Proc. Natl. Acad. Sci. USAVol. 90, pp. 8459-8463, September 1993Biochemistry

The role of active-site aromatic and polar residues in catalysis andsubstrate discrimination by xylose isomerase

(glucose isomerase/site-directed mutagenesis/enzyme-active site/protein engineering/catalytic effciency)

MENGHSIAO MENG*t, MICHAEL BAGDASARIAN*, AND J. GREGORY ZEIKUS*t§¶Departments of *Microbiology and *Biochemistry, Michigan State University, East Lansing, MI 48824; and Michigan Biotechnology Institute, Lansing,MI 48909

Communicated by T. Kent Kirk, June 11, 1993 (received for review March 18, 1993)

ABSTRACT The functions of individual amino acid resi-dues in the active site of Thernoanaerobacterium thermosulfu-rigenes D-xylose ketol-isomerase (EC 5.3.1.5) were studied bysite-directed substitution. The role of aromatic residues in theactive-site pocket was not limited to the creation of a hydro-phobic environment. For example, Trp-188 provided for sub-strate binding and Trp-139 aflowed for the discriminationbetween D-xylose and D-glucose. Substrate discrimination wasaccomplished by steric hindrance caused by the side chain ofTrp-139 toward the larger glucose molecule. Preference of theenzyme for the a-anomer of glucose depended on the His-101/Asp-104 pair. Wide differences observed in the catalyticconstant (kt) for a- versus i-glucose in the wild-type enzymeand the fact that only the kt for a-glucose was changed in theHis-101-+ Asn mutants strongly suggest that the substratemolecule entering the hydride-shift step is stiUl in the cydicform. On the basis of these results a revised hypothesis for thecatalytic mechanism of D-xylose isomerase has been proposedthat involves His-101, Asp-104, and Asp-339 functioning as acatalytic triad.

D-Xylose ketol-isomerase (EC 5.3.1.5) catalyzes the revers-ible isomerization of D-xylose to D-xylulose as part of thexylose metabolic pathway in microorganisms (1). Due to itsability to use D-glucose as substrate and convert it to D-fruc-tose this enzyme is widely used in industry for production ofsweeteners. Comparison of the primary structures of xyloseisomerases, deduced from the nucleotide sequences ofcloned genes reveal that amino acid residues, considered toplay important roles in the active site of the enzyme areconserved among all species studied to date (2-5). It isbelieved, therefore, that all known xylose isomerases useessentially the same mechanism of catalysis. Two classes ofxylose isomerases, however, may be distinguished. Class I,represented by the enzymes from Arthrobacter, Streptomy-ces rubiginosus, Streptomyces olivochromogenes, and Acti-noplanes missouriensis, has the N-terminal portion shorter incomparison with the class II enzymes, represented by theisomerases from Escherichia coli, Bacillus, and Thermoa-naerobacterium [formerly classified as Clostridium (3)].Crystal structures of the class I enzymes have been deter-mined (4-9), but no three-dimensional structures of the classII isomerases have been determined to date.On the basis of the crystal structure of the Arthrobacter

isomerase and its complexes with different substrates andinhibitors a reaction mechanism for aldose isomerization hasbeen proposed (6, 10). This mechanism included the follow-ing steps: (i) binding of a-D-pyranose substrate to the en-zyme, (ii) ring-opening, presumed to be catalyzed by theHis-53 residue, (iii) conformational rearrangement of sub-strate from pseudocyclic to an extended open-chain form, (iv)

The publication costs of this article were defrayed in part by page chargepayment. This article must therefore be hereby marked "advertisement"in accordance with 18 U.S.C. §1734 solely to indicate this fact.

hydride shift between Cl and C2, assisted by a divalent metalat position [II], (v) conformational rearrangement, ring-closure, and release of product. Our previous work on theisotope effect of D-[2-2H]glucose on the reaction velocity (3),as well as the work of other groups (10), indicated that thetransfer of hydrogen between Cl and C2 is the rate-limitingstep of the isomerization pathway. Indications were alsofound that Trp-139 residue (corresponding to Met-87 inArthrobacter) may be a steric hindrance for accommodationof glucose in the active-site pocket of the enzyme (11).

In the present work we provide further support for thehydride-shift hypothesis formulated by Collyer et aL (6) andextend it by proposing that the hydride shift occurs in thecyclic form of the substrate rather than in the extended form.We also examine the roles of aromatic amino acid residues inthe active site of Thermoanaerobacterium thermosulfuri-genes, including the residue that contributes to the discrim-ination between D-xylose and D-glucose.

MATERIALS AND METHODSStrains, Plasmids, and Chemicals. E. coli strain HB101 [F-

hsdS20 ara-1 recA13 proA12 lacYI galK2 rpsL20 mtl-l xyl-SJ(12) was used for expression of the T. thermosulfurigenesxylose isomerase gene as described (3). E. coli strain TG1 [thisupE hisD A(lac-proAB)/F' traD36 proA+B+ lacIqlacZAM15] and bacteriophage M13mp19 (13) were used forsite-directed mutagenesis and nucleotide-sequence determi-nation. a-Glucose and p-glucose were from Sigma.DNA Manipulation. Site-directed mutagenesis was per-

formed by the method of Sayers et al. (14) using the kit fromAmersham. Nucleotide sequences of the mutant genes wereconfirmed by the dideoxynucleotide chain-terminationmethod (15). His-101 -- Asn (3), Trp-139 -* Tyr, and Trp-139-- Phe (11) mutant enzymes were created previously.

Protein Purification and Steady-State Kinetics. The previ-ous protocol (3) was modified to purify xylose isomerase tohomogeneity (on SDS/PAGE). Briefly, crude cell extractwas incubated at 75°C for 15 min, precipitate was removed bycentrifugation, and supernatant was fractionated by DEAE-Sepharose followed by Sephacryl-300 chromatography (3,11). Phe-145 -- Lys and Trp-188 -* His mutant enzymes wereheated at 60°C for 20 min, and Trp-139-* Lys mutant enzymewas heated at 65°C for 30 min instead of 75°C. The enzymereactions in 1 ml contained 20 mM Mops (pH 7.0), 1.0 mMCoCl2, substrate at concentrations of 0.3-2.5 times the KM,and enzyme at 10-1500 pg. Temperature and reaction con-ditions are listed in the footnote to each table. Reactionproducts were determined by the cysteine/carbazole/

Abbreviations: kt, catalytic rate constant; KM(agjucosc)AFPP apparentKM for a-glucose; KM(pgIucose)AP, apparent KM for p-glucose.tPresent address: Department of Biochemistry, The University ofTexas, Southwestern Medical Center, Dallas, TX 75235.1To whom reprint requests should be addressed.

8459

Proc. Natl. Acad. Sci. USA 90 (1993)

Table 1. Experimental and corrected values for a-glucoseisomerization reaction velocitySubstate concentraion, mMP-Glucose a-Glucose*

Velocity, janol/min per mg

Vtowt V.* VP§300 6.6 0.56 0.33 0.23100 2.2 0.25 0.13 0.1260 1.3 0.17 0.08 0.0940 0.9 0.12 0.05 0.07

*Values were estimated from determination of mutarotation.tApparent values were determined experimentally.tVelocity was calculated from Eq. 4.Velocity was calculated from Yt.tj - V.

sulfric acid method (16). Kinetic constants were determinedfrom both Lineweaver-Burk and Eadie-Hofstee plots (17).We defined the catalytic constant (kc) as the turnovernumber per active site of enzyme at saturating substrateconcentration and determined it from the equation kat[E]o =V., where [EJ. = total active-site concentration.Correction for Spontaneous Mutarotation. In the determi-

nation of V. and KM for p-glucose, the interference froma-glucose, present as impurity and formed by spontaneousmutarotation, was considered because KM(algucose)App < <KM(p.glucose)App, where KM values represent apparent valuesfor a-glucose and p-glucose, respectively. If both anomersare present, fructose may arise from two different reactions:

Vp-glucose J fructose

mutarotation ] |

a-glucose Va fructose.

The initial velocity of fructose formation from p-glucose(VP) may be calculated by subtracting the initial velocity offructose formation from a-glucose (that exists as impurity oris formed from p-glucose by mutarotation), (Va), from theapparent velocity offructose formation (Vs + Va) determinedexperimentally. To account for the spontaneous mutarota-tion we have followed the change in optical rotation ofp-glucose under the same conditions as were used to deter-mine enzymatic glucose isomerization (20 mM Mops, pH7.0/1.0 mM CoCk2; 35°C; 1.5 min) with the Autopol IIautomatic polarimeter (Rudolph Research, Flanders, NJ).Because mutarotation is reversible and its rate constant isindependent of the concentration of sugar over a wide rangeof concentrations, at initial reaction conditions we have

d[a-glucose] - d[p-glucose]dt dt

= k[p-glucose]. [1]

I W136 K182:3 -: 18

D254

FtM87"D244

The reaction constant k, calculated from the measurement ofmutarotation and expressed in decimal logarithms and min-',was 0.0073 ± 0.0009. Because the spontaneous mutarotationrate is faster than the enzyme-catalyzed rate of glucoseisomerization, we assumed that

d[ja-glucose] d[a-glucose]dt (Enzyme) = dt (buffer) = 143glucose].

[2]

The content of p-glucose after 1.5 min of incubation underconditions used for enzymatic reaction was 96.6% of total.The content at 0 time, obtained by extrapolation of themutarotation curve, was 99.1%. As an approximation, there-fore, we can consider that during the first 1.5 min [p-glucose]= constant and

d[a-glucose]- ~~=constant. [31

dt

It is reasonable, therefore, to use the average content ofa-glucose = 2.2%, present in the solution of p-glucose duringthe initial 1.5 min of incubation with the enzyme, to calculatethe apparent initial velocity offructose formation from a-glu-cose, Va, from the equation:

V - Vmax(a-glucose) [a-glucose]

[a-glucose] + Km(,glucose)[4]

where we assumed Vn(aglucose) Vmax(cvgjucose)App andKM(a.glucose) " KM(a-gucose)App.The velocity of fructose formation from p-glucose at a

given concentration of p-glucose, Vq, could thus be calcu-lated and used to calculate Vma(frglucose) and KM(pgluwse). Toshow the corrections obtained by this method one set of datafor the wild-type enzyme is presented in Table 1.The same method was used to correct for mutarotation in

the determination of the V (aguwse) and KM(a-glucose). It wasfound that these corrections were very small.

RESULTS AND DISCUSSIONThe Role of Aromatic Residues in the Active Site of Xylose

Isomerase. In the active site of xylose isomerase from Ar-throbacter (Fig. 1), the substrate analogue (a-5-thio-D-glucose) is sandwiched between Trp-15 and Trp-136. In the T.thermosulfurigenes enzyme these residues correspond toTrp-49 and Trp-188. In addition, the residue Met-87 in theArthrobacter enzyme is conserved as tryptophan in all classII enzymes [Trp-139 in the enzyme from T. thermosulfuri-genes (3)]. This residue is very close to the C6-OH group ofglucose and may constitute a steric hindrance in the discrim-ination between D-xylose and D-glucose as substrate.

Wlj 361~18j K1823 8

<sK ~~D2!

FIG. 1. Stereostructure of the active site of D-xylose isomerase from Arthrobacter containing the substrate analog a-5-thio-D-glucose [fromthe coordinates of Collyer et al. (6)]. The metal ion sites are represented by crosses. Amino acids are indicated in one-letter code.

8460 Biochemistry: Meng et al.

Proc. Natl. Acad. Sci. USA 90 (1993) 8461

Table 2. Kinetic constants for wild-type and mutant D-xylose isomerases substituted in the active-site aromaticamino acids

Glucose Xylose

Enzyme kcat, S-1 KM, mM kcat/KM kcat, S-1 KM, mM kcat/KMWild type 11 ± 2.0 110 ± 8 0.10 23 ± 2.0 9.3 ± 1.8 2.5Trp-139 Tyr 9 ± 0.4 91 ± 12 0.10 10 ± 1.0 14 ± 1.0 0.7Trp-139 Phe 16 ± 1.0 65 ± 7 0.25 24 ± 2.0 16 ± 2.0 1.5Trp-139 Met 11 ± 1.0 55 ± 2 0.20 10 ± 1.0 9.2 ± 0.2 1.1Trp-139 Leu 8 ± 0.4 50 ± 4 0.16 14 ± 1.0 11 ± 1.0 1.3Trp-139 Val 5 ± 0.2 35 ± 1 0.14 8 ± 0.1 6.4 ± 0.1 1.3Trp-139 Ala 8 ± 0.2 31 ± 2 0.26 8 ± 0.2 5.9 ± 0.4 1.4Trp-139 Lys 3 ± 0.2 21 ± 1 0.14 4 ± 0.1 5.4 ± 0.7 0.7Trp-49 Phe 10 ± 0.3 330 ± 14 0.03 ND ND NDTrp-49 Ala 7 ± 0.3 710 ± 48 0.01 ND ND NDTrp-49 Arg 10 ± 1.0 110 ± 3 0.09 ND ND NDWild type (37°C)* ND ND ND 1 ± 0.1 1.1 ± 0.1 0.9Trp-188 His (3rC) ND ND ND 0.5 ± 0.02 820 ± 30 6 x 10-4Phe-145 Lys (37°C) ND ND ND 1 ± 0.1 53 ± 6 2 x 10-2ND, not determined.

*Reactions were performed at 37°C, as indicated in parentheses; all other reactions were run at 65°C. ND, not determined.

As shown in Table 2, substitution of Trp-139 by residueswith smaller side chains resulted in lower KMeUCOSC). Mostmutant enzymes thus obtained exhibited higher catalyticefficiency (kct/Km) for glucose and lower catalytic efficiencyfor xylose than the wild-type enzyme. The correlation be-tween the water-accessible surface of the hydrophobic resi-due in position 139 and the KMIUCOSc) (Fig. 2) suggests that,at least, part of the side chain of Trp-139 protrudes into thecavity of the active-site pocket and constitutes a sterichindrance against the binding of glucose. Reduction of thissteric hindrance creates mutant enzymes that accommodateglucose better than the wild-type enzyme. The Trp-139 -*

Lys mutant fails out ofthe proportionality rule. The polar andpositively charged side chain may cause a local structurechange and thus affect the affinity of substrate indirectly orincrease the affimity directly by hydrogen-bonding to thesubstrate. A methionine residue is present in the correspond-ing position of the enzymes from Arthrobacter (5, 6), Strep-tomyces (4, 8), and Actinoplanes (9). It is possible that theKM@UCOSC) may be reduced for these enzymes if this methi-onine is replaced with a smaller hydrophobic residue.

Substitution ofTrp-49 in the T. thermosulfurigenes enzyme(Trp-15 in the Arthrobacter enzyme) with phenylalanine oralanine residues resulted in a 3- and 6.5-fold increase in

4 on.

100-

0

80-

2t 60-

40 -

100 150 200 250Water-accessible surface area (A

300

FIG. 2. Correlation between the KMW1UCOSC) and the water-accessible surface area ofthe side chain ofthe residue in position 139in T. thermosulfurigenes xylose isomerase. Values for water-accessible surface areas are from ref. 18.

KM@UCOSC), respectively, with no appreciable change in k,.tSurprisingly, the substitution Trp-49 -) Arg did not changeappreciably either KM or k,at (Table 2). According to thestructure of the Streptomyces enzyme the NE1 of this indoleresidue contributes to the construction of the active-sitehydrogen-bonding network (8). The loss of one hydrogenbond in the network is expected to affect indirectly thebinding affinity for the substrate. It is possible that the Ng ofarginine may take a position in which it can perform the samefunction as the Nel of Trp-49.Mutant enzymes obtained by substitution of Trp-188 (Ar-

throbacter equivalent, Trp-136) with Lys, Asp, or Glu ex-hibited no detectable activity with either D-glucose or D-Xy-lose, although the proteins remained soluble under the con-ditions of assay (65°C). The mutant Trp-188 -- His showed alow activity only with D-xylose. The KM(xyIose) of this mutantenzyme at 65°C was >2 M, and thus the kinetic constantscould not be measured precisely. They could, however, bedetermined at 37°C. Under these conditions the KM(xy1ose) was800-fold higher than that of the wild-type enzyme, whereasthe k,at was only lower by a factor of2 (Table 2). These resultsare consistent with the prediction, deduced from the crystalstructure of xylose isomerase, that the hydrophobic interac-tions between the indole ofTrp-188 and the carbon backboneof pyranose contribute strongly to substrate binding in theactive site. The role of tryptophan residues in binding sub-strates has been demonstrated in maltose-binding protein(19), arabinose-binding protein, and galactose-binding pro-tein (20). Substitution of Phe-145 (Arthrobacter equivalent,Phe-93) with lysine resulted in a 50-fold increase of KM(XYlOSe)and an insignificant change in kcat (Table 2). This resultsuggests that Phe-145 also plays an important role in substratebinding. Phe-145 may orient the indole group ofTrp-188 in anoptimal position for substrate interaction.Anomeric Specificity of Xylose Isomerase. The models of

catalytic isomerization suggested by the crystallographicdata include a ring-opening step, presumed to be catalyzed byHis-53 residue acting as a base (6, 8). We have, therefore,tested, by mutagenesis, the role ofthe corresponding residue,His-101, and Asp-104, a residue that may assist the functionof His-101, in T. thermosulfurigenes xylose isomerase. Aswas shown in the previous work (3) and here in Table 3,substitution of the His-101 by residues incapable of acting asa base, but capable of accepting hydrogen bonds, resulted inmutant enzymes exhibiting considerable residual activitywith insignificant change of KM. Similar results have beenreported for the A. missouriensis enzyme (21).

Trp

Tyr

Phe

MetLeu .

Ala

lcu I

20 1 . I I

Biochemistry: Meng et aL

Proc. Natl. Acad. Sci. USA 90 (1993)

Table 3. Comparison of a-D-glucose and 3-D-glucose kinetic constants for variantD-xylose isomerases

a-n-Glucose P-D-Glucose kot/KM(a)

Enzyme k,,, s 1 KM, mM kcats -1 KM, mM kw/KM(pOWildtype 1.30 ± 0.03 24 ± 1 0.25 ± 0.03 136 ± 12 27His-101 Asn 0.15 ± 0.01 30 ± 2 0.26 ± 0.03 130 ± 16 2.5Asp-104 Asn 0.65 ± 0.01 33 ± 1 0.14 ± 0.02 204 ± 18 29Asp-104 Ala 0.08 ± 0.01 45 ± 2 0.27 ± 0.01 275 ± 20 1.8

Reactions were started with freshly prepared substrate and run at 350C for 1.5 min. The figures havebeen corrected as described in Materials and Methods.

The activity of the wild-type and mutant enzymes was alsodetermined for the a- and ,B-anomers of D-glucose. In thewild-type enzyme kcat( glucose) was 5-fold lower than thekCat(<,lU5C.). The substitution His-101 Asn reduced thekc.t(,,g..) to 12% of the wild-type value, whereas kw(plucose) did not change. The KM for either anomer remainedessentially unchanged (Table 3). If the hydrogen transfer,shown previously to be the rate-limiting step (3), occurredafter the opening of the pyranose ring, it would be expectedthat kctwglucose) and kcat(a.g1ucosc) would be of the same orderofmagnitude because anomers do not exist in the open-chainform. Moreover, it would be expected that both constantswould be affected to the same extent by the His-101 Asnmutation. The results presented in Table 3 suggest that thehydrogen transfer occurs when the structure of the substratemolecule is cyclic rather than linear. This conclusion isconsistent with the results of crystallographic studies doneunder steady-state conditions in a flow-cell by Farber et al.(4). The electron densities observed by these authors indi-cated that the rate-limiting step was preceded by a cyclic formof the substrate. We hypothesize that hydrogen transfer andring opening occur as a concerted single-step reaction. Twopossible mechanisms for this step may be considered. In oneof them, a base attracts the proton from the C2 carbon of thepyranose; this results in the formation of a cis-enediol inter-mediate and ring opening during the transfer of proton (22).

His-1011

.-1NH N --------- HO- Transition state

lAsn-1041

\e<°"NH2

lAsn-104Iw .NNH2

o

IHis-10l

N N-H HO- Transition state

FIG. 3. Schematic representation of the interaction betweenHis-101, Asn-104, and the Cl-OH of the transition state. Note thatAsp-104, which occupies this position in the wild-type enzyme, canlock His-101 in a particular tautomeric form, regardless of therotational position of its amide group, whereas Asn-104 can performthis function only when its carboxyl group is in the position repre-sented at the top of the figure.

In the second mechanism, a base attracts the proton from theC2-OH, and this is followed by a hydride shift and ringopening.Two arguments can be raised against the cis-enediol inter-

mediate mechanism: (i) no residue capable of acting as ageneral base has been seen near the C2 hydrogen in theavailable crystal structures; (ii) no exchange of proton withthe medium occurs during the isomerization reaction (23).Therefore, we propose the second mechanism, involving thehydride shift particularly because the crystal structure of theenzyme from Streptomyces (8) indicated that Asp-287 residue(corresponding to Asp-339 of T. thermosulfurigenes enzyme)is close enough to C2-OH of the substrate to attract itsproton.

If the position of His-101 in the T. thermosulfurigenesisomerase is indeed equivalent to the position ofHis-53 oftheArthrobacter enzyme, its major role would be as hydrogen-bond acceptor to stabilize the transition state of the rate-limiting step (3). The Asp-104 residue could assist thisfunction by stabilizing the His-101 residue. As shown inTable 3, substitution Asp-104 -o Asn resulted in a drop of kcatby 501O. The mutant obtained by the Asp-104 -* Alasubstitution, exhibited a kCt(,g1uwse) of only 6% of the wildtype, whereas the kcat(*glucose) remained unchanged. Thisresult suggested that anomeric specificity of xyloseisomerase depends not only on the presence of His-101 butalso depends on the position of this residue. With Asp-104 inplace, either oxygen of its carboxyl group can function ashydrogen-bond acceptor. By providing a hydrogen bond toHis-101, Asp-104 locks His-101 at one of the two possibletautomeric forms, thus ensuring that N82 of His-101 is alwaysa hydrogen-bond acceptor. In the Asp-104 -. Asn mutantonly the oxygen of the amide group can provide this function(Fig. 3). Without the hydrogen bond provided by Asp-104(e.g., in Asp-104 -- Ala mutant) the imidazole of His-101could rotate or take up the tautomeric form that is unfavor-able for the formation of the hydrogen bond to the transitionstate.

In the previously proposed models, the extended-chainmolecules ofthe substrate, identified in the crystal structuresof the enzyme, were interpreted as being intermediates thatprecede the hydride shift (6, 8). It is possible, however, thatthese extended-chain sugar molecules were not intermediatesthat preceded the rate-limiting step. They could be, forexample, the free ketose of xylulose because it is present asa 20.2% fraction in the aqueous solutions of xylulose at

Table 4. Kinetic constants of variant D-xylose isomerasestoward xylose

k:at/KM(mutant)Enzyme kw, s-1 KM, mM kcat/KM(wild type)

Wildtype 23 1.0 9.3 ±2.0 1.0Asp-309 Asn 2.7 ± 0.1 5.8 ± 1.0 0.2Asp-296 Asn 1.0 ± 0.1 140 ± 20 28 x 10-4Asp-339 - Asn 0.08 ± 0.01 70 ± 10 4 x 10-4

Reactions were performed at 65°C for 30 min.

8-162 Biochemistry: Meng et al.

Proc. Natl. Acad. Sci. USA 90 (1993) 8463

M[I] As|AsP-3391

0 H-O'OHOH.

NH N-H ---- -104

I His-1011 [Ap104 |

p-339 | M[IJ--.O1.0

OH"H OH

H,OHI

H~

M[I]MIIOH .

Asp-339 MOH

- -<ov wHOH2CO°\-<0 -----

HO0

N N-H

IHis-1011 Asp-104

NiN-H ----A,pO

I His.1071I | sp14

FIG. 4. Proposed catalytic mechanism for D-xylose isomerase involving the cyclic substrate, amino acid catalytic triad, and divalent metalin position [I].

equilibrium (24). Whitlow et al. (8) actually suggested that theextended species in the enzyme-xylose-MnCI2 structuresobserved in their diffraction pictures may be better describedas xylulose.The Role of Metal-Coordinating Residues. To test the func-

tion ofeach ofthe two metal ions in the active-site pocket, wehave substituted Asp-309, -296, and -339 residues (Arthro-bacter equivalents -256, -244, and 292, respectively; Fig. 1).Asp-309 corresponds to the residue binding the metal atposition [II], whereas Asp-296 and Asp-339 correspond toresidues that bind the metal in position [I]. In addition to itsmetal-coordinating function, Asp-257 in the Streptomycesand Actinoplanes isomerases (corresponding to Asp-256 inthe Arthrobacter and Asp-309 in Thermoanaerobacterium)was proposed to act as a general base that initiated thehydride shift on the linear form of the substrate (8, 25).

Substitution of each of these aspartate residues with as-paragine resulted in enzymes that still required Co2+ formaximal activity and thermostability suggesting that themetal-binding site still exists in these mutants. The Asp-309-- Asn mutant enzyme exhibited =w20%o of the wild-typecatalytic efficiency (kvat/KM; Table 4). This result arguesagainst the hypothesis that Asp-309 might be the essentialcatalytic base that initiates the hydride shift.

Substitutions Asp-296 -. Asn or Asp-339 -. Asn causeddrastic decreases in catalytic efficiency resulting from boththe increase in KM and decrease in ka (Table 4). Thus, thesetwo residues, or the metal[I], seem to play an important rolein the stabilization of the substrate and the transition state.The four order-of-magnitude decrease in kat/KM due to theAsp-339 -- Asn substitution supports the hypothesis thatAsp-339 is the essential base in the catalytic mechanisms ofxylose isomerase.The Proposed Catalytic Mechanism of Xylose Isomerase. In

the previously proposed models it was suggested that thehydride shift is catalyzed by the metal in site [I1 (6). Resultsof this work indicate that substitution of amino acid residuescoordinating to metal site [I] has a much more drastic effecton the activity than the substitution of the residues coordi-nating to metal site [1]. It is possible, therefore, that metal [I]stabilizes the substrate and the transition state by coordina-tion as well as by electrostatic interaction with the developingnegative charge of the transition state. Studies of the coor-dination sphere of the two metal-binding sites by spectro-scopic methods have also suggested that metal site A, cor-responding to the metal site [IE identified in the x-ray studies,is the site responsible for catalysis (26, 27).We would like, therefore, to propose, in Fig. 4, another

model for the reaction catalyzed by xylose isomerase. In thismodel, His-101 is locked in one tautomeric form by interac-tion with Asp-104, and it acts as a hydrogen-bond acceptor tostabilize the substrate and the transition state. Asp-339,acting as a base, attracts the proton from C2-OH of thesubstrate. This attraction facilitates the subsequent hydride

shift from Cl to C2 and simultaneously induces the ringopening. Metal [I] stabilizes the substrate and the transitionstate by coordination and, perhaps, provides the electrostaticforce to stabilize the developing negative charge at the C5-O.The involvement of three different amino acid residues in acatalytic triad is another hypothesis for the xylose isomerasemechanism. Nonetheless, the involvement of a specific cat-alytic triad has been reported for different hydrolytic sac-charidases, such as a-glucanases (28).We thank Paul Johnson for advice on mathematical correction for

mutarotation of anomeric substrate and David Blow for the discus-sions on the structure ofxylose isomerase. This work was supportedby a gant from the U.S. Department of Agriculture (90-34189-5014to Michigan Biotechnology Institute) and Research Excellence Fundfrom the State of Michigan.1.2.

3.

4.

5.6.7.

8.

9.

10.11.

12.13.14.

15.

16.17.

18.19.

20.21.

22.

23.

24.25.26.

27.

28.

Chen, W. P. (1980) Process Biochem. 15, 36-41.Dekker, K., Yamagata, H., Sakaguchi, K. & Udaka, S. (1991) J.Bacteriol. 173, 3078-3083.Lee, C., Bagdasarian, M., Meng, M. & Zeikus, J. G. (1990) J. Biol.Chem. 265, 19082-19090.Farber, G. K., Glasfeld, A., Tiraby, G., Ringe, D. &Petsko, G. A. (1989)Biochemistry 28, 7289-7297.Henrick, K., Coliyer, C. A. & Blow, D. M. (1989) J. Mol. Bio. 208,129-157.Collyer, C. A., Hemick, K. &Blow, D. M. (L990)J. Mol.Biol. 212,211-235.Carrell, H. L., Glusker, J. P., Burger, V., Manfre, F., Tritsch, D. &Biellmann, J. F. (1989) Proc. Nad. Acad. Sci. USA 86, 4440 -dd4.Wbitlow, M., Howard, A. J., Finzel, B. C., Poulos, T. L., Winborne, E.& GilCiland, G. L. (1991) Proteins 9, 153-173.Jenkins, J., Janin, J., Rey, F., Chiadmi, M., van Tilbeurgh, H., Lasters,I., De Maeyer, M., Van Belle, D., Wodak, S. J., Lauwereys, M.,Stanssens, P., Mrabet, N. T., Snawaert, J., Matthyssens, G. & Lam-beir, A.-M. (1992) Biochemistry 31, 5449-5458.Smart, 0. S., Akins, J. & Blow, D. M. (1992) Proteins 13, 100-111.Meng, M., Lee, C., Badasarian, M. & Zeikus, J. G. (1991) Proc. Nati.Acad. Sci. USA 88, 4015-4019.Boyer, H. W. & Roulland-Dussoix, D. (1969) J. Mol. Biol. 41,459-472.Yanish-Perron, C., Vieira, J. & Messing, J. (1985) Gene 33, 103-119.Sayers, J. R., Schmidt, W. & Eckstein, F. (1988) Nucleic Acids Res. 16,791-802.Sanger, F., Nicklen, S. & Coulson, A. R. (1977) Proc. Natd. Acad. Sci.USA 74, 5463-5467.Dische, Z. & Borenfreund, E. (1951) J. Biol. Chem. 192, 583-587.Fersht, A. R. (1985) Enzyme Structure and Mechanisms (Freeman, SanFrancisco), 2nd Ed.Chothia, C. (1976) J. Mol. Biol. 105, 1-14.Martineau, P., Szmelcman, S., Spurlino, J. C., Quiocho, F. A. & Hof-nung, M. (1990) J. Mol. Biol. 214, 337-352.Quiocho, F. A. (1989) Pure Appl. Chem. 61, 1293-1306.Lanbeir, A. M., Lauwrreyes, M., Stanssens, P., Marabet, N. T., Snau-waert, J., van Tilbeurgh, H., Matthyssens, G., Lasters, I., De Macyer,M., Wodak, S. J., Jenkins, J., Chiadami, M. & Janin, J. (1992) Biochem-istry 31, 5459-5466.Makkee, M., Kleboom, A. P. G. & van Bekkum, H. (1984) Recl. Trav.Chim. Pays-Bas 103, 361-364.Bock, K., Meldal, M., Meyer, B. & Wiebe, L. (1983)Acta Chem. Scand.Ser. B 37, 101-108.Wu, J. & Serianni, A. S. (1990) Carbohydr. Res. 2/6, 1-12.van Tilbeurgh, H., Jenkdns, J., Chiadmi, M., Janin, J., Wodak, S.,Mrabet, N. T. & Lambeir, A.-M. (1992) Biochemistry 31, 5467-5471.Sudfeldt, C., Schiffer, A., Kigi, J. H. R., Bogumil, R., Schulz, H.-P.,Wulif, S. & Witzel, H. (1990) Eur. J. Biochem. 193, 863-871.Bogumil, R., Hfttermann, J., Kappl, R., Stabler, R., Sudfeld, C. &Witzel, H. (1991) Eur. J. Biochem. 196, 305-312.Podkovyrov, S. M., Burdette, D. & Zeikus, J. G. (1993)FEBS Lett. 317,259-262.

Biochemistry: Meng et aL