Proton Magnetic Relaxation Studies of the Interaction of D-Xylose and Xylitol with D-Xylose Isomerase CHARACTERIZATION OF METAL-ENZYME-SUBSTRATE INTERACTIONS*

The interaction of n-xylose isomerase purified from two sources with Mn*+ and n-xylose or the competitive inhibitor xylitol has been examined by nuclear magnetic resonance. A greater paramagnetic effect of enzyme-bound Mn2+ on the (Y anomer of n-xylose than on the P anomer was observed, providing independent evidence for the specificity of n-xylose isomerase for the (Y anomeric form of n-xylose. The exchange rate of a-D-xylose into the ternary complex, determined from the normalized paramagnetic contribution to the transverse relaxation rate (l/fT,,) of the carbon 1 proton of a-D-xylose, exceeds V,,,,, for the enzymatic reaction by 3 orders of magnitude. The amount of xylitol necessary to displace a-n-xylose from the substrate.enzyme.Mn2+ complex is consistent with the K, value for a-D-XylOSe and the inhibitor constant K, for xylitol previously determined by the methods of enzyme kinetics. These results suggest that the NMR experiments observe complexes of n-xylose isomerase which are kinetically and thermodynamically

the inhibitor constant K, for xylitol previously determined by the methods of enzyme kinetics. These results suggest that the NMR experiments observe complexes of n-xylose isomerase which are kinetically and thermodynamically competent to participate in catalysis. From the frequency dependence of the paramagnetic contribution to the longitudinal relaxation rate (l/T,,) of the carbon 1 proton of a-n-xylose, the correlation time (7,) which modulates the dipolar interaction between enzyme-bound Mn*+ and cu-n-xylose has been determined (5.1 x lo-'OS). From these observations a range of calculated distances between enzyme-bound Mn2+ and the carbon 1 proton of a-D-xylose (9.1 f 0.7 A) has been found. The enzyme-bound Mn '+ has comparable effects on the carbon 1, carbon 2, and carbon 5 protons of a-D-xylose, suggesting that these protons of the enzyme-bound substrate are equidistant from the bound Mn*+. A similar distance (9.4 f 0.7 A) between the enzyme-bound Mn2+ and the terminal methylene protons of xylitol, an analog of the open chain intermediate in the reaction, has been determined. The results of the present substrate relaxation and previous water relaxation studies suggest that two small ligands such as water molecules or a large portion of the protein intervene between the bound metal ion and the bound substrate in the active ternary complex. n-Xylose isomerase (EC 5.3.1.5), an enzyme which catalyzes the aldose-ketose interconversion of n-xylose and D-xylulose, D-glucose and D-fructose, and D-ribose and D-ribulose, requires a divalent metal ion for activity' (2). Manganese (Mn*+) serves in this capacity for the xylose isomerases from Lactobacihs breois (3,4), and Streptomyces sp. (2), while magnesium (MgZ+) serves only for the latter enzyme (2). The possibility of  ( 1) an enzyme 'metal .substrate bridge complex has been suggested previously (5, .6) from magnetic resonance studies at a single frequency and from consideration of earlier kinetic data (3). The presence of substrates and inhibitors decreased the enhanced effect of the enzyme.Mn*+ complex on the relaxation rate of water protons (5). These observations are consistent with the replacement of water ligands on enzyme-bound MnZ+ and the formation of a bridge complex; alternative explanations have been presented, however (5). Investigation of the effects of an enzyme-bound paramagnetic metal ion on the nuclear relaxation rates of substrates provides a direct and general method for examining enzymemetal-substrate interactions (7,8). Studies at more than one frequency permit the precise estimation of correlation times and metal-substrate distances (9, 10). In this paper the effects at two frequencies of the enzyme.  (13,14). Estimates of molecular weight from gel electrophoresis and data from x-ray crystallographic studies (14) indicate that the Streptomyces enzyme is a tetramer (13,14).  (17) and assignments have been made for the C-l proton of the (Y anomer (o-C, proton) and all protons of the p anomeric form.
The metal-free n-xylose isomerase has no effect on the cu-C I or /3-C 1 resonance ( Fig. 1 and Table I) nor has the enzyme in the presence of the diamagnetic activator Mg'+ (not shown). The paramagnetic Mn*+ .xylose isomerase complex, however, broadens the a-C1 resonances due to an increase in the transverse (l/T,) relaxation rate and increases the radio frequency power required to saturate these resonances due to an increase in l/T,T, ( Fig. 1; Table I). The /3-C, proton resonances are virtually unaffected (Fig. l), which is consistent with the anomeric specificity of this enzyme (4,11). Moreover, the C, proton of cY-n-xylose remains a doublet despite the interaction of the substrate with active enzyme in D,O for several hours, indicating that the C, position remains protonated. This finding, which has been confirmed by direct observation of the C, proton using partially deuterated D-xylose indicates that the proton which is removed from C, is conserved by the enzyme, does not exchange with the solvent, and thus is reversibly transferred to C 1. Kinetic studies showed no incorporation due to enzymatic exchange of label from 'H,O into n-xylose (12). Addition of excess Mg*+, an effective activator of the Streptomyces enzyme (2), to the solution containing xylose isomerase, n-xylose, and Mn*+ ( Fig. 1    a-C, resonance, indicating a decrease in the relaxation rates of this proton. In separate experiments, the longitudinal (l/T,) and transverse (I/T,) relaxation rates of the C 1 proton of n-xylose have been examined by pulsed Fourier transform methods (9) in the presence of Mn*+ and both in the presence and absence of enzyme. The results indicate that the enzyme-bound Mn2+ is more effective than free Mna+ in relaxing the a-C, proton of n-xylose. These studies were then extended using pulsed methods (Fig. 2) to a wide range of concentrations of xylose isomerase (Table I). Previous EPR studies (5) revealed tight binding of Mna+ to the Streptomyces enzyme at 3.0 + 1.0 sites characterized by K, values (27 * 10 PM) which agreed with those obtained kinetically, as well as additional weak binding sites. By limiting our experimental conditions such that the total concentration of Mn*+ was 50.5 times the enzyme concentration we could be certain that the concentration of Mn'+ bound at ancillary sites (5) or free in solution would be negligible.
Under these conditions the paramagnetic effect of enzyme-bound MnZ+ on the longitudinal relaxation rate of the C, proton of a-o-xylose is given by: where l/T,* and l/T," are the relaxation rates of the paramag-  Table  I. From these data at varying enzyme concentrations a value of fT,, for the ternary xylose isomerase .Mn2+ .~u-Gxylose complex is obtained by extrapolation (Fig. 3) to infinite enzyme concentration using the procedure of Mildvan and Cohn (19). Since the concentration of free MnZ+ was negligible in these experiments, a linear extrapolation of the data is justified (19). At infinite enzyme concentration the ternary complex will be the only species contributing measurably to fT,, which under these conditions is equal to 4.45 x lo-* s. The lack of a strong dependence of fT,, on enzyme concentration (Fig. 3) is consistent with the high affinity of the enzyme and the low affinity of n-xylose for Mn*+ (5). The more limited data available on the enzyme from Lactobacillus breuis (Table II) indicate comparable paramagnetic effects on l/T, and a greater effect on l/T, of the bound substrate.
Displacement of D-Xyl OSe from Its Ternary Complex by Xylitol-As shown in Fig. 1, the addition of excess xylitol to the n-xylose isomerase . MnZ+ .n-xylose complex causes a narrowing of the (Y-C, resonance, and a decrease in the radio and l/T,,, respectively. The paramagnetic contribution to the longitudinal relaxation rate (l/T,,) determined by pulsed methods also decreases upon addition of xylitol (Table I). These observations indicate that the competitive inhibitor has displaced the substrate from the ternary complex.
A titration of the effect of xylitol on the line width of the (Y-C, proton resonance (Fig. 4) may be fitted by assuming simple competition between D-xylose (dissociation constant for the ternary complex, K, = 8 mM (5)) and xylitol (KS = 0.51 mM). The K, value for xylitol reported here is in reasonable agreement with the kinetically determined inhibitor constant (K, = 1.5 mM, determined in maleate buffer) and with the K, value obtained by PRR titrations (0.59 mM, determined in Mes buffer) (5). The K, value for cr-D-xylose used to fit the titration data was equal to that obtained by PRR titrations and is in reasonable agreement with the K, of the substrate (10 mM) (5). Thus, the dissociation constants of the ternary complexes of substrate or inhibitor determined by NMR agree with those found by kinetics and are consistent with both compounds binding to the active site of n-xylose isomerase. At saturating levels of xylitol (Figs. 1 and 4) at least 83% of the paramagnetic effects of the enzyme-bound MnZ+ on the relaxation rates of a-D-xylose are removed, indicating negligible outer sphere contributions to the relaxation rates (20).

Effect of D-&dose
Isomerase .Mnz+ on Relaxation Rates of Protons of Xylitol-The proton NMR spectrum of xylitol at 100 MHz includes a multiplet structure 4.1 ppm downfield from tetramethylsilane which can be assigned to the terminal (C, and C,) methylene protons. This assignment is based upon the chemical shift, the integrated area, and simplification of this signal using the shift reagent EuCl, or by deuteration at C, by the reduction of xylulose with NaBD,.
The longitudinal (l/T,) and transverse (l/T,) relaxation rates of the protons of xylitol have been examined by pulsed Fourier transform methods (9) in the presence of Mn2+ and both in the presence and absence of enzyme (Tables III and  IV). The results indicate that the enzyme-bound MnZ+ is more effective than free MnZ+ in relaxing the methylene protons of xylitol. These experiments were then extended to a wide range of concentrations of enzyme and of MnZ+ (Table III) The lack of a strong dependence of fT,, on enzyme concentration, analogous to the results in the presence of cu-D-xylose (Fig. 3), is consistent a Conditions and sample preparation as described in Table I  with the high affinity of the enzyme for Mn*+ and xylitol (5) and a low affinity of xylitol for Mn2+.
l/fT,, = q/U',H + rH) + UT..,. ( where 9 is the stoichiometry of bound ligand to the bound MnZ+, 71 is the residence time of the coordinated ligand, T,, and T,, are the relaxation times of the coordinated ligands, f = [paramagnetic speciesy [ligand], and l/T,.,. is the small outer sphere contribution to the relaxation rates. In the present case l/T,., is negligible as determined by the effect of the enzyme. manganese. xylitol complex on the relaxation rates of D-xylose (Fig. 4). The observed frequency dependence of l/T,p (Tables I  and IV) and the finding that l/fT,, is significantly less than l/fT1, (Tables I and III) indicate that r&T,,, i.e. that values of the relaxation rates are not limited by ligand exchange (8,22). Hence l/fT,, is well approximated by q/T,,. Since neither The experimental values of l/T,, at 100 and 220 MHz a Value at 100 MHz. (Tables I and IV) r distance is 10.4 * 0.7 A. The assumption of complete occuwhere r is the ion-proton internuclear distance and C is a panty by cr-n-xylose must be qualified, however. The isomerproduct of constants proportional to the spin state and average ase-catalyzed reaction is at equilibrium in the NMR experig value of the metal ion. The correlation function, f(r,), is ments, and the product n-xylulose will compete for binding to defined in Equation  5. the enzyme. The close agreement observed for the values for the K, of n-xylose (10 mM) and the dissociation constant for an f(r,) = 3rc + 77c (5) equilibrium mixture of ternary complexes (8.1 mM) (5) argues 1 + lJI,*r,* 1 + WJ*ZEz that n-xylose and n-xylulose bind with similar affinities. The correlation function includes terms for the Larmor preces-Furthermore, the K, values for n-glucose and n-fructose sion frequencies for nuclear (w,) and electron (wJ spins and the determined under other conditions are comparable.' correlation time for dipolar interaction (7,). For Mn*+ and its The anomeric specificity of the enzyme must also be complexes, ws * T=* >> 1 > 77, while w,* rcZ I 1. Hence the considered.
One conclusion from kinetic studies (4) was that second term or the right side of Equation 5 is negligible (7). any interaction of /3-n-xylose with xylose isomerase must be From Equations 4 and 5 it can be seen that the experimental very weak compared to that of the a anomer. On the other values of l/T,, at the two frequencies provide the ratio f(s,) hand, cu-methyl-n-xyloside and B-methyl-n-xyloside were lOO/f(T,) 220 and the correlation time T, can be calculated.
found by equilibrium measurements to bind to the The value of r. 5.12 x lo-i0 s for the enzyme .Mn*+ .(Y-nenzyme 'Mn *+ complex with equally low affinities (5), which xylose complex, which was calculated assuming that rc is suggests that the P-n-xylose may bind in a nonproductive independent of frequency between 100 and 220 MHz is in close manner. Kinetic investigations (4) showed that the enzymatiagreement with values reported for the paramagnetic effects of tally formed anomer of the product n-xylulose predominates Mn*+ on proton relaxation rates in enolase (9) and pyruvate (81%) at equilibrium, although this anomer has not been kinase (22) complexes. Furthermore, this value of T, is characidentified. teristic of values reported for 7, (the longitudinal electron spin Based on these considerations we can make corrections of the relaxation time) of Mn*+ (25). This is consistent with r, being calculated distances. A minimal correction is required if only the dominant term in TV, as reported for several enzyme .Mn*+ Lu-n-xylose and the enzymatically active anomer of n-xylulose complexes (9,26,27). Alternatively, if a maximal dependence compete with equal affinity for binding. Of the total concentraof T, on frequency between 100 and 220 MHz is assumed (28), a tion of sugars present at equilibrium, 84% will be n-xylose (29), value of rF equal to 2.11 x lo-"' s at 100 MHz is obtained. and the enzymatically active form constitutes 33% of this (18).

Analysis of Equations 4 and 5 shows that this range of values of
The enzymatically active anomer of n-xylulose will be only 13% to simplify the NMR spectrum suggest that the line-broadening and radio frequency powers required to saturate the C,, C,, and C, protons agree within experimental error indicating identical values of l/fT,, for these 3 protons. Since q and 7c are the same for these protons the enzyme-bound Mn2+ ion must be equidistant from these 3 protons. Hence no portion of the cY-n-xylose substrate is coordinated directly to the enzyme-bound Mn2+. In the case of xylitol the average distance from the enzymebound manganese to terminal methylene protons in the xylose isomerase.Mn*+ .xylitol complex may be calculated in a similar manner (Table V). Assuming q = 1 and complete occupancy of enzymatic sites by the inhibitor (K, = 0.51 mM (5)), and employing values of f(s,) and l/fT,, summarized in Table   IV, the calculated manganese-terminal protons distance is 9.4 i 0.7 A (Table V). Since only a single isomer of xylitol, an analog of the open chain intermediate, is present in solution no correction for occupancy is necessary. The calculated distance (9.4 + 0.7 A) overlaps with the corrected distance to the protons of a-n-xylose.
The role of the MnZ+ ion in activating xylose isomerase thus appears to be structural rather than catalytic. Danno had reached similar conclusions from kinetic studies of the enzyme from Bacillus coagulans (30).
The range of calculated distances suggests that an enzyme. Mn2+ .X .Y .a-n-xylose complex is formed in which two small ligands such as water molecules, or a portion of the protein intervenes between the bound metal ion and the bound substrate. The previously reported decrease in the relaxation rate of water protons when cw-n-xylose or its analogs bind to the enzyme. Mn2+ complex (5) may reflect a decrease in z. due to a conformational change in the protein, or a decrease in q, the number of rapidly exchanging water ligands on Mn*+ due to occlusion of the active site. From the value of zc determined here, the previously determined value of l/fT,, of water protons (l/fT,, = 1.12 x lo8 s-l) (5) and from the Mn*+ to water proton distance of 2.87 + 0.05 A from x-ray data (26), it is estimated that 1.4 + 0.3 fast exchanging water ligands remain coordinated to the enzyme-bound Mn2+ in the ternary Lu-n-xylose complex.
Hence the results of the present and previous study (5) together suggest that the intervening ligands X and Y may be water molecules.
With the enzyme from L. breuis the comparable paramagnetic effect of the enzyme-bound Mn2+ on l/T, of the C-l proton of cY-n-xylose (Table II) suggests a similar structure of its ternary complex. Kinetic and Thermodynamic Properties of Ternary Complexes of Xylose Isomerase, Mnz+ and cu-D-Xylose-The kinetic and thermodynamic properties of the substrate complexes detected by NMR are summarized in Table VI. As discussed below, these observations are consistent with the participation of these complexes in catalysis. First, from Equations 2 and 3 it can be seen that the fastest relaxation rate sets a lower limit on the ligand exchange rate l/7,,,: l/fT,, 5 UfT,, 5 IITM (6) The value of l/fT,, sets a lower limit on l/~~ or korr, the rate constant for dissociation of the ternary complex in the following kinetic scheme:  "From Ref. 5. b From the ratio k&K,. c In 16 mM Tris-maleate buffer as described under "Methods." d In 50 mM Mes buffer as previously described (5). e Based on the kinetic data of Yamanaka (3) at 35" and a molecular weight of 43,000 per active site as found for the Streptomyces enzyme (13,14).
For both the Streptomyces and L. breuis enzymes ( Table VI) the values of k,,, exceed the maximal turnover numbers by at least 3 orders of magnitude, consistent with Michaelis-Menten kinetics. Furthermore, the dissociation constants (K,) of the ternary xylose complexes of both enzymes (Fig. 4, Table VI) agree with kinetically determined Michaelis constants (2,5). Finally, from k,,JK,, the lower limit values of k,, for both enzymes can be calculated and are given in Table VI. With the enzyme from L. breuis, the greater values of l/fT,, (Table II) may reflect a greater exchange rate of cu-n-xylose into its ternary complex (Table VI). The magnetic resonance experiments have thus detected a ternary substrate complex which is kinetically and thermodynamically competent to participate in the xylose isomerase reaction.