinetic data can be misleading if the substitutions affect the properties of catalytically important residues other than those changed by mutagenesis. For this reason, we have shifted our mutagenic studies from the Escherichia coli D-xylose isomerase (28,29), which has not been successfully crystallized, to the S. rubiginosus enzyme which readily forms crystals diffracting x-rays beyond 2.0 Å (15,34,36). Here, we report the cloning and expression of S. rubiginosus xylA in E. coli, and investigate the roles of His 54 and Phe 94 in anomeric recognition, the putative role of His 54 in ring opening, and the functions of His 54 , Phe 94 , Lys 183 , and His 220 in isomerization.
Biochemical Reagents-All compounds were reagent grade and purchased from Sigma, except acetaldehyde and 5-thio-␣-D-glucopyranose (THG), 1 which were from Aldrich. The sugar 5-deoxy-D-xylulose was synthesized by aldol condensation of dihydroxyacetone and acetaldehyde, using rabbit muscle aldolase with the cofactor sodium arsenate as a catalyst, and was purified by ion-exchange chromatography, 2 using a scheme similar to that described by Durrwachter et al. (37) for the synthesis of 5-deoxy-D-fructose.
DNA Isolation, Transformation, and Manipulations-S. rubiginosus was grown in yeast and maltose extract medium supplemented with 34% (w/v) sucrose and chromosomal DNA was isolated as Hopwood et al. (38). Both plasmid and bacteriophage DNA were isolated from cul-tures of E. coli TG1 grown in LB medium as recommended by Sambrook et al. (39). Site-directed mutagenesis was conducted as described previously (34,40). His 54 was changed to Ser, Asn, and Asp; Phe 94 to Ser; Lys 183 to Met; and His 220 to Ser and Glu. DNA sequencing was carried out with the 7-deaza-dGTP sequencing kit from U. S. Biochemical Corp.
Cloning and Expression of S. rubiginosus xylA in E. coli-The reported S. rubiginosus xylA sequence (12) was used to construct two oligonucleotide primers (primer 1: ATGAACTACCAGCCCACCCCC-GAGGACAGG) (primer 2: TCAGCCCCGGGCGCCAGCAGGTGGTC-CAT) which were used in a PCR to amplify xylA from purified S. rubiginosus chromosomal DNA. The PCR reaction mixture was in 100 l, containing 2 units of Taq DNA polymerase, 250 M each of dATP, dTTP, and dCTP, 125 M each of dGTP and 7-deaza-dGTP, 0.5 M each of primers 1 and 2, 500 ng of S. rubiginosus chromosomal DNA, 50 mM KCl, 1.5 mM MgCl 2 , 0.05% Tween 20 (w/v), and 0.1 mg/ml bovine serum albumin in 10 mM Tris-HCl (pH 8.8). The PCR was conducted in a Hybaid thermocycler, using the following parameters: cycle 1, 90°C for 10 min, 94°C for 0.5 min, 65°C for 2 min, 72°C for 2 min; cycles 2-34, 94°C for 0.5 min, 65°C for 2 min, 72°C for 2 min; cycle 35, 94°C for 0.5 min, 65°C for 2 min, 72°C for 5 min. After the first PCR, 1 l was removed and used in a second PCR. The reaction mixture was as above except primers 1 and 2 were replaced by primers containing restriction sites for EcoRI and BspHI (bold and underlined, respectively; GGAAT-TCATGAACTACCAGCCCACCCCCGAG) and HindIII (italic; CCCA-AGCTTAGCCCCGGGCGCCCAGCAG). The PCR reaction was then as follows: cycle 1, 94°C for 1 min, 50°C for 1 min, 74°C for 2 min; cycles 2-34, 94°C for 0.5 min, 60°C for 0.5 min, 74°C for 2 min; cycle 35, 94°C for 0.5 min, 60°C for 0.5 min, 74°C for 10 min. The PCR product was purified with a Sephacryl G-200 spin column (Pharmacia) and concentrated by isopropyl alcohol/sodium acetate precipitation. The PCR product was restricted with EcoRI and HindIII, ligated into EcoRI and HindIII digested M13 mp19 and transformed into E. coli TG1. Only one transformant was selected that had a 1.2-kilobase insert. The insert DNA was sequenced and confirmed to be xylA. The plasmid pRDW100 was constructed by removing the xylA insert from M13 mp19 with BspHI and HindIII and subcloning it into NcoI-HindIII digested pET11d. The production and purification of D-xylose isomerase was as described previously (34).
Enzyme Assays-D-Xylose and D-xylulose solutions were prepared at least 12 h prior to use in kinetic assays in order to allow them to reach anomeric equilibrium. All mutants and wild-type D-xylose isomerase were preincubated with metal 12 h prior to measuring activity. The activity of His 220 and Lys 183 substitutions was assayed with D-xylose, as described previously (34), using the method of Dische and Borenfreund (41). The kinetics of His 54 and Phe 94 substitutions and of the wild-type D-xylose isomerase with D-xylose as a substrate were determined by the coupled sorbitol dehydrogenase assay (42). The reaction was carried out at 30°C in 100 mM KCl, 10 mM HEPES (pH 7.7), 10 mM MgCl 2 , 0.55 mM NADH, 1.5 units of sorbitol dehydrogenase, 1-100 mM D-xylose and D-xylose isomerase (0.05-1 M) in a total volume of 1 ml. The rate of NADH oxidation was followed on a Pharmacia UltraspecII spectrophotometer at 340 nm for 1.5 min. The kinetics of His 54 mutants and wild-type D-xylose isomerase with D-xylulose as a substrate was performed at 30°C in 100 mM KCl, 10 mM HEPES (pH 7.7), 10 mM MgCl 2 , 0.5-40 mM D-xylulose, and 0.1-0.7 M D-xylose isomerase in a total volume of 100 l. When 5-10% of the D-xylulose was converted to D-xylose, the reaction was terminated by the addition of 100 l of 0.1 M trichloroacetic acid (pH 2). D-Xylulose was separated from D-xylose by high performance liquid chromatography using a Bio-Rad Carbo-C column and the concentration of D-xylulose was determined by integration of the peak corresponding to its fraction. The kinetics of His 54 mutants and wild-type on 5-deoxy-D-xylulose was determined at 30°C in 100 mM KCl, 50 mM HEPES (pH 7.7), 10 mM MgCl 2 , 10 -200 mM 5-deoxy-Dxylulose, and 2-4 M D-xylose isomerase in 200 l. The reaction was stopped after 100 min with 25 l of 2 M trichloroacetic acid (pH 2), 275 l of D 2 O was added, and the amount of 5-deoxy-D-xylulose was determined by NMR (Brü ker AM 300). DEPC Inactivation of D-Xylose Isomerase-D-Xylose isomerase (18.6 M) was inactivated with 0.157 mM DEPC in 10 mM MgCl 2 and 50 mM potassium phosphate buffer (pH 5.5-7.5) at 25°C. Aliquots (10 l) were removed from 0 to 10 min, and activity with D-xylose as a substrate was measured using the sorbitol dehydrogenase assay described above. The rate of DEPC inactivation at each pH was determined by plotting the log D-xylose isomerase activity against time of incubation with DEPC. The inflection point (pK a ) was determined by taking the second derivative of the third order polynomial calculated by Cricket Graph version 1.3.2 (Malvern, PA).
Anomeric Specificity Assay-␣-D-Xylose (96%) was dissolved in water  (27). Residues relevant to this study (His 54 , Phe 94 , Lys 183 , and His 220 ) are shown, as well as Asp 287 , which is important in the cyclic-hydride transfer mechanism (b). Metal ions are labeled M1 and M2. In both mechanisms the ␣-pyranose (D-xylose is shown here) binds to the active site and after isomerization, the ␣-furanose form (D-xylulose is shown here) is released. In the metal-mediated hydride shift, ring opening is required prior to isomerization, while in the cyclic-hydride transfer ring opening and isomerization occur at the same step.

FIG. 2. Electron densities for D-xylose isomerase complexes with ligands.
For all of the panels, residues are labeled at CA, metal ions (asterisks) are labeled M1 and M2, and water molecules (asterisks) are labeled Wat. The dashed lines depict metal ligation and hydrogen bonds which have distances Յ3.2 Å and reasonable geometry. Cross-hatched lines represent electron density of the F o Ϫ F c "omit" map. The contour level is 3 in panels a and b, and 4 in panels c-f. a, wild-type/THG. THG was modeled in two conformations (ALT1 and ALT2) and each was refined at 0.5 occupancy. The main difference in ALT1 and ALT2 of THG is the position of O6. In ALT1, O6 is 2.8 Å from Thr 90 OG, and in ALT2, O6 is 2.5 Å from Wat 408 . The B factors for both models of THG, M1, and M2 are 17, 10, and 7 Å 2 , respectively. Although not indicated in this figure, the positions of both metals shift 0.6 Å compared to the wild-type structure without THG. The protein side chains liganded to M1 shift 0.3 to 0.8 Å as a result of the new position of the metal and also because of the proximity of the sugar hydroxyls O3 and O4. Shifts in the positions of side chains liganded to M2 are also observed, most likely because the ligands to M1 are repositioned. The indole group of Trp 137 rotated 9°about 2 and the phenyl ring of Phe 26* (*, from a neighboring subunit) rotated 5°about 1 when THG was added. The amino side chain of Lys 183 moves 0.5 Å in order to accommodate the new position of Wat 565 which is hydrogen bonded to O2 of THG. His 54 NE2 is 2.9 Å from the ring sulfur and 3.5 Å from the anomeric hydroxyl, and both O3 and O4 are liganded to M1 at 2.3 Å. b, His 54 -Ser/xylose. D-Xylose was placed in the electron density, but not added during refinement because the electron density of the sugar is weak and in the absence of sugar, water molecules are located close to the positions of the sugar hydroxyls and would complicate an accurate determination of sugar occupancy (not shown). Sugar binding would be comparable to a wild type-xylulose complex (36), except there is no hydrogen bond between Ser 54 and O5. Instead, O5 of sugar could hydrogen bond to Wat 585 and Wat 708 , a water not observed in a wild-type/xylulose structure (36). No water or metal were added to the electron density near M2, although we suspect that this electron density represents an alternate position for M2 (see Fig. 5). c, Lys 183 -Met/xylose. ␣-D-Xylopyranose was modeled and refined in Lys 183 -Met. The B factors for the sugar, M1, and M2 were 16, 12, and 13 Å 2 , respectively. The binding of ␣-D-xylopyranose is comparable to wild-type/THG, except ␣-D-xylopyranose lacks C6 and O6. The sugar hydroxyls O3 and O4 are ligated to M1 at 2.4 and 2.3 Å, respectively. d, and 100-l aliquots were removed at various times. The activity of wild-type, His 54 -Asn, or Phe 94 -Ser enzymes on the mutarotating D-xylose (200 mM final D-xylose concentration) was determined in the sorbitol dehydrogenase assay, described above.
Crystallography-Crystallization of D-xylose isomerase was as described previously (34). Crystals of D-xylose isomerase were stored in 1 mM MnCl 2 , 2 mM Pipes (pH 7.2), and 2 M ammonium sulfate. The concentrations of D-xylose, THG, and xylitol used in crystal soaks containing 1 mM MnCl 2 , 2 mM Pipes (pH 7.2), and 2 M ammonium sulfate are shown in Table I, along with crystallographic and refinement data. X-ray diffraction data from crystals of His 54 mutants with and without D-xylose were measured on a Nicolet X100A area detector, the data

FIG. 2-continued
His 220 -Glu/xylose. ␣-D-Xylopyranose was modeled and refined in His 220 -Glu. The B factor for the sugar is 13 Å 2 . Both metals are observed and metal ligation is similar in the absence of sugar (34), except M1 is liganded by O3 and O4 of sugar at 2.4 and 2.2 Å, respectively. Glu 220 does not ligand M2, but can interact with O3 of sugar at 3.0 Å. e, His 220 -Ser/xylose. ␣-D-Xylopyranose was modeled and refined in His 220 -Glu. The B factor for the sugar is 14 Å 2 . Both metals are observed and metal ligation is similar to that observed in the absence of sugar (34), except M1 is liganded by O3 and O4 of sugar at 2.3 and 2.2 Å, respectively. Two water molecules (Wat 700 and Wat 701 ) are located near Ser 220 and one water (Wat 700 ) replaces His 220 NE2 as a ligand to M2. f, His 220 -Ser/xylitol. The orientation of xylitol is such that O1 interacts with Lys 183 NZ and O5 hydrogen bonds to His 54 NE2. M1 is liganded by O2 and O4 at 2.2 and 2.3 Å, respectively. The B factor for xylitol is 6 Å 2 . M2 is absent, alternate conformations (ALT1 and ALT2) for the side chain of Asp 255 and two new water molecules (Wat 706 and Wat 707 ) are shown. Additional differences are shown and described in Fig. 7.
were integrated and scaled using the XDS software, and structures were refined using restrained least-squares refinement, as described in Cha et al. (34). Diffraction data from all other crystals of mutants were measured on a San Diego Multiwire systems area detector and the structures were refined using the X-PLOR suite of programs, as described previously (34).

RESULTS
Cloning and Expression of S. rubiginosus xylA-Two oligonucleotide primers, one of which was complementary to the 5Ј end and the other complementary to the 3Ј end of S. rubiginosus xylA (12), were used in a PCR of S. rubiginosus chromosomal DNA. The annealing temperature for the PCR was varied from 37 to 70°C, and MgCl 2 was added at 1.5-4 mM, but no products were amplified. Some DNA sequences with a high G ϩ C content cannot be amplified unless 7-deaza-dGTP is included in the reaction mixture (43). When 50% of the dGTP in the reaction mixture was substituted with 7-deaza-dGTP, a PCR product with the expected size (1.2 kilobases) was amplified. The DNA sequence of the cloned PCR product had differences at nucleotides 1735, 1857, and 2562 from the reported sequence of Wong et al. (12). The differences at 1857 and 2562 were silent mutations (both TTT instead of TTC), but the difference at nucleotide 1735 corresponded to a Phe3 Leu mutation (TTC to CTC) at residue 13 of the protein. The PCR product was cloned into pET11d, under the control of the T7 promoter and LacZ, and several transformants were obtained that expressed a 43-kDa protein only upon addition of isopropyl-1-thio-␤-D-galactopyranoside. This protein migrated the same in SDS-polyacrylamide gel electrophoresis as non-recombinant D-xylose isomerase, but this protein from the cell lysate was not soluble and had no enzymatic activity. To see if Leu 13 may have caused the insolubility of recombinant D-xylose isomerase, it was mutated to Phe. Also, to potentially improve protein expression, the AGG codon for Arg 10 , rarely observed in E. coli (44), was replaced with the more common CGT codon. The Phe 13 recombinant D-xylose isomerase (wild-type) was soluble and had kinetic properties identical to non-recombinant D-xylose isomerase (34). This expression system was used to overexpress D-xylose isomerases containing active site mutations. As summarized in Table I, six of these mutant enzymes have been kinetically characterized and crystallographically studied alone and in complex with D-xylose, and in one case, with xylitol.
Wild-type D-Xylose Isomerase Complexed with THG-The structure of the Arthrobacter D-xylose isomerase complexed with the cyclic inhibitor THG (K i ϭ 33 mM) showed that His 54 is approximately 3 Å from both the ring sulfur and atom O1, the anomeric oxygen (16). The structure of the S. rubiginosus D-xylose isomerase with THG was determined to assess the His 54 interactions in this enzyme with a cyclic ␣-pyranose sugar. THG can clearly be seen in the electron density map (Fig. 2a). The binding of THG in S. rubiginosus is comparable to that observed in the Arthrobacter enzyme, except His 54 NE2 is 2.9 Å from the ring sulfur and 3.5 Å from the anomeric hydroxyl and two alternate positions for O6 are seen in the electron density (Fig. 2a). One position of O6 (ALT1) is nearly the same as that reported by Collyer et al. (45) and forms a hydrogen bond to Thr 90 OG. In the other position (ALT2), O6 interacts with Wat 408 . The different positions of O6 did not affect the hydrogen bonding or metal ligation of the other sugar hydroxyls. The B factors for M1 and M2 were 10 and 7 Å 2 , respectively, indicating that the occupancy of both metals is high and their mobility is low.
pK a of His 54 -The ring-opening mechanism proposed by Collyer et al. (16) and Whitlow et al. (15) involves His 54 acting as base to remove a proton from the anomeric oxygen of the substrate. However, in the wild-type/THG structure described above, His 54 is closer (2.9 Å) to the ring sulfur than the anomeric hydroxyl (3.5 Å), suggesting that it might acid catalyze ring opening of the analogous sugar, D-xylose, via the protonation of the ring oxygen. Also, the His 54 imidazole is held in place by a hydrogen bond to the carboxylate of Asp 57 (36) which may help to raise the pK a of His 54 . The protonation state of His 54 was determined by chemical modification with DEPC, a reagent that reacts with histidines only in their deprotonated form (46). Previous studies had shown that His 54 is responsible for DEPC sensitivity (31,(47)(48)(49), and this was reconfirmed in this study. The wild-type enzyme retained only 9.3% activity when incubated with DEPC (0.157 mM) at 25°C for 10 min at pH 7.0, while His 54 -Ser and His 54 -Asn mutants (described below) retained 100% activity after incubation with DEPC. To determine the pK a of His 54 , the rate of DEPC inactivation was measured in the pH range of 5.5-7.5 (Fig. 3). The inflection point (pK a ) from two separate DEPC inactivation experiments was 6.40 Ϯ 0.01. Parallel studies, using the same enzyme preparation, yielded a pH optimum for catalysis of D-xylose between pH 8 and 9 (Fig. 3).
Kinetic Properties of His 54 Mutants-To examine the role of His 54 in anomeric recognition, ring opening, and isomerization, His 54 was changed to Ser, Asn, and Asp. The His 54 -Ser and His 54 -Asn mutants had activity (Table II), while His 54 -Asp was insoluble and crude cell extracts containing this mutant enzyme had no activity. The k cat values for the His 54 -Ser and His 54 -Asn enzymes on D-xylose were 12-and 5-fold lower and the K m values were 2-and 7-fold higher than that observed for wild-type. It has been suggested that the lower k cat of His 54 mutants is due to the loss of a hydrogen bond to O5 of the acyclic-extended substrate (30,31). The importance of this interaction was tested by examining the kinetics of D-xylose isomerase on the acyclic sugar 5-deoxy-D-xylulose. The His 54 mutants had k cat values on the 5-deoxy-sugar similar to D-xylulose, while the wild-type had a 20-fold lower k cat on the deoxy-sugar than on D-xylulose (Table II). The K m values on the 5-deoxy-sugar for the mutants (22.5 and 35.0 mM) were similar to the value for the native enzyme (22.0 mM).
Anomeric Specificity of D-Xylose Isomerase-D-Xylose isomerase preferentially binds the ␣-pyranose form of aldose (24,25). It has been reported that His 54 is important for recognizing the ␣-pyranose (30) and that the Phe 94 side chain prevents the ␤-pyranose form from binding by steric exclusion of the O1 hydroxyl (15,50). To possibly allow ␤-pyranose binding, the Phe 94 side chain was replaced with the smaller hydrophilic Ser residue. The kinetics of this mutant show a 5-fold decrease in the k cat and a 7-fold increase in K m with D-xylose as a substrate  (Table I). The roles of His 54 and Phe 94 in anomeric recognition were directly explored by dissolving crystalline ␣-D-xylose in water and assaying relative activity, using experimental conditions similar to Lambier et al. (30). Both wild-type and Phe 94 -Ser had the highest activity immediately after D-xylose was dissolved and showed diminished activity as the ␤-anomer was forming, until an activity equilibrium was reached (Fig. 4). His 54 -Asn retained 100% activity over time (Fig. 4), indicating that this mutant has no preference for the ␣-pyranose over the ␤-pyranose and is consistent with the results of Lambier et al. (30).
Structures of His 54 -Asn and His 54 -Ser-In wild-type D-xylose isomerase, His 54 NE2 hydrogen bonds to one water (Wat 413 ) and His 54 ND1 interacts with Asp 57 OD1 (36). The structures of His 54 -Ser and His 54 -Asn show only slight structural differences, and these differences are localized to the region around the substituted histidine. The electron densities of the His 54 -Ser and His 54 -Asn side chains appear well ordered and the average temperature factors for the Ser 54 and Asn 54 side chains are 17 and 13 Å 2 , respectively. In the His 54 -Ser structure, the side chain hydroxyl of Ser 54 rotated approximately 150°from the position observed for His 54 CG. In this position, Ser 54 OG hydrogen bonds to the amide nitrogen and the OD carboxylate of Asp 57 at 3.1 and 2.8 Å, respectively. One new water molecule (Wat 708 ) is observed 1.4 Å from the position normally occupied by His 54 NE2 and it forms a hydrogen bond to Wat 413 and Wat 591 . The latter solvent molecule is 2 Å from its position in the native enzyme. In His 54 -Asn, the Asn 54 ND hydrogen bonds to Asp 57 OD1 at 2.7 Å, replacing the His 54 ND1-Asp 57 OD1 interaction, and Asn 54 OD hydrogen bonds to Wat 585 at 2.9 Å. No other differences greater than 0.3 Å for either structure were observed.
Structure of His 54 -Ser Complexed with D-Xylose-In many structures of wild-type D-xylose isomerase complexed with D-xylose or D-xylulose, His 54 NE2 hydrogen bonds to O5, Lys 183 NZ interacts with O1, and both O2 and O4 are liganded to M1 of the extended sugar (15,36). Although 1.5 M D-xylose was used to soak a His 54 -Ser crystal, the electron density contributed by the sugar is weak; however, an acyclic-extended conformation of sugar can be seen in the F o Ϫ F c map (Fig. 2b). Placement of the sugar into the electron density reveals that hydrogen bonding to the protein and metal liganding are comparable to a wild-type/xylulose structure (36), except there is no hydrogen bond between Ser 54 and O5 (Fig. 2b). The B factors of both metals are significantly higher (36 Å 2 for both M1 and M2) than that observed in the wild type-xylulose complex (7 Å 2 for M1 and 6 Å 2 for M2, (36)) suggesting that either the metals are more mobile or their occupancy is lower. The higher temperature factors of the metals might be caused by destabilization of the sugar in the active site (see "Discussion").
There is significant, unaccounted for electron density 3.5 Å from M1 and 1.9 Å from M2 (Fig. 2b). Since metal cannot occupy both the modeled position for M2 and the position at the excess density, it appears that this is an alternate site for M2.
An equivalent, low occupancy site for M2 (M2Ј) has been described for the native S. rubiginosus D-xylose isomerase complexed with D-xylose (15) and is observed in our wild-type D-xylose complex. The M2Ј position has been proposed to stabilize the transition state of the acyclic-extended sugar in the metal-mediated hydride shift mechanism through interactions with O1 and O2 of sugar (15).
In our wild-type structure, the relative difference peak heights calculated from a F o Ϫ F c map with M2 and M2Ј omitted is 0.12 e/A 3 for M2 and 0.0092 e/A 3 for M2Ј. This yields an estimated M2:M2Ј ratio of 13:1. Since the occupancy of M2Ј is 13-fold lower and side chain atoms and water molecules have fewer electrons than Mn 2ϩ , low occupancy positions for the ligands of M2Ј would not be seen above the noise in the electron density, as noted by others (15). The potential liganding of the metal at M2Ј by Wat 409 , His 220 NE2, both carboxylate oxygens of Glu 217 , and O1 and O2 of sugar is shown in Fig. 5.
Structure of Phe 94 -Ser-In the wild-type enzyme, Phe 94 is in the active site near the side chains of Trp 137 and Phe 26* (*, from a neighboring subunit). As a result of replacing the larger Phe with the smaller Ser, two alternative positions for Ser 94 OG and two new waters (Wat 686 and Wat 687 ) are observed in the Phe 94 -Ser structure. In addition, the side chain of Trp 137 moved 0.5 Å toward Ser 94 and Phe 26* rotated Ϫ13 and 9°about 1 and 2 . No other changes greater than 0.3 Å were observed when compared to the wild-type enzyme.
Structure of Phe 94 -Ser Complexed with D-Xylose-In the wild-type/THG structure, CZ of Phe 94 is 3.8 Å from C1 of THG. In structures of S. rubiginosus D-xylose isomerase with an acyclic-extended pentose sugar bound in the active site (15,36), Phe 94 is far from C1 (6.6 Å) and is closest to C3 (5.4 Å) of the sugar. When D-xylose is added to Phe 94 -Ser, the electron density attributed to the sugar is disordered. Since several different conformations of cyclic, pseudo-cyclic (i.e. the sugar ring is cleaved and the sugar is somewhat cyclic) and acyclic-extended forms of D-xylose could fit into the electron density, sugar was omitted from the model. The position of M1 and M2 shift 0.2 and 0.4 Å, respectively, when D-xylose is added, but the B factors of M1 and M2 remain low (Ͻ10 Å 2 ). The other differences in the D-xylose complex are 0.4-Å shifts in the side chains of the metal ligands Glu 181 , Glu 217 , Asp 255 , and Asp 287 , equivalent to those observed in the wild-type/THG structure.
Structure of Lys 183 -Met-In the metal-mediated hydride shift mechanism, Lys 183 would assist in holding the substrate in the proper orientation and polarizing O1 by hydrogen bonding (15,16,30). In the cyclic-hydride transfer, Lys 183 has no direct catalytic role (27). Lys 183 was substituted with Met and

FIG. 4. Anomeric preference of wild-type (E), Phe 94 -Ser (⌬), and His 54 -Asn (f) D-xylose isomerase.
Crystalline ␣-D-xylopyranose (96%) was dissolved, aliquots were removed, and the relative rate on the mutarotating sugar was assayed at different time intervals by the coupled-sorbitol dehydrogenase assay, as described under "Materials and Methods." this mutant enzyme had no detectable activity (Table I). In wild-type D-xylose isomerase, Lys 183 NZ hydrogen bonds to Asp 255 OD1, Glu 186 O, and Wat 565 (Fig. 6, thin lines). Since Asp 255 is a M2 ligand, it was possible that the Met 183 substitution indirectly abolished activity by perturbing the structure around M2. The Lys 183 -Met structure was determined and no large differences (Ͼ0.4 Å) in the conformations of residue 183 or 255 were observed, however, major structural changes occurred outside the active site at Glu 186 and Pro 187 (Fig. 6) and to a lesser extent in the active site at Phe 26* and Trp 137 .
The loss of the Lys 183 -Glu 186 hydrogen bond and the increased van der Waals radii associated with the Met side chain leads to the most dramatic structural change, with the Glu 186cis-Pro 187 peptide bond fliped from the cis to the trans conformation (Fig. 6). The Glu 186 / angles change from 76/107°to 116/Ϫ65, both of which are angles normally not observed for non-glycine residues (51). The energy of a Xaa-trans-Pro peptide bond is roughly 5-fold more favorable than a Xaa-cis-Pro peptide bond (51).
Structure of Lys 183 -Met Complexed with D-Xylose-When D-xylose is soaked into a Lys 183 -Met crystal, electron density indicative of the cyclic ␣-D-xylopyranose is observed (Fig. 2c). The binding of ␣-D-xylopyranose is comparable to that of THG in wild-type, except ␣-D-xylopyranose lacks C6 and O6. There is no significant movement (Ͼ0.2 Å) of either metal in the Lys 183 -Met/xylose structure compared to either the Lys 183 -Met or the wild-type/THG structures. His 54 NE2 is 2.7 Å from the ringoxygen and is 3.4 Å from the anomeric oxygen. The positions of Glu 186 , Pro 187 , Phe 26* , and Trp 137 remain as in the unliganded Lys 183 -Met structure.
Structures of His 220 -Glu and His 220 -Ser Complexed with D-Xylose-In wild-type D-xylose isomerase, His 220 is the terminal residue on a short ␣-helix that includes residues 216 -220. At neutral pH, the imidazole ND1 is protonated and hydrogen bonded to Pro 182 O, and NE2 of the imidazole is ligated to M2. It was previously reported that substitutions to His 220 result in almost a complete inactivation of D-xylose isomerase (Table I) (34) and a decrease in thermostability. Structures of His 220 -Ser, His 220 -Asn, and His 220 -Glu show that both M1 and M2 are still bound, but there are some perturbations (0.3 to 0.4 Å differences from the wild-type) in the protein around M2 (34). Crystal structures of His 220 -Glu and His 220 -Ser with D-xylose were determined to discover why these mutants have almost no activity. Both His 220 mutants have clear ligand electron density into which ␣-D-xylopyranose (Fig. 2, d and e) was fitted and refined at full occupancy. The positions of M1 and M2 in the His 220 -Glu and His 220 -Ser do not differ greater than 0.3 Å from the sites they occupy in the wild-type/THG structure. The temperature factors for M1 are very similar in the wild-type/ THG, His 220 -Ser/xylose, and His 220 -Glu/xylose structures (all 8 to 10 Å 2 ); the temperature factor for M2 is slightly higher in His 220 -Ser/xylose (14 Å 2 ) than wild-type/THG (7 Å 2 ) and is significantly higher in His 220 -Glu/xylose (26 Å 2 ), indicating that M2 is either more mobile or has a lower occupancy in the His 220 mutants.
The largest change observed when D-xylose is added to His 220 -Glu, is a 0.4-Å shift in the position of Asp 287 OD. Glu 220 does not replace the function of His 220 NE2 in serving as a ligand to M2 (Fig. 2d), as previously observed in His 220 -Glu without sugar (34), but can interact with O3 at 3.0 Å. The position, coordination, and temperature factors for both metals is similar in His 220 -Glu with or without D-xylose (34).
The structure of His 220 Ser/xylose shows only slight rotation of Trp 137 and Phe 26* , but these positions are observed in the wild-type/THG structure. There were no other significant changes (Ͼ0.3 Å) in the positions of the metals or their ligands compared to His 220 -Ser without sugar (34). The two new water molecules seen near the Ser 220 side chain, Wat 700 and Wat 701 as well as the ligation of M2 by Wat 700 in His 220 -Ser/xylose (Fig. 2e), are also observed in His 220 -Ser without D-xylose (34).
Structures of His 220 -Ser Complexed with Xylitol-Xylitol is an acyclic polyol inhibitor of D-xylose isomerase. The K i value of xylitol for the Streptomyces violaceoruber D-xylose isomerase is 0.45 mM (52). The structure of wild-type S. rubiginosus D-xylose isomerase complexed with xylitol has been reported and xylitol is observed bound in an extended conformation and oriented such that O1 hydrogen bonds to Lys 183 NZ and O5 with His 54 NE2 (14,15). Both metals are observed and M1 is liganded by the sugar hydroxyls O2 and O4.
Xylitol was added to His 220 -Ser to mimic the acyclic sugar binding typically seen in native D-xylose isomerase-xylose complexes. The electron density of xylitol is well defined and xylitol binds in an extended conformation (Fig. 2f). However, significant structural changes are observed in and near the active site of His 220 -Ser/xylitol ( Fig. 2f and Fig. 7). These changes include the disappearance of M2, the addition of two new waters (Wat 706 and Wat 707 ), alternative conformations for the side chains Asp 255 and Glu 186 , rotation of Phe 26* , and changes in the position of both M1 and its carboxylate ligands. The electron density attributed to metal (Mn 2ϩ ) at both M1 and M2 in all of the structures described in this work can easily be seen at 12 to 15. However, in His 220 -Ser/xylitol only density around M1 is seen at 12 to 15, and no peaks greater than 5 are observed around the M2 site, strongly suggesting that M2 is not bound. In order to accommodate O1 of xylitol, Wat 700 has moved 1.8 Å, and is only 1.0 Å from the site once occupied by M2. Glu 217 OE2 shifted 1.2 Å and now hydrogen bonds to Wat 700 at 2.9 Å and to a new water molecule (Wat 707 ) at 2.7 Å. DISCUSSION The S. rubiginosus xylA was cloned via PCR and expressed in E. coli. The cloned PCR fragment contained several nucleotide sequence differences, one of which gives rise to a Phe 13 3 Leu mutation. In S. rubiginosus D-xylose isomerase, Phe 13 and the side chains of Val 18 , Leu 43 , Phe 286 , and Phe 288 are buried in a hydrophobic pocket, approximately 8.0 Å from the active site. Although we would not predict the Leu 13 substitution to affect activity, it resulted in the formation of insoluble inclusion bodies. After Leu 13 was changed back to Phe, soluble, active D-xylose isomerase was obtained, suggesting that the Phe 13 -Leu mutation caused problems with protein folding.
Isomerization-A cyclic-hydride transfer isomerization mechanism has been proposed ( Fig. 1b; Ref. 27) to unify the observations: (i) that rate-limiting step in the reaction is the C2-C1 hydride transfer and (ii) both the wild-type and mutant ␣-xylose isomerases from Clostridium thermosulfurogenes display a 2.5-fold difference in k cat for ␣-D-glucose as compared to ␤-D-glucose (45). Because ring opening of both ␣ and ␤-D-glucose produces chemically identical molecules, Meng et al. (27) argued that the different k cat values necessitate that the hydride shift must be concerted with ring opening.
While we do not have an alternative explanation for the results of Meng et al. (27) the data from our study and others are not readily consistent with the cyclic-hydride transfer mechanism. First, Lys 183 -Met, His 220 -Ser, and His 220 -Glu complexed with D-xylose and wild-type/THG show that O2 is 4.5 Å from the closest of the carboxylate oxygens (OD 2) of Asp 287 , and that this carboxylate oxygen is liganded to M1. It is highly unlikely that Asp 287 could be a base catalyst given the orientation of the ␣-pyranose. Second, in the cyclic-hydride transfer mechanism, M2 and Lys 183 would have no direct function in catalysis, even though our biochemical and crystallographic data indicate that both are important catalytically. The structure of His 220 -Ser and His 220 -Asn do not show any major structural perturbations at M1 or Asp 287 , yet these mutants are almost totally inactive (34). Structures of His 220 -Ser, His 220 -Glu, and Lys 183 -Met with D-xylose have ␣-D-xylopyranose in the active site and display binding similar to wild-type/THG. Additional evidence for a role of M2 in catalysis is shown by substitutions at residues located near M2 ligands. Substitutions to Lys 294 (Lys 289 in S. rubiginosus) and Glu 186 change the metal specificity and pH profile of the A. missouriensis D-xylose isomerase (30,35) and structures of the Glu 186 -Gln mutant with different metals show differences at the M2 site but not at M1 (35). Lys 183 is essential for activity. Replacement of Lys 183 with Met, Ser, Gln, or Arg renders the enzyme inactive (30). The structures of Lys 183 -Met reveals no dramatic changes to M1, M2, or their ligands and is further evidence against the cyclic-hydride transfer. However, the Lys 183 -Met structures should be interpreted with caution since other structural changes occur in this enzyme.
With the exception of the results of Meng et al. (27), the aforementioned data and observations are consistent with the proposed metal-mediated hydride shift mechanism (Fig. 1a) (15,16). When D-xylose, D-glucose, xylitol, or sorbitol are added to D-xylose isomerase, an acyclic-extended form of sugar is observed which may represent substrate, product, or intermediate(s) (15,16,18,19,21,26,35,36). The orientation of substrate is such that O1 is hydrogen bonded to Lys 183 NZ, O5 is hydrogen bonded to His 54 NE2, and O2 and O4 of the extended sugar are liganded to M1 (Fig. 1a). In the metal-mediated hydride shift mechanism proposed by Collyer et al. resulting negative charge on O2 would be stabilized by both M1 and M2 interactions. The C2 hydrogen would transfer directly to the partial positively-charged C1 and either a water molecule or Lys 183 NZ might protonate O1. Lys 183 NZ would help stabilize the sugar by interacting with O1 of the acyclic-extended substrate, intermediate(s), and product.
The results from this study and the possible functions of His 54 , Phe 94 , Lys 183 , and His 220 in the reaction are discussed below in terms of the metal-mediated hydride shift mechanism.
His 54 and Phe 94 -Prior to isomerization, sugar binding and ring-opening must occur. In solution, D-xylose and D-glucose exist predominately as hemiacetals, forming six-member pyranose rings with two anomeric forms (␣ and ␤) that differ in their stereochemistry at C1. NMR and kinetic experiments have shown that D-xylose isomerase prefers the ␣-pyranose form of hemiacetal (24,25). His 54 appears to be one determinant of the preference for the ␣-pyranose, by interacting with the anomeric hydroxyl (30). The results with His 54 -Asn suggests that this mutant enzyme has no preference for the ␣-pyranose over the ␤-pyranose and is consistent with the proposal of Lambeir et al. (30). It was proposed that Phe 94 could be involved in anomeric selection by preventing ␤-pyranose binding due to unfavorable interactions between the hydrophilic anomeric hydroxyl and the hydrophobic phenyl ring (15,50). Although changing the phenyl side chain to the smaller Ser decreased activity (Table I), this mutation did not affect anomeric specificity, indicating that Phe 94 is not a major determinant in anomeric recognition.
Phe 94 does have a role in maintaining the structure of the active site, sugar binding, and stabilization of the transition state. The structure of Phe 94 -Ser shows two new waters and changes in the positions of both the nearby hydrophobic side chains of Phe 26* and Trp 137 . Upon addition of D-xylose, disordered density is observed in the active site, into which different forms of sugar could be fit. As mentioned previously, the Phe 94 -Ser mutation has a reduced k cat and an increased K m (Table I). These results suggest that Phe 94 is involved in stabilization of the acyclic extended transition state, through hydrophobic interactions directly with the sugar or indirectly by interacting with the nearby Trp 137 and Phe 26* side chains which in turn contact the sugar.
After binding of the ␣-pyranose, D-xylose isomerase catalyzes ring opening (25). It was suggested that His 54 NE2 could act as a catalytic base, abstracting a proton from the anomeric oxygen (O1) of the ␣-pyranose, and facilitating sugar ring cleavage (15,16). However, in the wild-type/THG structure, His 54 NE2 is closer to the ring-sulfur of THG (2.9 Å) than the anomeric hydroxyl (3.5 Å), suggesting that the His 54 imidazole could act as a acid catalyst and thus protonate the ring oxygen. The pH dependence of DEPC modification indicates that the pK a of His 54 is 6.40 Ϯ 0.01. From kinetic studies of the Mg 2ϩ -activated Arthrobacter D-xylose isomerase with fructose as a substrate, it was reported that the pK a for a group controlling K m was 6.2 Ϯ 0.1 and it was suggested that this group was His 53 (His 54 in S. rubiginosus) (53). The pK a of His 54 determined by DEPC inactivation is consistent with the kinetic study of the Arthrobacter enzyme and indicates that at the pH optimum of the enzyme (near 8.0 -9.0), His 54 NE2 is deprotonated and probably could not be an acid catalyst in ring opening. Changing His 54 to non-basic residues (Ser, Ala, Asn, and Gln) reduces activity, but the rate-limiting step in the overall reaction has been reported to be isomerization, not ring opening (30,31). The acyclic sugar 5-deoxy-D-xylulose can be used as a substrate, indicating that enzyme-catalyzed ring opening is not a step absolutely required prior to isomerization. Other than crystallographic observations noting the proximity of His 54 to cyclic sugars (15,16), there is no biochemical evidence suggesting that His 54 catalyzes ring opening. Initial site-directed mutagenesis experiments of the E. coli D-xylose isomerase suggested that His 101 (His 54 in S. rubiginosus) might be catalytically important in a cis-enediol mechanism, as replacement with Arg or Tyr rendered the enzyme inactive (28). However, when His 54 was replaced with smaller residues (Ala, Ser, Asn, Asp, Glu, and Gln), these mutant enzymes retained activity (29 -31), proving that His 54 is not crucial for catalysis and providing further evidence against the cis-enediol mechanism.
His 54 does, however, interact with the transition state. In most crystal structures of D-xylose isomerase complexed with D-xylose, His 54 interacts with O5 of an acyclic-extended form of sugar. Biochemical evidence for a His 54 -sugar interaction is shown by comparing the kinetics of the wild-type enzyme and His 54 mutants on D-xylose, D-xylulose, and 5-deoxy-D-xylulose (Table II) on D-xylose and D-xylulose when compared to the native enzyme. This is likely due to the loss of a hydrogen bond from O5 of the extended sugar to the substituted side chain. The interaction with the C5 hydrogen of 5-deoxy-D-xylulose and His 54 NE2 of the native enzyme is clearly unfavorable, as reflected in both the diminished k cat and elevated K m . A possible reason for the His 54 mutants having a similar k cat on D-xylulose and 5-deoxy-D-xylulose is that their smaller side chains cannot hydrogen bond to O5 of D-xylulose, and the replaced side chains are not close enough to interact unfavorably with the C5 methyl group of 5-deoxy-D-xylulose.
Additional evidence for the importance of the hydrogen bond from His 54 to O5 of the extended substrate can be inferred from the His 54 -Ser/xylose structure. The Ser 54 side chain is 6.9 Å away from O5 of the extended sugar and thus cannot form a hydrogen bond to substrate. The electron density contributed by the sugar in the active site is weak and the temperature factors of the metals are high, possibly because the hydrogen bond between substrate is lost and/or the His 54 -Ser mutation makes the active site larger.
Lys 183 and His 220 -As mentioned previously, virtually all reports of wild-type D-xylose isomerase complexed with D-xylose have shown an acyclic conformation of sugar bound to the enzyme. The exception where ␣-D-xylopyranose was observed was attributed to the low occupancy of both metals (15). The ␣-D-xylopyranose form is observed in Lys 183 -Met, His 220 -Ser, and His 220 -Glu mutants none of which have appreciable activity (Table I). The Lys 183 -Met mutant is likely blocked at isomerization, since D-xylose exists in small amounts in solution as an acyclic form (0.3% in unbuffered solutions (54)).
There are several possible reasons for observing cyclic sugar in these mutants: (i) the extended form of sugar represents product, (ii) the mutated side chains directly or indirectly block the ring opening step, or (iii) the binding energy of cyclic (or pseudo-cyclic) sugar is improved to the binding energy of the extended substrate. Computer simulations of D-xylose isomerase with D-xylose estimated that there is an increase of 8 kcal/mol when D-xylose goes from pseudo-cyclic (O3 and O4 ligated to M1) to acyclic (O2 and O4 ligated to M1) conformations (55). Entropic contributions were not included in their calculation and might lower this estimate. One possibility is that the acyclic form of sugar observed bound to the wild-type D-xylose complexes is actually D-xylulose (15). Since Lys 183 -Met has no activity, and His 220 -Ser and His 220 -Glu have only 0.3 and 0.5% activity (Table I), respectively, it might be expected that ␣-D-xylopyranose would be observed. However, one S. olivochromogenes D-xylose isomerase mutant, Glu 180 -Lys (Glu 181 in S. rubiginosus), has no activity but shows an acyclicextended conformation of D-glucose bound in the active site (33). A direct role for either Lys 183 or His 220 in ring opening is unlikely since both are far (Ͼ7 Å) from the anomeric and ring oxygens of substrate. We postulate that ␣-D-xylopyranose is observed in the His 220 mutant structures because this form is energetically more stable in the mutants than the acyclicextended conformation as M2 has dissociated from His 220 -Ser/ xylitol and the altered binding site cannot provide the proper metal geometry and/or enough ligands. In the case of Lys 183 -Met, we believe ␣-D-xylopyranose is observed in the active site because the extended sugar is not stabilized by the absent Lys 183 NZ-O1 sugar interaction and thus the ␣-D-xylopyranose conformation becomes more stable.
Lys 183 could function catalytically in the metal-mediated hydride shift by assisting in the polarization of O1 of the transition state (15,16,30). Substituting the Lys side chain with Met rendered D-xylose isomerase inactive. In studies of the A. missouriensis D-xylose isomerase, substitution with Ser, Gln, and Arg also inactivated the enzyme (30). The absence of activity may, however, result from a structural defect that indirectly affects the function of another catalytically important residue. The structures of both Lys 183 -Met and Lys 183 -Met/xylose have large perturbations at and around Glu 186 and Pro 187 . Other changes are observed at Phe 26* and Trp 137 in the active site. Comparison of the wild-type enzyme complexed with D-xylose (i.e. acyclic-extended sugar) to the Lys 183 -Metxylose complex suggests that the extended sugar could be accommodated in the active site of Lys 183 -Met. These results clearly indicate that Lys 183 is structurally important but they also suggest that Lys 183 has a role in extending the pseudocyclic sugar, stabilization of the acyclic-extended sugar, and isomerization, by interacting with O1 of the sugar.
His 220 NE2 serves as a ligand to M2 when M2 is in either the high occupancy site (M2) or the low occupancy site (M2Ј) (15). Movement to the M2Ј may be concurrent with deprotonation of the C2 hydroxy (56). Perturbing the structure at M2 or M2Ј could affect sugar extension by altering the properties of the metal(s) or the metal-bound Wat 409 , both of which interact with O2 and/or O1 of the extended sugar (15,16,19,26,36). The reason why the His 220 mutants have little activity could be that the introduced side chains cannot stabilize the M2Ј site. Indirect evidence for M2Ј destabilization in His 220 mutants is shown by the weaker binding at M2, measured kinetically (34) and observed crystallographically. Further evidence is provided by the ability of xylitol to eject the M2 when it is soaked into the mutant enzyme.
In contrast to the A. missouriensis His 220 -Asn/xylose structure in which M2 is absent and acyclic-extended D-xylose is observed (19), M2 and ␣-D-xylopyranose are observed in the structures of His 220 -Ser (this study) and His 220 -Glu (this study). The difference in M2 occupancy might be due to different metals used in the two studies; they employed Mg 2ϩ , while in this study Mn 2ϩ was used. Depending upon the substitution, there is a 48 -200-fold decrease in metal affinity with Mg 2ϩ compared to Mn 2ϩ in S. rubiginosus His 220 mutants (34). The reason(s) that different sugar conformations are observed might be because different metals were used (as above) and/or that different substitutions might affect substrate binding or catalysis differently. It is unlikely these mutants operate using a different reaction mechanism. Attempts to recover secondsite mutations that restore partial activity in either the E. coli or S. rubiginosus enzymes have yielded only reversions back to the original amino acid residue (i.e. His 54 -Ser back to His 54 ; data not shown). Not all mutations at His 220 result in the same activity. His 220 -Gln has 3.4% activity, His 220 -Ser, His 220 -Asn, and His 220 -Glu have 0.5-0.8% activity, and His 220 -Lys has no activity (30,34).
His 220 appears to be important for maintaining the structure around M2 when the substrate is extended. Both the structure of His 220 -Ser/xylitol determined in this study and the structure of His 220 -Asn/xylose from A. missouriensis (19) display similar sugar binding, show that M2 is absent, and reveal new positions for metal ligands and other residues which are located in and near the active site. These differences are presumably caused directly and indirectly by bringing O1 and O2 of the extended sugar near a M2 site with lower affinity for metal.

CONCLUSIONS
The functions of active site residues in D-xylose isomerase was investigated kinetically and structurally. Some of the main conclusions and supporting observations are summarized here. 1) His 54 is not essential for catalysis, but appears to be responsible for anomeric recognition and to contribute to the stabilization of the transition state via hydrogen bonding to O5. His 54 may act as a base-catalyst but not likely as an acid catalyst in ring-opening where enzyme activity is maximal. 2) Phe 94 is not important for anomeric recognition or essential for enzyme activity, but it clearly contributes to the optimal binding of the transition state. 3) Lys 183 is important structurally and probably catalytically. Lys 183 -Met has structural perturbations at Glu 186 , Pro 187 , Trp 137 , and Phe 26 and no activity. Because residues located at the catalytic center only experience minor perturbations, we suggest that Lys 183 has a direct role in sugar extension and/or isomerization through interactions with O1 of substrate. 4) Additional, unaccounted for electron density is observed near M2 in the structure of His 54 -Ser/xylose and appears to represent an alternate position for M2 (M2Ј). Comparing the peak heights of M2 and M2Ј in the F o Ϫ F c map of the wild-type enzyme yields a ratio M2:M2Ј of 13:1. Metal at the M2Ј position could be a major contributor to catalysis coordinating both O1 and O2 of the acyclic-extended sugar. The low activity of His 220 -Ser may be due to the inability of Ser 220 to stabilize metal binding at the M2Ј position. 5) His 220 is likely important because it coordinates metal at both the M2 and M2Ј positions, as we suggest it is necessary for catalysis. His 220 -Ser and His 220 -Glu mutants have little activity and their structures have M1, M2, and ␣-D-xylopyranose bound to the active site. Xylitol binds in an extended form to His 220 -Ser, but the metal at the M2 position is lost from the enzyme rather than simply being shifted.