Two tyrosine residues, Tyr-108 and Tyr-503, are responsible for the deprotonation of phenolic substrates in vanillyl-alcohol oxidase

A number of oxidoreductases from the VAO/para-cresol methylhydroxylase flavoprotein family catalyze the oxidation of para-substituted phenols. One of the best-studied is vanillyl-alcohol oxidase (VAO) from the fungus Penicillium simplicissimum. For oxidation of phenols by VAO to occur, they must first be bound in the active site of the enzyme in their phenolate anion form. The crystal structure of VAO reveals that two tyrosine residues, Tyr-108 and Tyr-503, are positioned to facilitate this deprotonation. To investigate their role in catalysis, we created three VAO variants, Y108F, Y503F, and Y108F/Y503F, and studied their biochemical properties. Steady-state kinetics indicated that the presence of at least one of the tyrosine residues is essential for efficient catalysis by VAO. Stopped-flow kinetics revealed that the reduction of VAO by chavicol or vanillyl alcohol occurs at two different rates: kobs1, which corresponds to its reaction with the deprotonated form of the substrate, and kobs2, which corresponds to its reaction with the protonated form of the substrate. In Y108F, Y503F, and Y108F/Y503F, the relative contribution of kobs2 to the reduction is larger than in wild-type VAO, suggesting deprotonation is impaired in these variants. Binding studies disclosed that the competitive inhibitor isoeugenol is predominantly in its deprotonated form when bound to wild-type VAO, but predominantly in its protonated form when bound to the variants. These results indicate that Tyr-108 and Tyr-503 are responsible for the activation of substrates in VAO, providing new insights into the catalytic mechanism of VAO and related enzymes that oxidize para-substituted phenols.

A number of oxidoreductases from the VAO/para-cresol methylhydroxylase flavoprotein family catalyze the oxidation of para-substituted phenols. One of the best-studied is vanillylalcohol oxidase (VAO) from the fungus Penicillium simplicissimum. For oxidation of phenols by VAO to occur, they must first be bound in the active site of the enzyme in their phenolate anion form. The crystal structure of VAO reveals that two tyrosine residues, Tyr-108 and Tyr-503, are positioned to facilitate this deprotonation. To investigate their role in catalysis, we created three VAO variants, Y108F, Y503F, and Y108F/Y503F, and studied their biochemical properties. Steady-state kinetics indicated that the presence of at least one of the tyrosine residues is essential for efficient catalysis by VAO. Stopped-flow kinetics revealed that the reduction of VAO by chavicol or vanillyl alcohol occurs at two different rates: k obs1 , which corresponds to its reaction with the deprotonated form of the substrate, and k obs2 , which corresponds to its reaction with the protonated form of the substrate. In Y108F, Y503F, and Y108F/Y503F, the relative contribution of k obs2 to the reduction is larger than in wild-type VAO, suggesting deprotonation is impaired in these variants. Binding studies disclosed that the competitive inhibitor isoeugenol is predominantly in its deprotonated form when bound to wild-type VAO, but predominantly in its protonated form when bound to the variants. These results indicate that Tyr-108 and Tyr-503 are responsible for the activation of substrates in VAO, providing new insights into the catalytic mechanism of VAO and related enzymes that oxidize para-substituted phenols.
Vanillyl-alcohol oxidase (VAO) 2 (EC 1.1.3.38) from Penicillium simplicissimum is a flavin-dependent oxidase that cata-lyzes the oxidation of para-substituted phenolic compounds using molecular oxygen as an electron acceptor (1)(2)(3). It is a 64-kDa (560 amino acid) protein that forms octameric structures with each subunit containing an FAD cofactor that is covalently linked to the protein via an 8␣-N3-histidyl-FAD bond (4). VAO is a member of the VAO/para-cresol methylhydroxylase (PCMH) flavoprotein family, which consists of homologous oxidoreductases that catalyze a wide range of reactions, but share a conserved FAD-binding domain (5). Although VAO catalyzes a range of reactions, such as the oxidation of benzylic alcohols to aldehydes, the oxidative deamination of benzylic amines, the hydroxylation of 4-allylphenols, and the hydroxylation or dehydrogenation of 4-alkylphenols, it displays a strict selectivity toward the oxidation of para-substituted phenols at the C␣ atom of their side chain. The proposed catalytic mechanism of VAO provides an explanation for this selectivity (Fig. 1A) (2,6,7). First, the substrates of VAO are bound in the active site in their deprotonated phenolate form. Subsequently, a hydride anion is transferred from the C␣ atom of the side chain of the substrate to the N5 atom of FAD, yielding the reduced enzyme and a para-quinone methide intermediate. Finally, this intermediate either reacts with water or rearranges to yield the product of the reaction, and FAD is reoxidized by molecular oxygen, which is converted to hydrogen peroxide. The initial deprotonation of the phenolic hydroxyl group of the substrate is likely a crucial step in this mechanism, as it promotes the hydride-transfer reaction by enabling the formation of the para-quinone methide intermediate. Studies using the competitive inhibitor isoeugenol (Fig.  1B) revealed that its pK a is lowered from 10 to 5 upon binding to VAO (2). A potential explanation for this decrease is provided by examining the crystal structure of VAO in complex with isoeugenol (4). Three polar residues, Tyr-108, Tyr-503, and Arg-504, form a phenolate-binding pocket, with their side chains positioned within hydrogen bonding distance of phenolic hydroxyl group of the isoeugenol (Fig. 2A). These residues may facilitate deprotonation of the substrate either by promoting preferential binding of the deprotonated form or by one of the tyrosine residues acting as a base and actively deprotonating the substrate in the active site. All three residues are conserved in the homologous flavoenzymes eugenol oxidase (EUGO) and para-cresol methylhydroxylase (PCMH, EC 1.17.99.1), which also catalyze the oxidation of para-substituted phenols (8,9). The role of Arg-474 in PCMH, which corresponds to Arg-504 in VAO, has been studied by site-directed mutagenesis and it was found to be of importance in achieving covalent incorporation of the cofactor of the enzyme and in enabling efficient catalysis by modulating the redox properties of both the cofactor and enzyme-bound substrates (10). Here, we investigated the role played by Tyr-108 and Tyr-503 in VAO using sitedirected mutagenesis. Three VAO variants, Y108F, Y503F, and Y108F/Y503F, were created, expressed, and purified and their catalytic properties were studied. Our results reveal that the presence of at least one of these tyrosine residues is essential for efficient catalysis.

Results
The Y108F, Y503F, and Y108F/Y503F variants were successfully expressed in Escherichia coli and purified. Like wild-type (wt)-VAO, all three variants contained a covalently bound FAD cofactor, as judged from the yellow pellet and colorless supernatant obtained after precipitation of the proteins with trichloroacetic acid. In their native form, all variants displayed an octameric quaternary structure, as demonstrated by analytical size exclusion chromatography (data not shown). The flavin absorption spectra of the variants were highly similar to that of wt-VAO (see spectral studies presented below), demonstrating that the introduced mutations do not lead to any major changes in the electronic environment of their flavin cofactor. To determine whether the introduced mutations affect the stability of VAO, thermal midpoints of unfolding of the variants were determined using the ThermoFAD method (11). This yielded thermal midpoints of unfolding of 55°C for wt-VAO, 53°C for Y108F and Y503F, and 52°C for Y108F/Y503F, suggesting the introduced mutations have a minor deleterious effect on the thermostability of VAO.

Steady-state catalytic properties
To determine the effect of the introduced mutations on the catalytic efficiency of VAO, we measured the apparent steady-   2VAO (4))) and in the Y108F and Y503F variants (B and C, respectively, carbon in gray). The refined 2F o Ϫ F c electron density map (contoured at 1.2) is shown for the mutated residues Y108F and Y503F as blue chicken-wire. The wt-VAO crystal structure in complex with the competitive inhibitor isoeugenol (A, carbon in green) is taken as a reference to analyze the mutant proteins. As reported in Ref. 4, binding of isoeugenol does not significantly influence the geometry of the active site. In B and C, the side chains of Tyr-108 and Tyr-503, respectively, from wt-VAO are shown for comparison. As described in the text, in the Y503F mutant Phe-503 is slightly tilted with respect to the corresponding Tyr-503 in wt-VAO. Tyr-108, Tyr-503, Arg-504, Ser-426, isoeugenol, and the FAD cofactor (carbon in yellow) are displayed in stick representation with the surrounding protein backbone shown in schematic representation. Oxygen atoms are in red, nitrogen atoms in blue, and phosphorous atoms in magenta. Potential hydrogen bonds are depicted as black dashed lines.

Substrate activation in VAO
state kinetic parameters for the oxidation of two substrates: vanillyl alcohol, which is oxidized to the aldehyde vanillin, and chavicol ( Fig. 1B), which is hydroxylated to coumaryl alcohol. The kinetic parameters for the oxidation of these substrates are listed in Table 1. Interestingly, the effect of the mutations varied depending on the substrate. With vanillyl alcohol, the k cat drops to ϳ2% of that of wt-VAO for both the Y108F and Y503F variants. With chavicol, a more minor drop in the k cat value to ϳ45% of that of the wild-type enzyme was observed for the Y108F and Y503F variants. With chavicol, the Y108F/Y503F variant displayed a k cat value of less than 2% of that of wt-VAO and with vanillyl alcohol, the activity was too low to accurately determine kinetic parameters. This indicates that the presence of at least one of the two tyrosine residues is essential for effective catalysis. For each variant, the K m increased for each substrate. The magnitude of the increase ranged from 3-to 12-fold.

Stopped-flow kinetics of the reductive half-reaction
To determine whether the decreased catalytic efficiencies observed for the VAO variants were due to impaired reduction of their FAD cofactor, spectral changes during the anaerobic reduction of the enzymes by vanillyl alcohol or chavicol were followed by stopped-flow spectroscopy. Where possible, photodiode array detection was used to monitor changes over a spectral range from 254 to 725 nm. However, with long measurement times (Ͼ10 s), the high-intensity illumination required for this led to photoreduction of FAD. Therefore, in cases where the reduction reaction was slow, it was monitored using single wavelength detection at 439 nm, the absorption maximum of the oxidized FAD in VAO. This was the case for the reduction of the Y108F and Y503F variants by 2 mM vanillyl alcohol and for all reactions involving the Y108F/Y503F variant.
Observed rates for the anaerobic reduction of wt-VAO and the Y108F and Y503F variants by 2 mM vanillyl alcohol at pH 6.2 and 7.5 are given in Table 2. The anaerobic reduction of wt-VAO by vanillyl alcohol at pH 7.5 has been studied previously and was found to be a biphasic process with rate constants at saturating substrate concentrations of k red1 ϭ 24 s Ϫ1 and k red2 ϭ 3.5 s Ϫ1 (6). The two different rates of reduction were interpreted to be caused by the presence of two conformations of the enzyme. The rate of the slower phase, k red2 , lies close to k cat and is rate-limiting for catalysis. Our results with wt-VAO are in good agreement with these data. The effect of the Y108F and Y503F mutations on the rate of reduction by vanillyl alcohol was similar. In each variant, the reduction was a biphasic process. Each of the rates of reduction has dropped significantly as compared with wt-VAO. The rates of the two phases are in the same range and k obs2 is similar to the rate of steady-state catalysis, suggesting that flavin reduction is rate-limiting in these variants. In addition to the decrease in both reduction rates, the relative contribution of the two kinetic events to the drop in absorption at 439 nm (amplitude a) has changed. In wt-VAO, k obs1 has a greater contribution than k obs2 (a 1 :a 2 Ϸ 2:1), whereas in the Y108F and Y503F variants, k obs2 has a greater contribution than k obs1 (a 1 :a 2 Ϸ 1:5 for Y108F and 1:3 for Y503F). At pH 6.2, reduction by vanillyl alcohol was also biphasic for all variants. For wt-VAO, the rates of the two phases are very similar to those at pH 7.5. However, their contributions to the reduction have changed, with k obs2 having a greater contribution than k obs1 (a 1 :a 2 Ϸ 1:3). In the Y108F and Y503F variants, k obs1 is somewhat higher than observed at pH 7.5 and k obs2 is slightly lower in Y108F and does not change significantly in Y503F. As observed for wt-VAO, the contribution of the two rates to the change in absorption shifts, with k obs2 becoming more predominant (a 1 :a 2 Ϸ 1:32 for each variant).
Spectra obtained during the anaerobic reduction of wt-VAO by 100 M chavicol are shown in Fig. 3. After mixing wt-VAO with chavicol, the characteristic peak of the oxidized flavin cofactor of the enzyme (maximum at 439 nm) rapidly disappears, indicating that FAD is reduced by the substrate. At the same time, the absorption between 320 and 430 nm increases, yielding an intense peak with a maximum at 373 nm. This peak reaches its maximum intensity after ϳ600 ms, after which it decreases in intensity. The formation of a species with similar spectral characteristics has been observed during the anaerobic reduction of wt-VAO by vanillyl alcohol ( max ϭ 362 nm) or 4-(methoxymethyl)phenol ( max ϭ 364 nm) and was attributed to a complex between the quinone methide intermediate and the reduced enzyme formed after the electron-transfer reaction

Table 2 Observed rates of reduction of VAO variants by 2 mM vanillyl alcohol
Rates were determined by fitting appropriate exponential functions to single wavelength traces of the absorption at 439 nm obtained after mixing 10 M enzyme with 2 mM vanillyl alcohol in 50 mM potassium phosphate buffer, pH 6.2 or 7.5, under anaerobic conditions at 25°C using a stopped-flow apparatus. Values in parentheses give the relative contribution of k obs1 and k obs2, the two phases associated with flavin reduction, to the drop in absorption. The observed rates for the oxidation of 2 mM vanillyl alcohol under steady-state conditions (k ss ) at pH 7.5 are given for comparison.

VAO variant
a ND, not determined due to low reaction rate.

Substrate activation in VAO
has taken place (6). Intermediate spectra obtained upon spectral deconvolution of the stopped-flow data are shown in Fig. 3. The data were fit to a three-phase consecutive model, yielding four distinct spectral intermediates. Species 1 represents the oxidized enzyme. Species 3 displays the highest absorption at 373 nm and complete loss of the characteristic oxidized flavin spectrum. It likely represents the quinone methide-reduced enzyme complex. Species 2 displays incomplete formation of the quinone methide intermediate as indicated by its lower absorption at 373 nm. In addition, part of the FAD cofactor is still in its oxidized form in this species, as evidenced by the absorption in the range 440 -500 nm. The spectra of species 1, 2, and 3 display an isosbestic point at 423 nm and a spectrum calculated by assuming that the spectrum of species 2 is a simple composite of those of species 1 (20%) and 3 (80%) was virtually indistinguishable from the actual spectrum of species 2. This is consistent with species 2 not representing a real reaction intermediate, but rather reflecting the fact that there are two separate rates of conversion of species 1 to species 3. Species 4 displays a spectrum that is typical of a reduced flavoprotein, suggesting that this species is formed by the breakdown of the quinone methide intermediate by its reaction with water, yielding the product coumaryl alcohol. Thus, the anaerobic reduction of wt-VAO by chavicol proceeds as follows. First, the sub-strate binds to the enzyme. Subsequently, the FAD cofactor is reduced by the substrate leading to the formation of a quinone methide-reduced enzyme complex with two kinetic phases that both contribute to this reduction being distinguished. Following reduction, the quinone methide intermediate reacts with water to yield coumaryl alcohol. Rate constants for these processes were obtained by fitting curves of the absorption at 440 nm to a triphasic exponential function. Rate constants for the two reduction rates (k obs1 and k obs2 ) are given in Table 3. k obs2 is similar to the steady-state reaction rate, suggesting it is ratelimiting for catalysis. The observed rate of decay of the quinone methide intermediate, k obs3 , was similar for all variants (0.1-0.2 s Ϫ1 ) and is too slow to be catalytically relevant. Presumably, breakdown of this intermediate is faster in the presence of oxygen, as has been observed previously for the quinone methide intermediate formed upon reduction of wt-VAO by 4-(methoxymethyl)phenol (6). Upon reduction of the Y108F and Y503F variants by chavicol, the spectral changes observed were similar to those seen with wt-VAO, with reduction of FAD occurring concomitantly with the appearance of a putative quinone methide intermediate (Fig. 3). The absorption maximum of this intermediate is shifted to lower wavelengths in the variants (363 nm for Y108F and 368 nm for Y503F) compared with wt-VAO (373 nm). Bottom panels show spectra of reaction intermediates obtained after spectral deconvolution of the stopped-flow data for wt-VAO, Y108F, and Y503F. Data were fit to a three-phase model (1323334), yielding four spectrally distinct species: 1 (O), 2 (---), 3 (-⅐-⅐-), and 4 (⅐⅐⅐). One-or two-phase models were not able to satisfactorily explain the data. The rates obtained for the three phases (average of at least 3 measurements) are: k 1 ϭ 134 s Ϫ1 , k 2 ϭ 3.7 s Ϫ1 , and k 3 ϭ 0.093 s Ϫ1 for wt-VAO, k 1 ϭ 29 s Ϫ1 , k 2 ϭ 1.7 s Ϫ1 , and k 3 ϭ 0.051 s Ϫ1 for Y108F, and k 1 ϭ 13 s Ϫ1 , k 2 ϭ 2.7 s Ϫ1 , and k 3 ϭ 0.083 s Ϫ1 for Y503F. These rates are in good agreement with those obtained by fitting to single wavelength traces at 440 nm (see Table 3).

Substrate activation in VAO
Maximum abundance of the intermediate was observed after 2.2 s for the Y108F variant and after 1.5 s for the Y503F variant, after which the intermediate decays. Spectral deconvolution again revealed a three-phase process (Fig. 3). As with wt-VAO, species 1 and 3 represent the oxidized enzyme and the quinone methide-reduced enzyme complex, respectively. Species 2 again displays an absorption spectrum that is a composite of that of species 1 and 3 and is likely not a real reaction intermediate. Species 4 appears to represent a mixture of quinone methide-reduced enzyme complex and free reduced enzyme. This is likely due to the decay of the quinone methide not being complete at the end of the measurement (120 s). Fitting the data at 440 nm to triple exponential functions revealed that k obs1 and k obs2 are both decreased in each variant (Table 3). However, neither rate in either variant decreased more than 10-fold. Thus, the effect of the Y108F and Y503F mutations on the rate of reduction of VAO by chavicol is less severe than that on the reduction by vanillyl alcohol (k obs1 and k obs2 decreased Ͼ100fold in each variant). Interestingly, a similar effect on the relative contributions of k obs1 and k obs2 to reduction was observed with chavicol as with vanillyl alcohol, with k obs1 contributing most in wt-VAO, but k obs2 contributing most in the Y108F and Y503F variants. Reduction of the Y108F/Y503F variant by chavicol was monitored using single wavelength detection at 439 nm. In this case, the data fit best to a double exponential function. Interestingly, k obs1 is in the same range as for the Y108F and Y503F variants, whereas k obs2 is an order of magnitude smaller. This brings it into the same range as k obs3 observed for wt-VAO and the Y108F and Y503F variants. This provides an explanation for the absence of a third phase, as the decay of the quinone methide intermediate is now likely no longer distinguishable from the second flavin reduction phase. As with Y108F and Y503F, k obs2 makes the largest contribution to flavin reduction. For all variants, k obs2 lies close to the observed rate of steady-state catalysis, suggesting it is rate-limiting.
Upon reduction of wt-VAO and the Y108F and Y503F variants by 100 M chavicol at pH 6.2, similar spectral changes were observed as at pH 7.5. Formation of a quinone methide intermediate and concomitant reduction of FAD were followed by the decay of the quinone methide intermediate, which was too slow to be catalytically relevant. The observed reduction rates are given in Table 3. For wt-VAO, the reduction is a biphasic process. The two rates are similar to those observed at pH 7.5, however, as with vanillyl alcohol, the contribution of the rates to the drop in absorbance changes, with k obs2 becoming more predominant (a 1 :a 2 Ϸ 1:2 at pH 6.2 compared with a 1 :a 2 Ϸ 3:1 at pH 7.5). For the Y108F and Y503F variants, only a single reductive phase was observed, with rate constants that are similar to those of k obs2 at pH 7.5, suggesting that the observed rate can be attributed to the same process as k obs2 and that the process associated with k obs1 does not make an observable contribution to the reduction at pH 6.2. For the Y108F/Y503F variant, the reduction was observed to be biphasic with the rates of the two phases and their relative contributions being similar to those observed at pH 7.5. However, the contribution of the slower phase to the reduction may be underestimated. After reduction of Y108F/Y503F a residual absorbance of 0.03 was measured (usually ϳ0.015). With the Y108F and Y503F variants it was observed that after the reduction was complete the absorption at 440 nm increased, likely due to the formation of a charge-transfer interaction between the reduced enzyme and a bound substrate molecule. Due to the slow rate of reduction of the Y108F/Y503F variant, formation of such a charge-transfer complex could occur on a similar time scale as the reduction, which would cause part of the absorbance drop associated with the reduction to be masked.
To study the pH dependence of the reaction in more detail, the anaerobic reduction of wt-VAO by 50 M chavicol was studied in the pH range 6.1-9.5. The overall spectral changes, as monitored using photodiode array detection, were similar to those observed during the experiments presented in Fig. 3 at all pH values. Reduction of FAD and concomitant formation of a quinone methide-reduced enzyme complex was followed by the decay of the quinone methide intermediate to yield the fully reduced enzyme spectrum. As before, the absorption at 440 nm was fit to a triphasic exponential function, yielding two rates that are associated with flavin reduction: k obs1 and k obs2 . The values of k obs1 and k obs2 remained similar in this pH range, with the lowest and highest values never differing more than a factor 2 (Fig. 4, inset). The pH dependence of the fractional contribution of k obs1 to the reduction is shown in Fig. 4. The contribution of k obs1 increases with pH. Attempts to fit the data to the Henderson-Hasselbach equation, which allows the determination of the pK a of an ionizable group were unsuccessful. Although the data display a sigmoidal form, the slope of the sigmoid is lower than described by the Henderson-Hasselbach equation. To estimate the midpoint at which the contribution of k obs1 and k obs2 to reduction is equal, the data were fit to a sigmoidal curve with the slope as a variable, yielding a midpoint of pH 6.9.

Isoeugenol-binding studies
To evaluate whether the introduced mutations affect the protonation state of phenolic compounds bound to VAO, we studied the binding of the competitive inhibitor isoeugenol to Table 3 Observed rates of reduction of VAO variants by 100 M chavicol Rates were determined by fitting appropriate exponential functions to traces of the absorption at 440 nm (from photodiode array detection (wt-VAO, Y108F, and Y503F)) or 439 nm (from single wavelength detection (Y108F/Y503F)) obtained after mixing 10 M enzyme with 100 M chavicol in 50 mM potassium phosphate buffer, pH 6.2 or 7.5, under anaerobic conditions at 25°C using a stopped-flow apparatus. Values in parentheses give the relative contribution of k obs1 and k obs2 , the two phases associated with flavin reduction, to the drop in absorption. The observed rates for the oxidation of 100 M chavicol under steady-state conditions (k ss ) at pH 7.5 are given for comparison.

Substrate activation in VAO
the VAO variants. In solution, the phenolic hydroxyl group of isoeugenol has a pK a of ϳ10 and its absorption spectrum is dependent on its protonation state (2). In particular, deprotonated isoeugenol displays significantly higher absorption at 320 nm than protonated isoeugenol (⑀ 320 ϭ 0.21 mM Ϫ1 cm Ϫ1 at pH 6 and 8.8 mM Ϫ1 cm Ϫ1 at pH 12). Therefore, the absorption of isoeugenol at 320 nm upon binding to VAO can act as a reporter on its protonation state. The pK a of isoeugenol bound to wt-VAO is ϳ5 (2). To evaluate whether this has been perturbed by the introduced mutations, the VAO variants were titrated with isoeugenol at pH 7.5. At this pH, isoeugenol is almost entirely protonated in solution, but almost entirely deprotonated when bound to wt-VAO. Spectral changes observed upon titration of wt-VAO with isoeugenol are shown in Fig. 5. Upon addition of isoeugenol, the intensity of the absorption peak of oxidized FAD at 439 nm decreases, whereas the absorption at wavelengths below 335 nm increases, which is due to the absorption of the added isoeugenol. Plots of the absorption at 439 and 320 nm against the concentration of isoeugenol displayed a curvature indicative of a binding event (Fig. 5, inset). The data at each of these wavelengths could be satisfactorily fit to a model assuming a simple 1:1 complex formation between isoeugenol and VAO. This yielded a dissociation constant of 6.8 Ϯ 0.5 M from the data at 439 nm and 6.8 Ϯ 1.0 M from the data at 320 nm. The increase in the absorption at 320 nm associated with binding of isoeugenol to VAO is indicative of the fact that, as expected, it is primarily in its deprotonated form upon binding to the enzyme.
Somewhat different spectral changes were observed upon titration of the Y108F variant with isoeugenol (Fig. 5). As with wt-VAO, the intensity of the peak at 439 nm decreases and absorption below 335 nm increases. In addition, a broad absorption band appears at wavelengths between 500 and 680 nm. Such a broad absorption band at high wavelengths is indicative of a charge-transfer interaction between isoeugenol and FAD, likely caused by -stacking of the aromatic rings of isoeugenol and isoalloxazine. Similar charge-transfer bands have been observed upon binding of various aromatic compounds to the flavoprotein subunit of PCMH (12). Fitting of the data at 439 nm yielded a dissociation constant of 18 Ϯ 2 M, suggesting binding of isoeugenol is slightly impaired in the Y108F variant as compared with wt-VAO. The absorption at 550 nm also displayed curvature indicative of a single binding event and fitting yielded a dissociation constant of 20 Ϯ 1 M. This suggests that the drop in absorption at 439 nm and chargetransfer band at wavelengths above 500 nm are both caused by the same binding event, the formation of a 1:1 complex between isoeugenol and Y108F. The absorption at 320 nm, however, did not display the major increase associated with binding to wt-VAO. This suggests that isoeugenol is primarily in its protonated form when bound to Y108F and that the protonated form of isoeugenol forms a charge-transfer complex with the FAD cofactor of Y108F.
For Y503F, the absorption peak at 439 nm also decreased in intensity upon addition of isoeugenol (Fig. 5). Fitting to the data at 439 nm yielded a dissociation constant of 91 Ϯ 7 M, indicating that binding of isoeugenol is significantly weaker than for wt-VAO. Similar to Y108F, the appearance of a charge-transfer band at wavelengths above 500 nm was observed. However, the measured absorption of this band was too low to reliably determine the dissociation constant. The absorption at 320 nm essentially increased linearly with the isoeugenol concentration, suggesting it is due to the presence of deprotonated isoeugenol in solution and that isoeugenol is predominantly in its protonated form upon binding to Y503F.
With Y108F/Y503F, the results of the titration strongly resemble those with Y108F (Fig. 5). Dissociation constants of 11 Ϯ 1 and 7.0 Ϯ 0.4 M were obtained by fitting to the data at 439 and 550 nm, respectively. The appearance of a chargetransfer band is observed and the absorption at 320 nm does not show the large increase associated with binding to wt-VAO. This suggests that, like for the Y108F and Y503F variants, isoeugenol is predominantly in its protonated form when in complex with Y108F/Y503F.

Crystal structures
To evaluate the effects of the mutations on the architecture of the active site of VAO, the crystal structures of the Y108F and Y503F variants were solved at 2.8-Å resolution, by molecular replacement using the VAO monomer (PDB code 2VAO) (4) as a search model. Attempts to solve the structure of the Y108F/ Y503F variant were hampered by fragility of the crystals. The overall structures of both Y108F and Y503F are essentially identical to that of wt-VAO, as indicated by an r.m.s. deviation of 0.5 Å for 1105 and 1108 pairs of C␣ atoms, respectively. The asymmetric unit contains two monomers, which form the functional octamer through crystallographic symmetry, as in wt-VAO.
The geometry of the active site in the mutated enzymes is largely conserved, with the aromatic ring of Phe-108 or Phe-503 located at the same position as the corresponding tyrosine residue in wt-VAO, where they each form a hydrogen bond with the phenolic hydroxyl group of isoeugenol (Fig. 2) (4). Additionally, Tyr-108 interacts with Ser-426 via a hydrogen bond. This interaction is disrupted in the Y108F variant, causing a Anaerobic substrate and enzyme solutions in Britt-Robinson buffers of appropriate pH values were mixed in a stopped-flow apparatus and absorbance changes were monitored using a photodiode array detector. Fitting curves to the absorbance at 440 nm yielded two reduction rates: k obs1 (fast) and k obs2 (slow). The graph shows the fraction of the drop in absorbance associated with reduction that is contributed by k obs1 (fraction k obs1 ϭ a obs1 /(a obs1 ϩa obs2 ), where a is the amplitude). The solid line is a fit of a sigmoidal function to the data, from which a midpoint pH value (fraction k obs1 ϭ fraction k obs2 ϭ 0.5) of 6.9 was estimated. The inset shows the values of k obs1 (OE) and k obs2 (F). The error on each data point is less than 5%.

Substrate activation in VAO
shift of the Phe-108 side chain toward the FAD cofactor (Fig.  2B). In the Y503F variant, the plane of the aromatic ring of Phe-503 is slightly tilted compared with Tyr-503 in wt-VAO (Fig. 2C). This may partly be due to the lack of the hydrogen bond between Tyr-503 and Arg-504 that is present in wt-VAO.
In summary, both overall structures of the Y108F and Y503F variants and the geometry of their active sites are extremely similar to wt-VAO. Therefore, changes in reactivity are not due to major structural perturbations, but can be attributed to disruption of the hydrogen bonding capacity conferred by the two tyrosine residues.

Conservation of the phenolate-binding pocket in VAO homologues
To determine whether the residues that form the phenolatebinding pocket in VAO are also present in homologous proteins, we evaluated their conservation in a group of homologues of VAO, EUGO, PCMH, and the related flavocytochrome eugenol hydroxylase from Pseudomonas sp. OPS1 (13) that we identified in a previous study and named the 4-phenol oxidizing (4PO) subfamily (14). Additionally, we determined whether they are present in the decarboxylases CndG and FeeG, which catalyze the oxidative decarboxylation of N-substituted tyrosine derivatives (15,16). These enzymes display homology to 4PO subfamily members and catalysis is believed to proceed via a mechanism whereby the initial formation of a para-quinone methide intermediate is followed by the expulsion of carbon dioxide. This analysis revealed that Tyr-108, Tyr-503, and Arg-504 are strictly conserved suggesting they are of importance for catalysis in these related enzymes.

Discussion
Our results demonstrate that the presence of at least one of the active site tyrosine residues, Tyr-108 or Tyr-503, is essential for efficient catalysis by VAO, with the k cat values for Y108F/ Y503F being less than 2% of those for wt-VAO for each tested substrate. The effect of the single tyrosine to phenylalanine mutations on catalytic efficiency was dependent on the substrate. The oxidation of vanillyl alcohol was severely impaired in both the Y108F and Y503F variants, whereas the oxidation of chavicol was only slightly impaired as compared with wt-VAO.
Stopped-flow kinetics revealed that reduction of the FAD cofactor of VAO by vanillyl alcohol or chavicol proceeds at two distinct rates: k obs1 (fast) and k obs2 (slow). k obs2 lies close to the rate of steady-state catalysis and, therefore, is probably ratelimiting. The observed reduction in the rate of steady-state catalysis for the tyrosine to phenylalanine variants corresponds with the observed reduction in k obs2 , demonstrating that impaired flavin reduction is the cause of the reduced rate of catalysis. Titration with the competitive inhibitor isoeugenol indicated that, at pH 7.5, this molecule is predominantly in its deprotonated phenolate anion form when bound to wt-VAO, but predominantly in its protonated form when bound to Y108F, Y503F, or Y108F/Y503F. In combination, these results

Substrate activation in VAO
suggest that Tyr-108 and Tyr-503 play a role in activating the substrates of VAO for their reaction with its FAD cofactor by enabling their deprotonation in the active site of the enzyme. Interestingly, the relative contributions of the two reduction rates, k obs1 and k obs2 , to the change in absorption associated with flavin reduction changes upon the introduction of the tyrosine to phenylalanine mutations. At pH 7.5, k obs1 contributes more to the drop in absorption in wt-VAO, whereas in the Y108F, Y503F, and Y108F/Y503F variants, k obs2 contributes more to the drop in absorption. The relative contributions of the two rates shift when the reduction is performed at pH 6.2, with the relative contribution of k obs2 becoming greater. The reduction of the Y108F/Y503F variant by chavicol was an exception to this observation. However, this may be due to the contribution of k obs2 to the reduction at pH 6.2 being underestimated due to the concomitant formation of a charge-transfer interaction between the reduced enzyme and substrate. For the reduction of wt-VAO by chavicol, the contribution of k obs1 increases with pH in the pH range 6.1-9.5. The data were fit to a sigmoidal curve with a midpoint pH, at which k obs1 and k obs2 contribute equally to the reduction, of 6.9.
Based on these results, we propose a somewhat modified model for the reaction mechanism of VAO ( Fig. 6). Substrates can bind in the active site of VAO with their phenolic hydroxyl group either in its protonated or deprotonated form. The affinity of the enzyme for the substrate depends on the protonation state of the substrate, with wt-VAO displaying a higher affinity for deprotonated substrates. Subsequently, the substrate reacts with VAO, a reaction that occurs at different rates depending on the protonation state of the substrate. With the deprotonated substrate, the reaction simply involves the transfer of a hydride anion to the FAD cofactor to yield the para-quinone methide intermediate. Reaction with the protonated substrate requires the deprotonation of the phenolic hydroxyl group of the substrate either prior to or at the same time as the hydridetransfer reaction. The proton is transferred to either Tyr-108 or Tyr-503, which may be transiently deprotonated in the active site of the enzyme. This is stimulated by Arg-504, which is positioned so as to form a cationinteraction with Tyr-108, lowering its pK a . This model explains the two rates of flavin reduction observed for each substrate, with k obs1 representing the reaction with deprotonated substrate and k obs2 representing the reaction with protonated substrate. This is supported by the fact that the contribution of k obs1 to flavin reduction increases with pH, as at higher pH values a larger percentage of the substrate is in its deprotonated form. The results of the isoeugenol titrations support the idea that substrates bind to Y108F, Y503F, and Y108F/Y503F predominantly in their protonated form at pH 7.5, as opposed to wt-VAO where it predominantly binds in its deprotonated form. This explains the change in the relative contributions of k obs1 and k obs2 to flavin reduction upon introduction of the tyrosine to phenylalanine mutations. The mutations prevent the preferential binding of the deprotonated form of the substrates observed with wt-VAO due to the loss of hydrogen-bonding interactions that stabilize the negative charge on the phenolate oxygen. In support of this, the crystallographic analysis revealed that, although the mutations do not produce any major structural perturbation of the active site architecture, they modify the hydrogen-bonding network that is essential to activate the substrate for the reaction with the flavin cofactor. The fact that the curve of the fractional contribution of k obs1 to the reduction versus pH does not obey the Henderson-Hasselbach equation may be attributable to the effects of functional groups near the phenolic hydroxyl group on its protonation state, which can have a distorting effect on the titration behavior of ionizable groups in proteins (17). Additionally, interconversion of the enzyme-protonated substrate (E ox -SH) and enzyme-deprotonated substrate (E ox -S Ϫ ) complexes may involve more significant changes than simply the deprotonation of the phenolic hydroxyl group of the substrate. The fact that a charge-transfer interaction is observed between the enzyme and protonated isoeugenol during titrations with this molecule, but not between the enzyme and deprotonated isoeugenol, suggest that the ligand may bind to the enzyme in a slightly different conformation depending on its protonation state. Thus conversion of the E ox -SH complex to the E ox -S Ϫ complex may require a reorientation of the substrate within the active site, and possibly minor changes in the conformation of active site residues. This could explain the reasonably low rates observed for k obs2 . Although in our view this model provides the most satisfactory explanation of our observations, an alternative mechanism where the two observed reduction rates reflect the presence of two forms of the enzyme, one in which both Tyr-108 and Tyr-503 are protonated and one in which one of the two tyrosine residues is deprotonated, cannot be fully ruled out.
The effect of the single Y108F and Y503F mutations on the rate of reduction of VAO was dependent on the substrate used. With the benzylic alcohol vanillyl alcohol, a far more severe drop in both k obs1 and k obs2 was observed than with the 4-allyphenol chavicol. However, the change in the relative contributions of k obs1 and k obs2 to flavin reduction is similar. Possibly, the large effect on the reduction rate seen with vanillyl alcohol is due to incorrect positioning of the substrate in the active site of Y108F and Y503F. Hydrogen bonds between the phenolic hydroxyl group of the substrate and the two tyrosine residues may be of importance in maintaining it in the correct orientation for the hydride-transfer reaction to take place. This effect may be less important for chavicol than for vanillyl alcohol, as interactions involving its longer side chain may lock it into position in the active site. In this context, it is interesting to note Substrates can bind to the enzyme with their phenolic hydroxyl group in its protonated (SH) or deprotonated (S Ϫ ) form. Upon binding of the deprotonated form, the enzyme is reduced, yielding the reduced enzyme-quinone methide complex. Reduction by deprotonated substrate corresponds to k obs1 from the stopped-flow kinetic experiments presented here. Upon binding of the protonated form, the substrate must be deprotonated to allow flavin reduction to occur, i.e. the E ox -SH complex must be converted to the E ox -S Ϫ complex. This corresponds to k obs2 from the stopped-flow kinetic experiments. After the reduction, the product is released from the reduced enzyme. This step is too slow to be catalytically relevant, suggesting product release is faster in the presence of dioxygen.

Substrate activation in VAO
that the K m for the oxidation of chavicol by wt-VAO is significantly lower than that for the oxidation of vanillyl alcohol, indicating tighter binding of chavicol.
In addition to the 4PO subfamily members mentioned above, a catalytic motif composed of two tyrosine residues has been proposed to be involved in substrate activation in other enzymes in which catalysis proceeds through a para-quinone methide intermediate, such as the cofactor-independent hydroxycinnamoyl-CoA hydratase-lyase from Pseudomonas fluorescens AN103 (18) and bacterial phenolic acid decarboxylases (19) (although an alternative mechanism has also been proposed (20,21)). This suggests that these structurally unrelated proteins have obtained similar mechanisms for the activation of the phenolic hydroxyl groups of their substrates through convergent evolution.
Catalytic motifs composed of two tyrosine residues have also been implicated in substrate activation in a number of enzymes that catalyze oxidation reactions that do not involve a parasubstituted phenol. They have been proposed to facilitate the activation of hydroxyl groups for their oxidation to carbonyls in other oxidoreductases from the VAO/PCMH family, such as aclacinomycin oxidoreductase from Streptomyces galilaeus ATCC 31615 (22), Dbv29 from Nonomuraea sp. ATCC 39727 (23), GilR from Streptomyces griseoflavus (24), and the monolignol oxidoreductase BBE-like protein 15 from Arabidopsis thaliana (25). These enzymes all belong to the same clade of the VAO/PCMH family, which primarily contains enzymes with a bicovalently bound FAD cofactor (26). Despite their homology with VAO, a sequence and structural alignment of the five proteins demonstrated that the tyrosine pair found in these enzymes is not equivalent to that found in VAO, suggesting the motifs evolved separately.
In summary, the results presented here indicate that two tyrosine residues, Tyr-108 and Tyr-503, are responsible for activating the phenolic substrates of VAO for oxidation by facilitating the binding of their deprotonated phenolate form in the active site, shedding light on the catalytic mechanism of VAO and other homologous flavoenzymes that oxidize parasubstituted phenols.

Materials
Vanillyl alcohol and isoeugenol were from Sigma. Chavicol was from Quest International (Naarden, NL). All other chemicals were from commercial sources and of the purest grade available.

Cloning, protein expression, and purification
Constructs encoding for the Y108F and Y503F variants were created by linear whole-plasmid amplification from the pBC11 (27) plasmid using the mutagenic primers CATCTCTATTG-GAAGGAATTCCGGATTTGGCGGTGCTGCGCC (Y108F) and ATGGATGGGGCGAATTTCGAACCCATCTGGC (Y503F) (nucleotides in italics are the sites of mutation for introduction of the tyrosine to phenylalanine mutations, the underlined nucleotide indicates the site of a silent mutation that introduces an EcoRI site). After amplification and concomitant ligation, template DNA was digested using DpnI and the mutagenized plasmids were transformed into DH5␣ E. coli. Correct plasmids, as confirmed by sequencing, were transformed into BL21 E. coli for protein expression. The construct encoding for the Y108F/Y503F variant was created in a similar manner, using the plasmid encoding for the Y108F variant as the template for amplification and the primer for the Y503F variant. All variants were expressed and purified as described previously for wt-VAO (14).

Analytical methods
Unless specified otherwise, all experiments were performed in 50 mM potassium phosphate buffer, pH 7.5. Absorption spectra of purified proteins were recorded using a Hewlett Packard 8453 photodiode array spectrophotometer (Agilent Technologies, Santa Clara, CA) and enzyme concentrations were determined using the extinction coefficient of wt-VAO at 439 nm (⑀ 439 ϭ 12,500 M Ϫ1 cm Ϫ1 (1)). For trichloroacetic acid precipitations, 5% (w/v) trichloroacetic acid was mixed with 10 M enzyme and samples were incubated on ice for 30 min. Subsequently, the precipitated protein was pelleted by centrifugation (21,000 ϫ g, 15 min, 4°C), and the presence of FAD in the pellet or supernatant was judged from its color. The absence of FAD from the supernatants was confirmed by measuring their absorption spectra. Analytical gel filtration was performed using a Superdex 200 HR 10/30 column on an Ä kta pure chromatography system (GE Healthcare). Samples of enzyme at a concentration of 10 M were loaded onto the column and subsequently eluted in 50 mM potassium phosphate buffer, pH 7.5, containing 150 mM potassium chloride. Molecular masses were determined based on a calibration curve of proteins of known mass. Thermal unfolding of the VAO variants was monitored by the ThermoFAD method (11) using a MiniOpticon realtime PCR system (Bio-Rad). Protein solutions at a concentration of 7 M were subjected to a temperature gradient from 25 to 70°C and the fluorescence of the FAD cofactor was measured every 0.5°C with excitation and emission wavelengths of 485 Ϯ 30 and 625 Ϯ 30 nm, respectively. Thermal midpoints of unfolding were determined by finding the maximum of the derivative of the unfolding curves.

Steady-state kinetics
All experiments were performed at 25°C in 50 mM potassium phosphate buffer, pH 7.5. The oxidation of vanillyl alcohol to vanillin was followed by monitoring the absorption of the product at 340 nm (⑀ 340 ϭ 14,000 M Ϫ1 cm Ϫ1 ). The oxidation of chavicol was followed by monitoring oxygen consumption using a Hansatech Oxytherm system (Hansatech Instruments, King's Lynn, UK). Steady-state kinetic parameters were determined by fitting the obtained data to the Michaelis-Menten equation (vanillyl alcohol) using IGOR Pro (Wavemetrics, Lake Oswego, OR) or to a modified version of the Michaelis-Menten equation that includes a term to take substrate inhibition into account (chavicol),

Substrate activation in VAO
where k obs is the measured initial rate, k cat is the turnover number, K m is the Michaelis constant, K i is the inhibition constant for substrate inhibition, and [S] is the substrate concentration.

Stopped-flow kinetics
The reduction of VAO variants by vanillyl alcohol or chavicol was followed using an SX20 stopped-flow system (Applied Photophysics, Leatherhead, UK) equipped with a photodiode array detector or a photomultiplier detector as applicable. Enzyme and substrate solutions in 50 mM potassium phosphate buffer, pH 6.2 or 7.5, as appropriate, containing 5 mM glucose were made anaerobic by first flushing them with nitrogen gas for 10 min and then adding 0.3 M glucose oxidase. Subsequently, the enzyme and substrate solutions were mixed to final concentrations of 10 M enzyme and 2 mM (vanillyl alcohol) or 100 M (chavicol) substrate and spectral changes were monitored at 25°C. Rate constants were obtained by global fitting of multiple single wavelength traces to appropriate exponential functions using IGOR Pro. Spectral deconvolution of multiwavelength data were performed using Pro-Kineticist (Applied Photophysics). To study the pH dependence of the reduction of wt-VAO by chavicol in more detail, the reaction was performed at various pH values in the pH range 6.1-9.5. Experiments were performed in Britt-Robinson buffers consisting of 40 mM acetic acid, 40 mM phosphoric acid, and 40 mM boric acid set to the correct pH using potassium hydroxide. The ionic strength of the buffers was set to 0.2 M by the addition of potassium sulfate. Subsequently, enzyme and substrate solutions were made anaerobic as described above, mixed to final concentrations of 10 M enzyme and 50 M chavicol, and the reduction was monitored at 25°C and data were analyzed as described above.

Isoeugenol-binding studies
Changes in the spectral properties of the VAO variants upon binding of the competitive inhibitor isoeugenol were monitored using a Cary 4000 spectrophotometer (Agilent Technologies, Santa Clara, CA). Isoeugenol was titrated to 10 M enzyme in 50 mM potassium phosphate buffer, pH 7.5, and an absorption spectrum was recorded after every addition. Dissociation constants were determined by fitting the data at a specific wavelength to a model assuming formation of a 1:1 enzyme-isoeugenol complex using the following formula with IGOR Pro, with ⑀ E , ⑀ EI , and ⑀ I being the molar extinction coefficients of the free enzyme, the enzyme-inhibitor complex, and the free inhibitor, respectively, [E] f being the concentration of free enzyme, [EI] the concentration of enzyme-inhibitor complex, [I] f the concentration of free inhibitor, [E] t the total enzyme concentration, [I] t the total inhibitor concentration, and K d the dissociation constant.

Crystallization and structure determination
VAO variants were crystallized by the hanging drop vapor diffusion method. Protein solutions at a concentration of 180 -220 M in 50 mM potassium phosphate buffer, pH 7.5, were mixed 1:1 (v/v) with a mother liquor containing 6% (w/v) PEG4000 in 100 mM sodium acetate buffer, pH 4.6, and equilibrated against the mother liquor at 20°C. Prior to data collection, crystals were transferred to a cryoprotectant solution containing 10% (w/v) PEG4000 and 20% (w/v) glycerol in 100 mM sodium acetate buffer, pH 4.6, and cryo-cooled in liquid nitrogen. X-ray diffraction data were collected at the X06DA beamline of the Swiss Light Synchrotron in Villigen, Switzerland (SLS), and at the ID30B beamline of the European Synchrotron Radiation Facility in Grenoble, France (ESRF). Images were integrated using MOSFLM (28) and data were scaled using programs of the CCP4 suite (29). Detailed data processing statistics are shown in Table 4. Structures of VAO mutants were solved by molecular replacement using PHASER (30) with the structure of wt-VAO (PDB code 2VAO (4)) devoid of all ligands and water molecules as a search model. Subsequently, structures were refined by alternating refinement using REFMAC5 (31) and manual model building using COOT (32). Figures were prepared with CCP4mg (33). Atomic coordinates and structure factors were deposited in