O2 Reactivity of Flavoproteins

Molecular dynamics simulations and implicit ligand sampling methods have identified trajectories and sites of high affinity for O2 in the protein framework of the flavoprotein d-amino-acid oxidase (DAAO). A specific dynamic channel for the diffusion of O2 leads from solvent to the flavin Si-side (amino acid substrate and product bind on the Re-side). Based on this, amino acids that flank the putative O2 high affinity sites have been exchanged with bulky residues to introduce steric constraints. In G52V DAAO, the valine side chain occupies the site that in wild-type DAAO has the highest O2 affinity. In this variant, the reactivity of the reduced enzyme with O2 is decreased ≥100-fold and the turnover number ≈1000-fold thus verifying the concept. In addition, the simulations have identified a chain of three water molecules that might serve in relaying a H+ from the product imino acid =NH2+ group bound on the flavin Re-side to the developing peroxide on the Si-side. This function would be comparable with that of a similarly located histidine in the flavoprotein glucose oxidase.

The early proposal by Lakowicz and Weber (1) that O 2 diffuses just about freely through proteins has forged the thinking of biochemists for decades. However, evidence is accumulating that O 2 access is guided and controlled. It thus appears that our understanding of the mechanisms of O 2 migration to catalytic centers is currently experiencing a shift of paradigms. Many of the early works on O 2 migration between affinity sites concern myoglobin (2-5) and enzymes from the respiratory chain (6,7). Studies on lipoxygenase-1 dealt with the effect of steric perturbations on the reaction with O 2 and suggested a distinct O 2 pathway (8,9). Later, the role of O 2 channels in the stereochemistry of the 15-lipoxygenase reaction was addressed using a combination of computational and experimental methods (10). A similar ap-proach led to the identification of O 2 access pathways in copper amine oxidase (11).
In monooxygenases and oxidases from the flavoprotein family, O 2 (re)oxidizes the reduced flavin cofactor. Although there is agreement on the physical mechanisms of the initial step(s) of electron transfer from the flavin to O 2 (12)(13)(14), key questions regarding the control of O 2 reactivity with the reduced flavin cofactor are still open. There are examples for channels in flavoproteins that guide O 2 to the reaction site (15)(16)(17). In this context, several questions emerge. To what extent are these channels controlling O 2 reactivity? Do they have a relevant O 2 affinity? What role do H 2 O molecules play? This study addresses these issues using D-amino-acid oxidase (EC 1.4.3.3, DAAO), 4 a prominent member of the flavoprotein oxidase family that has played a central role in mechanistic studies of this enzyme class (18,19). DAAO catalyzes the oxidative deamination of D-amino acids to yield ␣-keto acids, ammonia, and hydrogen peroxide.
In DAAO, the catalytic center is located at the end of a funnel serving as entrance for the substrate D-amino acid. In this channel, the cofactor flavin is shielded by the bound D-amino acid substrate or by the product imino acid from access of small molecules originating from the solvent region (18,19). Because catalysis by DAAO with the best substrates follows a ternary complex mechanism (20), it is unlikely that O 2 accesses the reduced flavin via the substrate channel. Inspecting DAAO three-dimensional structures (18,19), putative channels can be envisaged that would connect solvent and the flavin pocket (see supplemental Fig. 1). What makes a good O 2 access channel? Ideally, it is a continuous region between solvent and the active site where the average probability of finding O 2 is high. This probability depends on many factors like the dynamics of protein side chains and H 2 O molecules or the hydrophobicity of the environment.
Implicit ligand sampling (ILS) is an efficient method, based on molecular dynamics (MD) simulations, to compute the probability density for small, weakly interacting particles such as O 2 at all positions in a molecular system (4). From this probability, useful quantities can be calculated as follows: (i) occupancy, the probability to find a particle in a certain volume region; (ii) Gibbs free energy costs ⌬G(O 2 ) of relocating O 2 from a reference state (e.g. solvent) into a certain region; and (iii) affinity for O 2 for a given region expressed as an apparent dissociation constant K d,app .
Our approach combines the strengths of computer simulations and biochemical experiments. Using ILS, we predict high affinity regions for O 2 inside the protein and in the vicinity of the flavin and extract the most likely diffusion pathways. To verify these predictions, the protein side chains expected to interfere sterically with the high affinity sites were subjected to mutagenesis, and their kinetic parameters are compared with those of the parent DAAO. In addition, the simulations have uncovered a H 2 O chain that is proposed to play a key role in the chemical reaction of O 2 .

EXPERIMENTAL PROCEDURES
Computational Procedures-The sites of high O 2 affinity were determined using the ILS method (4). The underlying MD trajectories were based on the model of DAAO from Rhodotorula gracilis built from the three-dimensional structure 1c0p in which the ligand D-Ala was replaced by IP (19). Flavin is in the reduced anionic state (FAD red H Ϫ ), i.e. the species that reacts with O 2 (20). In the MD simulations, the protein is solvated in physiological NaCl solution, and periodic boundary conditions were applied; no explicit O 2 is present in the system. The ILS calculations were based on 26-ns MD runs after a 4-ns equilibration period. Before applying the ILS procedure, the protein in the 26,000 frames of each trajectory had to be aligned. To improve the alignment, the ILS free energy maps were generated for each monomer P1 and P2 of dimeric yeast DAAO independently. The two maps compare well and were combined into one, thereby doubling the number of samples. To compute the free energy maps, an oxygen probe molecule was placed in 20 different orientations on each position on a fine grid with grid points 0.33 Å apart. The results of the ILS calculation of each frame were averaged, and the resulting free energy grid was down-sampled to 1 Å resolution. All MD simulations were performed using the program NAMD (21). The ILS runs and all molecular graphics were carried out with VMD (22).
Biochemical Procedures-Site-saturation mutagenesis at position 52 or 201 was carried out using the DAAO cDNA subcloned into the pT7-HisDAAO plasmid as template; Ϸ280 clones were screened using D-Ala as substrate and at low (2.5% ϭ 30 M) O 2 saturation (23). The G52C/G52T/G52V DAAO variants were identified based on their limited activity in the screening procedure compared with WT DAAO; the clones encoding for T201L/T201V variants were identified based on their enhanced activity compared with WT DAAO. Production and purification of the variants were according to Ref. 23. Purification yields were Ն65%; purity was Ͼ90% by SDS-PAGE analysis. Activity was determined using an oxygen consumption assay at pH 8.5, 25°C, with 28 mM D-Ala and 0.253 mM dioxygen.
Spectral experiments were carried out at 15°C in 50 mM potassium phosphate, pH 7.5, containing 2 mM EDTA, 10% glycerol, and 5 mM 2-mercaptoethanol. Semiquinone forms of Ϸ10 M DAAO were produced by light irradiation under anaerobic conditions containing 5 mM EDTA and 0.5 M 5-dea-zaflavin. The thermodynamic stability of the semiquinone forms and reduction potentials of DAAO variants were evaluated according to Ref. 24

DAAO Has Two High Affinity Sites for O 2 Close to the Flavin-
The ILS calculations take the three-dimensional structure of the DAAO-imino pyruvate (IP) complex (19) as a starting point. They render the three-dimensional distribution of the free energy potential ⌬G(O 2 ) of O 2 relative to its potential in the bulk solvent. The lower the free energy for a given point in space, the higher is the affinity for O 2 and therefore the average O 2 occupancy. Fig. 1A shows ⌬G(O 2 ) in the vicinity of the active site in the form of an isoenergy surface. On the Si-side of flavin, there are two high affinity regions: site A is adjacent to the N(5)-C(4a) locus, whereas site B is close to the xylene edge of the flavin (Fig. 1A). Both sites correspond to small voids in the crystal structure with enough volume to hold a molecule such as O 2 or H 2 O. Remarkably, there is no high affinity site on the substrate side (Re) in direct contact with flavin, the closest one being found behind the IP product at Ϸ9 Å (site D, Fig. 1A); it has less affinity than the ones on the Si-side. It is conceivable that the IP product with its hydration shell plugs the entrance and effectively shields the isoalloxazine Re-side.
Several Paths Lead O 2 to the Active Center-The three-dimensional structures of DAAO from yeast (19), human (26), and pig (27) have in common a narrow water-filled channel connecting the solvent and the Si-side of the flavin (see Figs. 1 and 2 and supplemental Fig. 1). In yeast DAAO, the entrance is flanked by Val-41, Ser-42, Pro-140, Leu-203, and Arg-227 and Arg-279 (Fig. 1C). To assess if this channel could serve as access path for O 2 , we extracted from the ILS free energy map the minimum energy paths connecting the O 2 high affinity sites with the solvent (11). These are displayed in Fig. 1A with the respective energy profiles in Fig. 1B. The green path leading from the bulk solvent to site A corresponds to the one inferred from the three-dimensional structures. Three branches origi-nate from affinity regions at the protein surface and converge in a free energy minimum just under the protein surface (site C) before reaching site A. The access to site B initially coincides with the green path, but at site C it branches off (blue path). For the alternative red path to site A, there is no crystallographic indication. Among the different ways, the green route to site A implies the lowest energy barriers corresponding to a high probability of passage, whereas the blue and red paths are energetically less favorable. Passage from site D to A entails surmounting a rather high barrier of 18 kJ/mol therefore making a transgression from the flavin Re-to the Si-side unlikely. Moreover, site D corresponds to the location of the side chain of larger amino acid substrates (19).
The energy level along the access channel toward site A, including the barriers, is below the solvent level, whereas the path toward site B exhibits maxima that are insignificantly higher than solvent level, i.e. inside the channel, particularly at sites A and B in the vicinity of flavin, the virtual [O 2 ] is substantially higher than that in an equivalent solvent volume thus speeding up the flavin reoxidation process. The ⌬G(O 2 ) of sites A and B (Ϫ15.3 and Ϫ14.6 kJ/mol) correspond to apparent dissociation constants K d,app Ϸ2.2 and Ϸ2.8 mM, respectively, in fairly good agreement with known kinetic parameters for yeast DAAO (19,20).
Note that the movement along the pathways should be understood as guided, reversible diffusion inside the channel system; consequently, all sites along the pathways may be considered connected. Because of the low barriers and small distances, the diffusion time between any two places inside the channel system can be assumed much shorter than the mean time for the access of another O 2 , the entire system contributes A, section through the protein (gray surface representation) with flavin cofactor and product IP in the binding pocket. The substrate entrance is on the right; the putative O 2 access channels are on the left. Areas that are considered to be part of the solvent have a light blue tint and dashed outline. Regions of low free energy potential ⌬G(O 2 ) denote high affinity sites for O 2 and are shown in the form of the isosurface at Ϫ3 kJ/mol (transparent blobs in green, red, blue, and yellow). There are two high affinity sites for O 2 close to the flavin on the Si-side; site A is Ϸ3.5 Å from the N(5)-C(4a) FAD locus; site B is Ϸ5 Å from the flavin benzene ring. Site D is further away from flavin (Ϸ9 Å) on the Re-side and is part of the solvent phase at the substrate entrance. Minimum energy paths between the different sites and the solvent are displayed as tubular structures whose thickness scales with ⌬G(O 2 ). The pathway connecting site A with the solvent via site C is shown in green, and the one connecting sites B and C in blue. At the protein surface, the path branches out into several catchment areas for O 2 . There is an alternative path (red) to site A on the Si-side with three different entrances. The clipping plane through the protein is rendered transparent in an oval region to show the red path where it dips behind the surface. B, free energy profiles along the different O 2 channels. The colors and the location of the flavin plane correspond to the ones in A. The  Ϫ11 kJ/mol beginning from the outermost surface) using the same color codes as in Fig. 1, A and B. Four H 2 O molecules around the complex may serve as a H ϩ relay chain between the Re-and Si-sides as described in the text and in Fig. 5. The H-bond network is shown in pink. W1 corresponds to a H 2 O seen in the crystal structure (19).
to the increase of the O 2 "effective concentration" at the reaction center.
Protein Dynamics Reveals a Proton Relay System-Comparing the x-ray structure (100 K) with the MD trajectories (298 K) yields interesting clues. The putative O 2 channel on the flavin Si-side is flexible, and the H 2 O molecules inside are not fixed but in constant exchange with the bulk water phase. No H 2 O is observed at site A in any of the four different DAAO x-ray structures (18,19,26,27) Fig. 1B).
Obstructing O 2 Access, the G52X DAAO Variants-The O 2 access hypothesis was tested computationally as well as experimentally. Based on the three-dimensional structure (19), it can be predicted that mutation of the residue Gly-52 would interfere with the high affinity site A (Fig. 3) by partially filling the space expected to be occupied by O 2 in a complex prior to electron transfer. G52C/G52V/G52T DAAO variants were prepared by site-saturation mutagenesis, and their biochemical properties were investigated with particular focus on O 2 reactivity (see supplemental material). The effect of the G52V substitution was assessed computationally. Biochemical Properties of G52X DAAO Variants-The general properties of the G52X variants are very similar to those of WT DAAO (see supplemental material). From this, we conclude that there are negligible structural differences in the immediate flavin environment. A difference between WT and G52X DAAO variants is in the midpoint redox potential (E m ) that is lowered from Ϫ109 mV (WT) to Ϫ185 mV in G52V DAAO. Note that a decrease of E m by Ϸ80 mV is expected to increase the rate of reaction with O 2 by at least one order of magnitude (28), an effect that would thus be opposite to those resulting from the steric factors.
The substitution of Gly-52 significantly affects the steady state kinetic parameters in that, compared with WT DAAO, k cat , K m,D-Ala , and K m,O2 , are 1000-, 70-, and 20-fold lower (supplemental Table 1). The catalytic equations of DAAO consist of two half-reactions as shown in Scheme 1 (20,29), where a) is the reductive half-reaction (substrate dehydrogenation ϩ flavin reduction) and b) is the oxidative half-reaction (O 2 reaction with the reduced enzyme flavin). To verify that in the G52X DAAOs the observed consequences of the exchange are due to E-Fl red + O 2 E-Fl ox k 6 SCHEME 1. Minimal kinetic scheme for the catalytic cycle of DAAO with D-Ala as substrate (sequential mechanism). S, amino acid substrate; P, imino acid product that dissociates in solvent; ox/red, oxidation state of the flavin (20,29). In parentheses are reported the kinetic steps related to the alternative ping-pong mechanism, as observed only for basic D-amino acids (20,29). the reactivity of reduced flavin with O 2 , we have compared the rates of the half-reactions with those of WT DAAO. Thus, the rate of flavin reduction by D-Ala, the best substrate (20), is 2-fold faster for G52V DAAO compared with WT DAAO (Fig.  4A and supplemental material). K d and the rate of product dissociation (Scheme 1a, k 5 ) are identical (supplemental Table 1). Because the chemical steps underlying substrate dehydrogenation are likely the chemically "difficult" ones, we assume that the corresponding catalytic machinery is essentially unaltered in DAAO following substitution of Gly-52.
Reactivity of Free, Reduced G52V DAOO with O 2 Is Lowered ≈100-Fold-In contrast to the reductive half-reaction, the reactivity of the DAAO variants in the oxidative half-reaction is drastically lowered (Fig. 4B). The profile that represents the reaction of the free reduced variant with O 2 (k 6 ) reflects a Ն100-fold lower rate (Table 1, Fig. 4B, and supplemental Fig. 6). The rate of the reaction of the reduced enzyme-IP complex with O 2 for G52V DAAO could not be assessed directly because of the unfavorable equilibria for the formation of the complex itself (see supplemental material). However, a lower limit for k 6 corresponding to an Ϸ100-fold decrease compared with WT DAAO was obtained by simulation of steady state measurements (see supplemental material). Because the parameters for the reductive half-reaction are essentially the same for WT and G52V DAAO, the Ϸ1000-fold decrease in k cat should result from effects on parameters connected with the oxidative halfreaction (Table 1). Because similar catalytic parameters have been determined for the three G52X DAAO variants (supplemental Table 1), we assume that the decrease of O 2 reactivity is not due to a specific effect of the valine isopropyl side chain. The drastic difference in catalytic properties of G52V versus WT DAAO is perhaps best illustrated by the data in Fig. 4C. In this, the absorbance at 455 nm (ordinate) reflects the relative concentrations of oxidized and reduced enzyme species present during turnover. The absorbance value at a given point during turnover also corresponds to the net ratio of the rates of steps involved in enzyme reduction and enzyme (re)oxidation under the specific conditions at this point in time. In these experiments, the enzyme is present in the oxidized form at the very beginning of the reaction, and its absorbance is indicated by Start in Fig. 4C. Upon mixing WT DAAO with D-Ala in the presence of O 2 , there is a very rapid absorbance decrease by some 10% that occurs in the first Ϸ10 ms of the reaction (data not shown, difference between "Start" absorbance and initial absorbance data points). Then the enzyme enters a stationary phase (turnover) up to Ϸ5 s and subsequently gradually returns to the original absorbance value within 20 s. Of importance in the present context is the observation that during turnover WT DAAO is present mainly (Ϸ90%) in its oxidized form. Consequently, for WT DAAO, the net velocity of the reductive half-reaction is Ϸ1/10 that of the oxidative half-reaction. In sharp contrast to this, for G52V DAAO essentially total conversion to the reduced form occurs within 1-2 s (Fig. 4C). Then the enzyme remains reduced until consumption of the reductive substrate D-Ala and subsequently returns to the oxidized state within 100 s (Fig. 4C). For G52V DAAO, the ratio of the net velocities for the reductive and oxidative half-reactions is Ն100 (see also supplemental material). The kinetic constants that were extracted from these experiments (Table 1 and supplemental Fig. 7) are consistent with a catalytic mechanism for the G52V variant in which the decrease of k cat compared with WT DAAO . The inset to C shows the data points for WT DAAO at an expanded absorbance scale.

TABLE 1 Steady state and pre-steady state kinetic parameters of WT and G52V DAAOs
Steady-state kinetic parameters were determined with the enzyme-monitored turnover method (25) at 15°C. The rate constants refer to those defined in Scheme 1.  Fig. 7).

Oxygen Diffusion in D-Amino Acid Oxidase
AUGUST 6, 2010 • VOLUME 285 • NUMBER 32 results largely from the decrease of the rate of reoxidation of the reduced enzyme product complex with O 2 (k 3 in Scheme 1). In a similar approach used to test the role of site B, residue Thr-201 was exchanged into Leu and Val. In contrast to the G52X variants, however, the overall catalytic properties of the T201L/ T201V DAAOs are not significantly different from those of WT enzyme (supplemental Table 2). Fig. 4C (see also supplemental Figs. 4 and 7), it is apparent that during turnover the DAAO variants are largely present in the reduced form, whereas WT DAAO is mainly in the oxidized form during the course of the same experiment. These observations and the evidence from the rapid reaction studies indicate that the major effect resulting from the substitution of Gly-52 in yeast DAAO leads to an alteration of the rate(s) of reaction of the reduced enzyme flavin with O 2 . The validity of the estimated rate constants and of the overall kinetic mechanism was tested by simulation of the time courses of flavin absorbance changes during enzyme-monitored turnover experiments. For this, the sequences of kinetic steps of Scheme 1 and the experimentally employed D-Ala and oxygen concentrations were used. An example of the quality of the simulation procedure is given in supplemental Fig. 7 for G52V DAAO. From this analysis, values for k 3 ϭ 2.4 Ϯ 0.8 ϫ 10 3 M Ϫ1 s Ϫ1 and k 6 Ϸ 300 M Ϫ1 s Ϫ1 were estimated. This confirms that the main kinetic change following substitution of Gly-52 in DAAO is related to a Ϸ100-fold decrease in rate constants involved in the reoxidation of reduced flavin by O 2 , k 3 , and k 6 in Fig. 4. A and B (Fig.  1) lie close to the flavin on the Si-side. Site A is in contact with the isoalloxazine system, with its center being Ϸ3.5 Å from the flavin C(4a) position. Because the highest density of the reduced flavin negative charge is at positions C(4a)-N(5) (30), site A would constitute an ideal location for efficient O 2 reactivity. However, the negative charge is delocalized over the whole isoalloxazine system, with a substantial fraction being found in the C(7)-C(8) area (31). Site B at Ϸ5 Å from the flavin xylene edge would thus also be suitable for electron transfer. The relative importance of sites A and B can be inferred from the mutagenesis results. Obstruction of site A as in G52X DAAOs reduces the O 2 reactivity up to 100-fold, which sets an upper limit for a chemical reactivity via site B in the low percent range compared with site A. This is supported by the absence of major kinetic effects upon the insertion of a sterically demanding substituent in site B (T201L DAAO). On the other hand, sites A and B are close (Ϸ8 Å); they are connected via site C, and intertransfer of O 2 proceeds through a low activation energy barrier (Fig. 1B) The additional sites of O 2 affinity (Fig. 1A) could also play relevant roles. Site C is located just under the protein surface at the entrance of the channel leading to the high affinity sites A and B. Together with the affinity sites at the protein surface, it constitutes a region where O 2 is "collected/concentrated" for further transport and utilization (catchment area). The energetic profiles of the various paths of O 2 between solvent and the sites of high affinity in the vicinity of the flavin (Fig. 1B) emphasize that once O 2 has entered the protein frame, it can move just about freely and get to the point of chemical reactivity with a high frequency; the O 2 storage functionality can be ascribed to the entire channel system. Furthermore, storage should not be understood in the sense that a single O 2 molecule finds its way into the channel and is kept there until the reaction occurs. Instead, the O 2 equilibration between channel and solvent is much faster than the reaction time. Hence, many O 2 molecules enter and leave the channel before eventually one of them participates in the reaction. The affinity of the channel system increases the probability for each O 2 to stay close to flavin thus enhancing its bimolecular reaction term.

Implications for the Reaction Mechanism-Sites
An unexpected finding is the uncovering of a chain of ordered H 2 O molecules that could represent a H ϩ relay system. Its importance becomes evident in view of a key conclusion from the work of Klinman and co-workers (32) on the reactivity of (reduced) glucose oxidase with O 2 . This is best described in their statement: "…a protonated active site histidine at low pH accelerates the second-order rate constant for one electron transfer to dioxygen through electrostatic stabilization of the superoxide anion intermediate." One difference between the active sites of DAAO and glucose oxidase is that in the former there is no amino acid residue at the active site or its vicinity that might have a role analogous to that of the His in glucose oxidase (32). This role is proposed to be exerted by the H ϩ relay chain and to consist in the transfer of a H ϩ from the ϭNH 2 ϩ function of bound IP to the developing, negatively charged O 2 species. This H ϩ relay is shown in Figs. 2 and 5.
This interpretation also provides an answer to the long standing puzzle that the reactivity of the reduced DAAO-IP complex with O 2 is substantially higher than that of the free form (Scheme 1). The opposite would be expected because in the free enzyme there is no apparent hindrance in O 2 access. An important feature of the mechanism shown in Fig. 5 is its rearrangement and balance of charges. In the E-Fl red -P complex (Fig. 5, left), the opposed charges on reduced flavin and IP are an important thermodynamic factor for complex stability, whereas the negative charge on the reduced flavin is required to enhance reactivity with O 2 (32,33). Upon transfer of electrons and H ϩ to O 2 , the resulting complex (Fig. 5, right) is now neutral. Consequently, the rate of product dissociation is enhanced providing a molecular rationale for the sequential kinetic mechanism of DAAO (and related enzymes) in turnover with neutral D-amino acids (19,29).
Because H 2 O 2 is only somewhat larger than O 2 or H 2 O, the O 2 channel and the H ϩ relay channel might in principle also serve for its release in a way similar to the "in-out" diffusion of H 2 O. Furthermore, if a path via a Si 3 Re migration and via the H ϩ channel would be feasible, once H 2 O 2 is on the Re-side of the flavin, it would likely leave through the substrate channel following IP release. In the three-dimensional structure of DAAO, a small molecule on the flavin Re-side near the N(5)-C(4a) position was detected at low occupancy (19). This observation could be reconciled with the presence of H 2 O 2 that might have become trapped at that position before leaving the active center via the substrate entrance.
Comparison within the Flavoprotein Oxidase Family-For an efficient activation of O 2 , the reduced flavin ought to be in its anionic form (14,34,35), a condition found in most oxidases. A second prerequisite is the presence of a system that provides a H ϩ that neutralizes the developing O 2 . /HOO Ϫ . This role has been assigned to a His in glucose oxidase (12,32) and to a Lys in monomeric sarcosine oxidase (36) and in monoamine oxidase (37,38). In DAAO, this would be implemented by the proposed H ϩ relay system. Intriguingly, in L-amino-acid oxidase a lysine (Lys-326) is linked via a (crystal) H 2 O to the flavin N(5) (supplemental Fig. 2), and it is placed such as to suggest a H ϩ donor role in O 2 activation also on the flavin Si-side. Thus, Dand L-amino-acid oxidase would share the flavin Si-side for O 2 activation; however, they would use two variants of the same theme for H ϩ transfer in the chemical activation of O 2 .
Conclusions-This work demonstrates the great potential of joint computational/experimental research. MD simulations allowed individuating specific paths and positions suitable for the diffusion and reactivity of O 2 and thus the prediction of effects resulting from steric constrictions. Meanwhile, biochemical experiments provided the required verification. The uncovering of steric and chemical factors that affect O 2 reactivity brings up the question of their relative importance. Although a clear-cut differentiation is not possible, we conclude the following.
(i) In free, reduced WT DAAO, O 2 has unrestricted access via both the O 2 and the substrate channels. There is no H ϩ relay chain. The observed O 2 reaction rate should thus be the intrinsic one (step k 6 in Scheme 1).
(ii) In G52V DAAO, there is a steric hindrance to O 2 access and no H ϩ relay chain. The Ϸ100-fold difference of the rates in i and ii should thus be due to steric factors. Note that this decrease might be up to 10-fold higher because the (75 mV) more negative E m in G52V variant compared with WT DAAO partially counteracts the steric effects.
(iii) A lower limit estimate of the effect of the H ϩ relay chain can be obtained by comparing the rates of reoxidation of reduced WT DAAO in the free (no H ϩ relay) and IP complexed state (with H ϩ relay), corresponding to k 6 and k 3 rate constants (Scheme 1). This difference is Ϸ10-fold for yeast DAAO (20).