Enzymes in the glucose–methanol–choline (GMC) oxidoreductase superfamily catalyze the oxidation of an alcohol moiety to the corresponding aldehyde. In this review, the current understanding of the sugar oxidation mechanism in the reaction of pyranose 2-oxidase (P2O) is highlighted and compared with that of other enzymes in the GMC family for which structural and mechanistic information is available, including glucose oxidase, choline oxidase, cholesterol oxidase, cellobiose dehydrogenase, aryl-alcohol oxidase, and pyridoxine 4-oxidase. Other enzymes in the family that have been newly discovered or for which less information is available are also discussed. A large primary kinetic isotope effect was observed for the flavin reduction when 2-d-d-glucose was used as a substrate, but no solvent kinetic isotope effect was detected for the flavin reduction step. The reaction of P2O is consistent with a hydride transfer mechanism in which there is stepwise formation of d-glucose alkoxide prior to the hydride transfer. Site-directed mutagenesis of P2O and pH-dependence studies indicated that His548 is a catalytic base that facilitates the deprotonation of C2–OH in d-glucose. This finding agrees with the current mechanistic model for aryl-alcohol oxidase, glucose oxidase, cellobiose dehydrogenase, methanol oxidase, and pyridoxine 4-oxidase, but is different from that of cholesterol oxidase and choline oxidase. Although all of the GMC enzymes share similar structural folding and use the hydride transfer mechanism for flavin reduction, they appear to have subtle differences in the fine-tuned details of how they catalyze substrate oxidation.
Regiospecific oxidation of organic molecules at the –CH–OH moiety is useful for many biotechnological and diagnostic applications . Flavoenzymes in the glucose–methanol–choline (GMC) oxidoreductase superfamily are particularly useful biocatalysts for this process, as they employ a cheap and clean oxidant, molecular oxygen, to generate oxidized organic products and H2O2 [1, 2]. For example, glucose 1-oxidase (GO) and cholesterol oxidase (ChO) have long been used in conjunction with peroxidases as biosensors for detecting blood d-glucose and cholesterol, respectively [3, 4]. Pyranose 2-oxidase (P2O) is useful for the production of 2-keto-aldoses, which are precursors for the synthesis of the low-calorie and noncariogenic sugar, d-tagatose, or the β-pyrone antibiotic cortalcerone [1, 5]. The mechanism by which the –CH–OH moiety is oxidized by the GMC enzymes has been investigated by many research groups [6, 7], and considerable progress has been achieved over the past few years [4, 8-12]. The aim of this review is to highlight recent advances in our understanding of sugar oxidation by P2O, and compare it with the catalytic features of other enzymes in the GMC family [GO, ChO, choline oxidase (CO), aryl-alcohol oxidase (AAO), cellobiose dehydrogenase (CDH), and pyridoxine 4-oxidase (PNO)] for which structural and mechanistic information is available. Mechanistic studies of methanol oxidase (MO) are also discussed. For more comprehensive reviews of ChO, CO, AAO, and MO, please refer to other available references [4, 6, 8, 9, 13]. In addition, biochemical characterizations of recently discovered enzymes in the GMC class, including compound K oxidase and a new alcohol oxidase, are discussed.
General properties of P2O
P2O (pyranose:oxygen 2-oxidoreductase; EC 1.13.10) is a fungal flavoenzyme that catalyzes the oxidation by molecular oxygen of several aldopyranoses at the C2 position to yield the corresponding 2-keto-aldoses and H2O2  (Scheme 1). P2O contains FAD as a cofactor, like other enzymes in the GMC family. The FAD cofactor in P2O is covalently linked through a histidyl linkage (His167) . The physiological role of P2O is thought to be to generate H2O2 as a substrate for lignin-degrading enzymes. In some species of white rot fungus, P2O is involved in a secondary metabolic pathway that produces antibiotics such as β-pyrone-cortalcerone, which helps in protecting the fungi from bacterial attack [5, 16]. P2O is a useful biocatalyst in several biotechnological applications, including biotransformation of carbohydrates such as d-glucose and d-galactose to generate 2-keto-sugars that can be further reduced at the C1 position to yield d-fructose and d-tagatose, respectively [1, 5]. In addition, P2O has the potential to be useful for biofuel cell applications .
P2O from Trametes multicolor was isolated, purified, and biochemically and biophysically characterized. This enzyme is a homotetrameric flavoprotein with a native molecular mass of 270 kDa . Crystal structures of P2O with or without ligand bound in many forms are available [15, 18, 19]. The overall folding of P2O is similar to that of other enzymes in the GMC superfamily (Fig. 2). The enzyme structures indicate that binding of a sugar substrate is compatible with open or semi-open conformations of the substrate-recognition loop (residues 452–457) [15, 18, 20], and this loop undergoes conformational rearrangement into a fully closed form when a small ligand such as acetate binds. The closed loop conformation seems to be the most likely conformation for the reaction of the enzyme with oxygen [18, 21].
Overall catalysis by P2O and the unique feature of the oxidative half-reaction
The catalytic reaction of P2O can be divided into a reductive half-reaction, in which two electrons are transferred as a hydride equivalent from a sugar substrate to the oxidized FAD to generate the reduced FAD and a 2-keto-sugar, and an oxidative half-reaction, in which two electrons are transferred from the reduced flavin to oxygen to form H2O2 [21, 22]. The catalytic reaction of P2O is described in Scheme 2, and kinetic constants associated with each reaction step are summarized in Table 1 [11, 12, 21, 22]. The steady-state kinetics of P2O can be classified as a ping-pong bi-bi type, because the 2-keto-sugar product is released prior to the oxygen reaction, as with GO, ChO, and CDH . An exception was found for the T169A mutant, in which a hydrogen bond between Thr169 and the N5 atom of FAD is absent . The kinetic mechanism of the T169A mutant with d-glucose or d-galactose indicates that a 2-keto-sugar product remains bound at the active site during the oxidative half-reaction, because an intersecting pattern of the double-reciprocal plot was obtained when both substrate concentrations were varied .
Table 1. Rate constants of WT P2O according to the kinetic mechanism described in Scheme 2 at 4 °C and pH 7.0 [21, 22]
5.8 × 104m−1·s−1
2.5 × 103 s−1
5.8 × 104m−1·s−1
P2O is the only enzyme for which C4a-hydroperoxyflavin has been detected as an intermediate during the oxidative half-reaction (Scheme 2) [12, 21]. It is still not known whether C4a-(hydro)peroxyflavins are common among other flavoenzyme oxidases as well, but the other enzymes simply fail to stabilize the intermediate, or whether this intermediate is unique to P2O [21, 24]. The C4a-hydroperoxyflavin intermediate is common for flavoprotein monooxygenases, because it is required as an oxygenating reagent [24-31]. However, for P2O, the C4a-hydroperoxyflavin merely eliminates H2O2 to generate oxidized FAD (Scheme 2; Table 1) . Mutation of His167 (H167A) of P2O, which removes the flavin covalent linkage, results in a mutant enzyme that behaves similarly to the wild-type (WT) enzyme in the oxidative half-reaction [10, 21, 22]. However, mutation of Thr169, where Oγ of the hydroxyl group of the side chain interacts with the flavin N5, to Ser, Asn, Ala or Gly resulted in the abolition of a quasi-stable C4a-hydroperoxyflavin intermediate . Recent investigations on the effects of pH on the oxidative half-reaction of P2O indicate that, at pH values > 7.6, flavin oxidation switches to proceed via a mode that does not stabilize C4a-hydroperoxyflavin .
Investigations using transient kinetics and isotope effects have shown that the breakage of the flavin N(5)–H bond controls the overall process of H2O2 elimination from C4a-hydroperoxyflavin . The H2O2 elimination from the intermediate (k5 in Scheme 2) shows a solvent kinetic isotope effect (SKIE) of 2.8, and has a linear relationship with the proton inventory plot, suggesting that a single proton transfer causes the SKIE on H2O2 elimination . The key to identifying the factor responsible for controlling the elimination of H2O2 came from a specific labeling experiment in which only the reduced flavin N5 was labeled with deuterium . The data indicated that the N5-deuterated reduced flavin generates a similar KIE to that found in the experiment performed with the enzyme pre-equilibrated in D2O buffer. Therefore, the N–H(D) cleavage at the flavin N5 position is responsible for the SKIE, providing the proton-in-flight that is transferred during the transition state.
Mechanism of sugar oxidation in P2O
Formation of the enzyme–substrate complex
For the sugar oxidation process (reductive half-reaction), the reaction begins with binding of d-glucose, which results in an increase in absorbance at ~ 395 nm . This type of spectroscopic change upon substrate binding has not been documented for other enzymes in the GMC oxidoreductase family. When 2-d-d-glucose was used instead of d-glucose, stopped-flow detection showed that a greater increase in absorbance at 395 nm occurred because of the subsequent slower flavin reduction step, indicating that the absorbance at 395 nm corresponds to the formation of a Michaelis complex of d-glucose and P2O [10, 22] (Fig. 4). The rate of the increase in absorbance at 395 nm approaches a limiting value at high d-glucose concentrations, indicating that the species observed at 395 nm is formed in at least two steps (Scheme 2); little or no absorbance change occurs in the first step. The step causing the absorbance change is possibly isomerization of the enzyme–substrate complex or formation of the alkoxide intermediate . The isomerization step showed an inverse isotope effect (k2H/k2D of ~ 0.60), implying that the C2–H bond of d-glucose is more rigid in the d-glucose–P2O complex than when d-glucose is free in solution . The cause of this inverse isotope effect is currently unclear. As the d-glucose alkoxide intermediate presumably forms prior to the flavin reduction step  (see more details below), we propose that the second isomerization step may be involved in d-glucose alkoxide generation (Scheme 2). The binding of d-galactose to P2O also resulted in an absorbance increase at 395 nm, with the rate constants of this step being linearly dependent on d-galactose concentration. The linear dependence on d-galactose concentration may be attributable to a much larger Kd value if thed-galactose binding is also involved in a two-step binding process. Alternatively, the binding of d-galactose may simply be involved in a one-step process .
For the reaction of the H167A mutant, in which the FAD covalent linkage was removed, with d-glucose, a multistep binding process was observed. This phenomenon may be related to the presence of different sugar-binding modes in this mutant, as observed in its X-ray structures. The crystal structure of the H167A mutant in complex with 2-fluoro-2-deoxy-d-glucose (2FG) indicates that the ligand binds with the sugar C3 position close to the flavin N5, and that the substrate loop is in the open conformation (Fig. 5B) . This binding mode would promote sugar oxidation at the C3 position (C3 oxidation mode). The structure of the H167A mutant in complex with 3-fluoro-3-deoxy-d-glucose (3FG) indicates that the ligand binds to the enzyme with the sugar C2 position close to the flavin N5 (C2 oxidation mode), and that the substrate loop is in the semi-open conformation (Fig. 5C) . As the experiments with d-glucose and deuterated d-glucose (discussed below) indicated that the H167A mutant oxidizes d-glucose regiospecifically at the C2 position, we propose that d-glucose may initially bind to the open loop conformer (C3 oxidation mode). Then, a subsequent isomerization step may lead to rearrangement of the initial complex into the C2 oxidation mode (semi-open loop conformation). Phe454 and Tyr456 on the substrate loop are key residues that allow optimized interactions with the sugar (Fig. 5D).
Unlike for the WT enzyme, when 2-d-d-glucose was used as a substrate in the reaction of the H167A mutant, no inverse kinetic isotope effect was observed for the d-glucose binding step [10, 22]. This indicates that the increased bond constraint at the C2–H bond of d-glucose upon WT enzyme binding could not be detected in the complex of the H167A mutant and d-glucose. This difference may be attributable to greater dynamics and flexibility of the noncovalent FAD cofactor in the active site of the H167A mutant.
Activation of sugar oxidation and flavin reduction
Comparison of the active sites of P2O from T. multicolor and other enzymes in the GMC oxidoreductase superfamily shows that His/Asn (P2O, ChO, CO, and CDH) or His/His (GO and AAO) are highly conserved, and are positioned near the isoalloxazine ring (Figs 6-10). Studies of the reductive half-reaction of P2O at various pH values with WT and mutant enzymes suggest that the conserved His548 is a catalytic base that is important for the deprotonation of d-glucose in P2O . Mutations of His548 to various amino acids, e.g. H548A, H548N, H548S, H548D, H167A/H548A, H167A/H548N, H167A/H548S, and H167A/H548D, resulted in inactive enzymes at all pH values. The catalytic impairment of these mutants is not caused by drastic structural changes or unfavorable thermodynamics. For example, the structure of the H548N mutant is very similar to that of the WT enzyme . The redox potentials of the H167A/H548A and H167A/H548N mutants are approximately −200 mV (~ 100 mV lower than that of the WT enzyme, and ~ 50 mV lower than that of the H167A mutant) (Table 2). This level of redox potential should not cause a problem for the flavin reduction, because the T169A mutant, which has a redox potential of −197 mV, can be reduced by d-glucose, but at a lower rate of 0.03 s−1 at 25 °C (Table 2) . All of the data suggest that the impairment of H548 mutants must be caused by the absence of a base (His548) that is important for catalysis .
Table 2. Summary of the redox potential values and the rate constants for flavin reduction of WT P2O and mutants at 25 °C and pH 7.0 [11, 18, 23]. The kred value could not be determined, because the enzymes are inactive for the reductive half-reaction. ND, not determined
Reduction rate constants (k3) (s−1) for d-glucose at 25 °C
−105 ± 1
48 ± 2
−150 ± 2
3.97 ± 0.29
−197 ± 3
0.03 ± 0.007
−119 ± 1
0.42 ± 0.07
+ 17 ± 3
0.018 ± 0.001
−190 ± 2
−217 ± 2
Evidence for the role of His548 as a catalytic base in the P2O reaction is given by results from studies with the H167A/H548R mutant . The single amino acid substitution H548R resulted in a nonhomogeneous mixture of covalently linked and non-covalently bound FAD. Thus, double mutants were constructed with H167A along with various substitutions at His548 in order to obtain a homogeneous protein with non-covalently bound FAD (Table 3). An additional effect of having the H167A mutation in the His548 mutants is merely to decrease the flavin reduction rate constant (~ 22-fold), with other catalytic properties being conserved . The results from the stopped-flow investigation indicate that the observed rate constant for flavin reduction of the H167A/H548R mutant is 0.018 s−1 at pH 7.0 (Table 2). As other His548 mutants are inactive (discussed previously), the flavin reduction ability of the H167A/H548R mutant must be attributable to the participation of Arg548 in the reaction.
Table 3. Percentage of covalently linked FAD found in WT P2O and mutants [11, 19]
Covalently linked (%)
98 ± 2
63 ± 3
42 ± 4
85 ± 3
49 ± 2
26 ± 1
The rate constant associated with the flavin reduction (sugar oxidation) step (k3, Scheme 2) of the H167A/H548R mutant also increases with pH. As the increase in pH is likely to decrease the reduction potential, the increase in rate with the increase in pH is not attributable to an increase in the thermodynamic driving force. It is more likely that the pH increase causes Arg548 in the H167A/H548R mutant to be unprotonated . Upon an increase in the pH, the reduction rate constant of the H167A/H548R mutant increases ~ 360-fold, and is consistent with a pKa value of > 10.1 (Table 4) . This pKa value is in the range of an Arg in aqueous solution (~ 12.4). The data suggest that Arg548 in the mutant and His548 in the WT enzyme must be in the unprotonated state in order to act as good catalytic bases. Because the flavin reduction rate constants of the WT enzyme and the H167A mutant are pH-independent between pH 5.5 and pH 10.5, the data also imply that the His548 in the WT enzyme must be unprotonated (neutral) throughout this pH range. The unusually low pKa of His548 may be the result of the rather hydrophobic active site environment of P2O.
Table 4. Summary of the reduction rate constants for the reactions of oxidized H167A/H548R mutant with d-glucose at various pH values 
Reduction rate constants (s−1) for d-glucose at 25 °C
0.006 ± 0.001
0.018 ± 0.001
0.057 ± 0.004
0.63 ± 0.03
1.60 ± 0.06
2.13 ± 0.03
Studies of SKIEs on the transient kinetics of the WT and H167A enzymes provided insights into the mechanism of sugar activation . There is no SKIE on the flavin reduction step in either the WT or H167A enzyme, suggesting that the deprotonation of the C2–OH bond does not coincide with the hydride transfer from the C2–H bond . If the deprotonation of the C2–OH bond occurs simultaneously with the hydride transfer, the reaction performed in D2O should have a lower rate constant, owing to the deprotonation of C2–OD instead of C2–OH. The data indicate that a d-glucose alkoxide intermediate exists prior to the transfer of a hydride equivalent from the C2–H group to the flavin N(5) (species 3, Scheme 2) . Therefore, the flavin reduction mechanism of P2O is consistent with a mechanism by which the unprotonated form of His548 acts as a catalytic base to abstract a proton at the C2–OH bond of d-glucose to generate an alkoxide intermediate prior to the hydride transfer step (Scheme 2) . The resulting protonated His548 may also contribute to stabilization of the alkoxide intermediate (Scheme 2).
Regiospecificity and selectivity of sugar oxidation
Pre-steady-state kinetic studies with 2-d-d-glucose as a substrate for WT P2O showed a large primary kinetic isotope effect (kH/kD = 8.84) on the flavin reduction step , whereas no additional kinetic isotope effect was observed when all-deuterated d-glucose (1,2,3,4,5,6,6-d7-d-glucose) was used in the reactions with the H167A mutant . These results clearly indicate that P2O oxidizes the C–H moiety of d-glucose regiospecifically at the C2 position. Pre-steady-state and steady-state kinetics have also shown that flavin reduction is a rate-limiting step in the overall reactions of the WT enzyme and the H167A mutant [10, 22].
The recently reported crystal structure of the H167A mutant in complex with 3FG at 1.35-Å resolution shows interactions of the substrate analog with the active site residues, which might be important for the regiospecific oxidation of d-glucose at the C2 position (Fig. 5C) . In the previously reported structure , the C2 position of 2FG is oriented far away (> 4 Å) from His548 and the flavin N(5), which is not ideal for oxidation at the C2 position, but favors oxidation at the C3 position (Fig. 5B). A major difference between the C2 and C3 oxidation conformations lies in the position of a flexible loop (residues 452–456) that forms part of the substrate-recognition site (Fig. 5D). In the complex of the H167A mutant and 2FG (C3 oxidation), the substrate-recognition loop swings further away from the active site to assume the open conformation, whereas in the C2 oxidation complex of the H167A mutant and 3FG, the gating segment (residues 452–456) swings towards the active site to provide a binding pocket for the substrate (green; Fig. 5D) . Mutagenesis of Phe454 and Tyr456 resulted in inactive enzymes, indicating that this region is functionally important for P2O .
Site-directed mutagenesis and kinetic studies of Thr169 mutants  also indicate that mutation at this position can broaden the substrate specificity of P2O to accommodate the C2 oxidation of d-galactose. Thr169 is positioned close to the flavin N5 and located in the closed loop conformer, with its Oγ involved in a hydrogen bond interaction with the flavin N5 [19, 23]. Stopped-flow experiments with the WT enzyme and Thr169 mutants showed that P2O binds d-galactose in a one-step binding process, which is different from the binding of d-glucose. Most of the Thr169 mutants have Kd values for d-galactose that are ~ 10-fold lower than that for d-glucose (Table 5). For the T169S, T169N and T169G mutants, the flavin reduction rate constant (kred) with d-galactose is larger (approximately nine fold for the T169G mutant) than that of the WT enzyme with d-galactose (Table 5). This result implies that d-galactose binds to the T169 mutants with a more optimal geometry than to the WT enzyme, allowing sufficient mobility of d-galactose to enable it to be more feasible for the hydride transfer reaction to proceed. This improvement is presumably attributable to the removal of steric hindrance close to the axial C4–OH of d-galactose in the T169 mutants.
Table 5. Summary of the kinetic parameters for the reactions of Thr169 mutants with d-glucose and d-galactose at 4 °C and pH 7.0 [22, 23]. The Kd value could not be determined, because the absorbance increase at 395 nm of the first phase was not significant . ND, not determined
15.3 ± 0.4
0.3 ± 0.03
13.8 ± 0.4
0.5 ± 0.04
9.7 ± 0.1
0.9 ± 0.01
0.7 ± 0.01
2.7 ± 0.02
Most of the evidence suggests that the proper binding of substrate with respect to the crucial residues in the active site is key for achieving sugar oxidation. The thermodynamic driving force of the oxidation, i.e. the reduction potential of enzyme-bound FAD, is not a major factor determining the rate constant of flavin reduction (Table 2). For example, the H167A/H548R mutant, with a reduction potential higher than that of the WT enzyme (+17 versus −105 mV), reacts poorly with d-glucose (Table 2). Of the T169 mutants, only the enzymes that possess hydrogen-bonding ability similar to the that of the WT enzyme, i.e. the T169S and T169N mutants, react with d-glucose with similar Kd and kred values as those of the WT enzyme (Table 5).
Comparison of P2O and other enzymes in the GMC family
Enzymes in the GMC family share similar overall three-dimensional structures, in which FAD-binding and substrate-binding domains can be distinguished. When their active site residues are aligned with respect to the FAD moiety, a few conserved residues can be identified (Figs 6-11). The most notable feature is the conserved His/His, His/Asn or His/Pro pair: His516/His559 in GO from Aspergillus niger, His502/His546 in AAO from Pleurotus eryngii, His447/Asn485 in ChO from Streptomyces, His466/Asn510 in CO from Arthrobacter globiformis, His689/Asn732 in CDH from Phanerochaete chrysosporium, His548/Asn593 in P2O from T. multicolor, and His462/Pro504 in PNO from Mesorhizobium loti. The functional roles of the His/His, His/Asn and His/Pro pairs in catalysis have been investigated (discussed below). Interestingly, the conclusions drawn about their functional roles in different systems are not the same. For example, the conserved His in the His/Asn(His/Pro) pair of P2O, AAO, GO, CDH and PNO is thought to be a catalytic base for substrate oxidation, on the basis of site-directed mutagenesis, pH-dependence studies or theoretical calculations. For CO and ChO, results from an imidazole-rescue experiment and a high-resolution X-ray structure do not support the role of this residue as a catalytic base, and the identity of the catalytic base for these systems remains elusive. The reaction mechanism of MO was proposed to be hydride transfer, based on studies of kinetic isotope effects. As results of site-directed mutagenesis and X-ray structure determination are not available, no putative base was identified.
GO catalyzes the oxidation of d-glucose at the C1 position to form glucono-lactone (Scheme 3). Stopped-flow studies of the reductive half-reaction with d-glucose as a substrate indicate that the rate constant of the flavin reduction is very high and is directly dependent on substrate concentrations in a second-order fashion . When 1-d-d-glucose was used, a large primary kinetic isotope effect (~ 15 at 3 °C) on the flavin reduction was observed, and the rate constant reached a limit of 67 s−1 at 3 °C. . Bright et al. proposed that the substrate oxidation involves a direct hydride transfer from the C1 position of glucose to the N5 position of FAD . The crystal structure of glucose oxidase from A. niger at 2.3-Å resolution was solved, and d-glucose was modeled into the active site . The structure showed three potential catalytic residues located near the reaction center: Glu412 and two conserved His residues (His516 and His559) (Fig. 6). Although the distances between the C1–OH proton of the modeled glucose and the nitrogen atoms of His516 and His559 are similar, the His516 (equivalent to His548 in P2O) has been proposed to be a catalytic base, owing to its lower pKa value according to computational calculations . Scheme 4 summarizes the current reaction mechanism proposed for of GO [35, 37].
CDH catalyzes the oxidation of cellobiose at the C1 position to form cellobiono-δ-lactone (Scheme 5). The crystal structure of CDH in complex with cellobiono-1,5-lactam (1.8-Å resolution) , an anomer of cellobiose, indicates that the C1 hydrogen of the substrate, which is transferred as a hydride equivalent during the reductive half-reaction, is located 3.0 Å from the N5 position of FAD. The conserved His/Asn pair (His689 and Asn732) is positioned near the isoalloxazine ring (Fig. 7). His689 is the only basic residue in the vicinity, and is proposed to be the base that abstracts a proton to initiate a hydride transfer reaction. This idea is supported by the fact that the kcat values of the His689 mutants are > 1000-fold lower than that of the WT enzyme when cellobiose and its epimer (lactose) are used as substrates . The crystal structure also shows the hydrogen bonding of Asn732 with the anomeric hydroxyl group, suggesting that this residue may contribute to substrate binding . The effects of the Asn732 mutation are varied, depending on the identities of the amino acid substitutions in the mutants. kcat values of the mutants for cellobiose oxidation were 5–4000-fold lower than that of the WT enzyme . Therefore, Asn732 is thought to be a residue that facilitates binding of the substrate (Scheme 6).
CO catalyzes the double oxidation of choline to form glycine betaine via the formation of betaine aldehyde (Scheme 7). The enzyme reaction mechanism has been investigated extensively by Gadda  Comprehensive reviews on the reductive half-reaction and oxidative half-reaction  of the enzyme are available. The crystal structure of CO at 1.86 Å indicates that the two conserved His residues (His351 and His466) are located near the flavin ring  (Fig. 8). Mutation of these residues to Ala decreased kcat values by ~ 60-fold [42, 43]. The activity of the H466A mutant could be rescued by adding exogenous imidazolium but not imidazole, indicating that the positive charge of His is important for catalysis . The kcat of the reaction increases upon an increase in pH, with a pKa of ~ 7.5 . Gadda et al. interpret these data to be the result of deprotonation of the choline substrate to form an alkoxide intermediate [8, 45] (Scheme 8). However, a different catalytic model based on theoretical energy profiles and protonation predictors was suggested by the Martinez group. The computational results predict that His466 in CO acts as a base to generate an alkoxide intermediate .
The results of studies on the solvent and primary kinetic isotope effects of the WT enzyme [D2O(kred) = 0.99 and D(kred) = 8.9] indicate that there is no SKIE for the flavin reduction step, suggesting that formation of the alkoxide species is decoupled from a hydride transfer step , which is similar to the P2O reaction discussed above. It was proposed that stabilization of the alkoxide species is achieved through electrostatic and hydrogen-bonding interactions with His466 and Glu312 [8, 41, 43] (Scheme 8). Glu312 is important for the reductive half-reaction, because the E312A mutant is inactive. The E312Q mutant showed a Kd value that was ~ 500-fold higher than that of the WT enzyme. The E312D mutant had a kcat/KO2 value similar to that of the WT enzyme, but the kcat/Km(choline) value and the flavin reduction rate were ~ 30-fold and ~ 260-fold lower, respectively, than those of the WT enzyme. The data indicate that both a negative charge and a proper side chain length of Glu312 are required for the reductive half-reaction. The negative charge of Glu312 may be important for binding and positioning the alcohol moiety for the hydride transfer reaction .
AAO catalyzes the oxidation of p-methoxybenzyl alcohol to form p-anisaldehyde (Scheme 9). The crystal structure of P. eryngii AAO was solved at 2.55-Å resolution, and shows the conserved catalytic residues of His502 (equivalent to His548 in P2O) and His546 (equivalent to Asn593 in P2O) located near FAD  (Fig. 9). Both His residues are important for substrate binding and oxidation. The Kd value of the H502A mutant is 122-fold greater than that of the WT enzyme, and that of the H546A mutant is 50-fold greater. Mutation of His502 caused a drastic decrease in the kcat and kred (~ 1250–2500-fold lower than that of the WT enzyme), wheras mutation of H546 caused only moderate effects (35–50-fold reduction in kcat and kred) [46, 48].
A plot of the pH-dependence rate profile for WT AAO does not show clear pH-dependent behavior, but a bell-shaped profile was observed for the H546S mutant, consistent with the presence of two ionizable groups. One group, with a pKa of ~ 3.5, must be unprotonated, and another, with a pKa of ~ 8.3, must be protonated in order for maximum activity to be achieved. For the H502S mutant, only a group with a pKa of ~ 7.5 could be observed, suggesting that the residue that is required to be unprotonated with a pKa of ~ 3.5 in the H546S mutant must be His502. The theoretical energy profiles and protonation predictors are also in support of His502 acting as a base [46, 48] (Scheme 10).
Although both AAO and P2O use conserved His residues as catalytic bases, their hydride transfer mechanisms are somewhat different. For AAO, hydride transfer occurs via a concerted mechanism, [49, 50], whereas for P2O it probably occurs via a stepwise mechanism (discussed previously). Transient kinetics and substrate and solvent kinetic isotope effects of the AAO reaction showed a kinetic isotope effect of ~ 9 when α-bideuterated substrate was used, and a kinetic isotope effect of ~ 13 when both α-bideuterated substrate and D2O solvent were used. These results are consistent with a synchronous mechanism in which the proton abstraction from a substrate hydroxyl group and the hydride transfer step occur simultaneously  (Scheme 10). This conclusion is also supported by theoretical calculations, because the quantum mechanics/molecular mechanics energy profiles of AAO do not identify a stable intermediate such as alkoxide along the proton and hydride transfer coordinates . Interestingly, the hydride transfer reaction of AAO is highly stereoselective. When (R)-α-deuterated and (S)-α-deuterated p-methoxybenzyl alcohol were used, a primary kinetic isotope effect of ~ 6 was observed only for the R-enantiomer. Docking of p-methoxybenzyl alcohol at the active site, together with quantum mechanics/molecular mechanics calculations, suggests that this stereoselectivity is a result of the arrangement of the hydride-receiving atoms (flavin N5 and His502 Nε) in relation to the alcohol Cα-substituents .
ChO catalyzes the oxidation of cholesterol to 5-cholesten-3-one and the subsequent isomerization to form 4-cholesten-3-one as the final product (Scheme 11). Originally, His447 of the conserved His/Asn pair was proposed to act as a general base to abstract a proton from the 3-hydroxyl group of cholesterol to facilitate the hydride transfer of the cholesterol oxidation. Mutagenesis of His447 resulted in a significant reduction in cholesterol oxidation activity, suggesting that His447 is important for the catalytic mechanism. The kcat values of the H447E and H447D mutants are > 100 000-fold lower than that of the WT enzyme, while those of the H447Q and H447N mutants are 120-fold and 4400-fold lower, respectively, than that of the WT enzyme . However, this proposal was modified when the crystal structure of ChO at 0.95-Å resolution was solved. The structure shows that His447 exists in the neutral imidazole form, with the Nε2-H group of His447 pointing towards the substrate hydroxyl group  (Fig. 10). Therefore, Vrielink et al. proposed that His447 acts as a hydrogen bond donor, in order to optimize the substrate binding so that the correct orbital alignment for hydride transfer is achieved . A general base that assists in proton abstraction to promote the hydride transfer has not been identified  (Scheme 12). Additional structural studies have suggested that Tyr446 facilitates flavin reduction by exerting steric pressure on the isoalloxazine ring  (Fig. 10). Asn485 also undergoes hydrogen bond interactions with the flavin π-system  (Fig. 10), and the N485L mutation decreases kcat/Km by ~ 1300-fold. It has been proposed that this residue modulates the electrostatic potential of the flavin to enhance cholesterol oxidation .
PNO is the first enzyme in the vitamin B6 degradation pathway. It contains FAD, and catalyzes the oxidation of pyridoxine to pyridoxal (Scheme 13). PNO has been purified from three bacteria: Pseudomonas sp. MA-1 , Microbacterium luteolum , and M. loti . The enzymes from M. loti and Mi. luteolum are monomeric, and have molecular masses of 55 and 54 kDa, respectively. PNO shows high substrate specificity, because three pyridinium compounds (2,6-dihydroxypridine, 3-pyridinemethanol, and 4-pyridinemethanol) gave only ~ 0.15% of the activity as compared with pyridoxine, and the enzyme cannot use pyridoxal 5′-phosphate as a substrate . The crystal structures of PNO at 2.2-Å resolution and the PNO–pyridoxamine complex at 2.1-Å resolution have been recently reported . His460, His462 and Pro504 are located near the flavin (Fig. 11). PNO is the first enzyme in the GMC family in which the His/Pro (His462/Pro504) pair has been identified instead of the His/His or the His/Asn pair. The side chains of His462 and His460 are located ~ 2.7 and 3.1 Å from the N4′ atom of pyridoxamine. The kcat/Km values of the H462A mutant (1.5 × 104m−1·s−1) and the H460A mutant (6 × 103m−1·s−1) are approximately 100–200-fold lower that of the WT enzyme (1.3 × 106m−1·s−1). The H460A/H462A double mutant showed no activity . On the basis of these results, Mugo et al. suggested that His462 (equivalent to His548 in P2O) may act as a general base to abstract a proton from the 4′-OH of pyridoxine, and the hydride from C4′ may be transferred to N5 of the flavin. His460 and Pro504 may be involved in the binding and positioning of the pyridoxine substrate (Scheme 14) .
Other enzymes in the GMC family
MO from the yeast Hansenula polymorpha catalyzes the oxidation of short-chain primary alcohols to the corresponding aldehydes, with concomitant reduction of molecular oxygen to H2O2 (Scheme 15) . MO belongs to the GMC family, and the enzyme is an octamer with a subunit molecular mass of 83 kDa. Seven molecules of flavin were found per octamer . Only two of the flavin cofactors are present in the oxidized state and participate in the reaction. EPR spectra indicate that the other five flavins are present in the reduced semiquinone state and are not involved in the catalytic reaction . As the X-ray structure of MO is currently not available, it is not possible to compare the structure and mechanistic roles of the active site residues of MO with those of other enzymes.
Incubation of MO with cyclopropanol results in enzyme inactivation and covalent attachment of the ring-opened inhibitor to the flavin . Two mechanisms have been proposed for this inactivation: a radical mechanism and an ionic mechanism . Kinetic isotope effects on the reaction of MO from H. polymorpha with 2-substituted ethyl alcohols (2-chloroethanol and 2-bromoethanol) have been investigated. The results showed a kinetic isotope effect of 5 on the Vmax/Km value, and an SKIE very close to 1 . These results were interpreted as indicating a hydride transfer mechanism in which the –OH bond cleavage occurs prior to the –CH bond cleavage in a stepwise fashion [suggesting the formation of a putative alkoxide intermediate (Scheme 16)] .
Compound K oxidase
A new FAD-bound enzyme, compound K oxidase from Rhizobium sp. GIN611, oxidizes ginsenoside (compound K) to form (S)-protopanaxadiol. The enzyme is a member of the GMC family, and consists of two subunits with molecular masses of 63.5 and 17.5 kDa. Sequence analysis has indicated that His493 and Asn536 are the conserved His and Asn commonly found in the GMC family. A small-subunit protein is important for expression of a soluble larger protein, which catalyzes oxidation and deglycosylation. The deglycosylation reaction is thought to occur spontaneously after oxidation of the glucose moiety of ginsenoside .
A new alcohol oxidase from Bjerkandera sp
A new alcohol oxidase from Bjerkandera sp. is a monomeric enzyme containing a non-covalently bound FAD with a molecular mass of 76 kDa . This enzyme can be classified as a member of the GMC family . The enzyme can use several compounds as substrates, including aromatic and polyunsaturated aliphatic primary alcohols, similarly to AAO from P. eryngii. However, this enzyme is different from AAO, in that it can efficiently oxidize phenolic benzyl and cinnamyl alcohols, and chlorinated benzyl alcohols. These alcohols are typical substrates for another flavoenzyme, vanillyl alcohol oxidase (VAO), which is not a member of the GMC family. However, the pH optimum (pH 6.0) for p-anisyl and vanillyl alcohol oxidation by this enzyme is different from that of VAO, which has a pH optimum in the alkaline range . The data suggest that the mechanism of alcohol oxidation of the new GMC oxidase is different from that of VAO.
Conclusions and future perspectives
Recent investigations of P2O and other GMC enzymes have advanced our understanding of the reaction mechanisms of enzymes in this superfamily. Most of the data are consistent with the –CH–OH moiety of substrates being oxidized via a hydride transfer to generate the reduced flavin cofactor and oxidized products. The His/His, His/Asn and His/Pro pairs, which are conserved among these enzymes, are important for substrate oxidation. Results from site-directed mutagenesis and pH-dependence studies of P2O and AAO have identified the conserved His as a catalytic base. For GO, CDH and PNO, because mutation of the conserved His resulted in a large deleterious effect on substrate oxidation, and because there are no other basic residues available near the substrate-binding site, the conserved His is also thought to be a catalytic base. For CO, the results suggest that the His in the conserved His/Asn pair is not the main catalytic base, because mutation of His only causes a moderate decrease in enzyme activity. In addition, protonation of His may be required for stabilization of the putative alkoxide intermediate. For ChO, the atomic resolution X-ray structure indicates that the lone pair of His is located 7.7 Å away from the –OH group of the steroid dehydroisoandrosterone substrate, leading the investigators to conclude that the conserved His might not be the catalytic base. For P2O and CO, the data also imply the stepwise formation of an alkoxide species before the hydride transfer step, whereas for AAO these two steps are concerted.
Although the current reaction mechanisms of these enzymes are consistent with hydride transfer mechanisms, details of the involvement of active site residues in catalysis are not fully understood, and seem to be different among different enzymes. Future studies to gain an in-depth understanding of active site residues in the catalysis should be helpful for rational engineering of the GMC enzymes to make them more suitable for biotechnological or biocatalytic applications, e.g. to broaden substrate specificity or to increase reactivity towards particular compounds. The discovery of new GMC enzymes and investigation of their structures and mechanisms should also be helpful to broaden our knowledge and to provide new perspectives for biocatalysis. Finally, the mystery of why P2O, but not other enzymes, can stabilize C4a-hydroperoxyflavin remains to be explored.
This work was supported by The Thailand Research Fund through grant No. BRG5480001 and grants from the Faculty of Science, Mahidol University (to P. Chaiyen) and the National Science and Technology Development Agency (to T. Wongnate).