Probing the Catalytic Mechanism of Copper Amine Oxidase from Arthrobacter globiformis with Halide Ions*

Background: Copper amine oxidases catalyze amine oxidation using copper and a quinone cofactor. Results: Halides bind axially to the copper center, preventing the reduced cofactor from adopting an on-copper conformation. Conclusion: The cofactor undergoes large conformational changes during the catalytic reaction that enable transitions between different types of chemistry. Significance: Molecular details of cofactor movement have been unveiled based on structural and kinetic evidence. The catalytic reaction of copper amine oxidase proceeds through a ping-pong mechanism comprising two half-reactions. In the initial half-reaction, the substrate amine reduces the Tyr-derived cofactor, topa quinone (TPQ), to an aminoresorcinol form (TPQamr) that is in equilibrium with a semiquinone radical (TPQsq) via an intramolecular electron transfer to the active-site copper. We have analyzed this reductive half-reaction in crystals of the copper amine oxidase from Arthrobacter globiformis. Anerobic soaking of the crystals with an amine substrate shifted the equilibrium toward TPQsq in an “on-copper” conformation, in which the 4-OH group ligated axially to the copper center, which was probably reduced to Cu(I). When the crystals were soaked with substrate in the presence of halide ions, which act as uncompetitive and noncompetitive inhibitors with respect to the amine substrate and dioxygen, respectively, the equilibrium in the crystals shifted toward the “off-copper” conformation of TPQamr. The halide ion was bound to the axial position of the copper center, thereby preventing TPQamr from adopting the on-copper conformation. Furthermore, transient kinetic analyses in the presence of viscogen (glycerol) revealed that only the rate constant in the step of TPQamr/TPQsq interconversion is markedly affected by the viscogen, which probably perturbs the conformational change. These findings unequivocally demonstrate that TPQ undergoes large conformational changes during the reductive half-reaction.

The catalytic reaction of copper amine oxidase proceeds through a ping-pong mechanism comprising two half-reactions. In the initial half-reaction, the substrate amine reduces the Tyrderived cofactor, topa quinone (TPQ), to an aminoresorcinol form (TPQ amr ) that is in equilibrium with a semiquinone radical (TPQ sq ) via an intramolecular electron transfer to the active-site copper. We have analyzed this reductive half-reaction in crystals of the copper amine oxidase from Arthrobacter globiformis. Anerobic soaking of the crystals with an amine substrate shifted the equilibrium toward TPQ sq in an "on-copper" conformation, in which the 4-OH group ligated axially to the copper center, which was probably reduced to Cu(I). When the crystals were soaked with substrate in the presence of halide ions, which act as uncompetitive and noncompetitive inhibitors with respect to the amine substrate and dioxygen, respectively, the equilibrium in the crystals shifted toward the "off-copper" conformation of TPQ amr . The halide ion was bound to the axial position of the copper center, thereby preventing TPQ amr from adopting the on-copper conformation. Furthermore, transient kinetic analyses in the presence of viscogen (glycerol) revealed that only the rate constant in the step of TPQ amr /TPQ sq interconversion is markedly affected by the viscogen, which probably perturbs the conformational change. These findings unequivo-cally demonstrate that TPQ undergoes large conformational changes during the reductive half-reaction.
Copper amine oxidases (CAOs 2 ; EC 1.4.3.6) catalyze the oxidative deamination of various primary amines to produce the corresponding aldehydes and ammonia, coupled with the reduction of molecular oxygen to hydrogen peroxide (1)(2)(3). CAOs play distinct physiological roles in prokaryotes and eukaryotes. Prokaryotic CAOs mainly function to assimilate primary amines as carbon and nitrogen sources for growth. Eukaryotic CAOs have versatile functions, being involved in the detoxification of bio-active amines, such as histamine (4); cell adhesion (5); cell death (6); collagen cross-linking in animals (7); and germination, root growth, and healing of wounded cell walls in plants (8). It has been reported that human serum CAOs cause angiopathy in diabetes (9). Therefore, various CAO inhibitors have been developed as therapeutic drugs (10,11).
In the initial reductive half-reaction, the C5 carbonyl group of the oxidized cofactor (TPQ ox ) undergoes nucleophilic attack by amine substrate to form the substrate Schiff base (TPQ ssb ). Stereospecific proton abstraction by a conserved base (Asp-298 in AGAO) (29) converts TPQ ssb to the product Schiff base (TPQ psb ). Concomitantly with the release of the corresponding aldehyde, TPQ psb is hydrolyzed to the reduced cofactor (an aminoresorcinol form, TPQ amr ) that is in equilibrium with a semiquinone radical form (TPQ sq ) via the single electron transfer reduction of Cu(II) to Cu(I). In these steps of the reductive half-reaction, the cofactor ring is not directly ligated to the copper atom (a configuration designated the "off-copper" conformation), whereas in the TPQ sq ⅐Cu(I) state, the cofactor ring is assumed to be directed toward the metal in the so-called "oncopper" conformation to facilitate electron transfer. This suggests that the equilibrium between TPQ amr ⅐Cu(II) and TPQ sq ⅐Cu(I) is accompanied by a substantial conformational reorganization of the TPQ ring. Indeed, the conformational flexibility of TPQ within the active site of CAOs has previously been suggested to be important in catalysis (24,30,31). In the subsequent oxidative half-reaction, the reduced cofactor is reoxidized by dioxygen to produce hydrogen peroxide and an iminoquinone intermediate (TPQ imq ), which is further hydrolyzed to form the oxidized cofactor, releasing ammonia in the following step (1)(2)(3). In the presence of excess substrate, it is assumed that TPQ imq reacts directly with the substrate amine to form TPQ ssb (via a trans-imination reaction) rather than regenerating TPQ ox (Scheme 1).
Depending on the enzyme sources and reaction conditions, such as pH and temperature, various amounts of TPQ sq ⅐Cu(I) are known to be formed by adding an amine substrate under anaerobic conditions, which is assumed to induce the reductive half-reaction. However, the mechanistic role of the TPQ sq ⅐ Cu(I) form in the subsequent O 2 reduction remains unclear and controversial (32)(33)(34)(35)(36)(37)(38)(39). Two reaction pathways for the O 2 reduction have been proposed, depending on the enzyme sources. One is an inner sphere mechanism in which O 2 is coordinated on Cu(I) and reduced by the transfer of two electrons from TPQ sq ⅐Cu(I) to ultimately produce a Cu(II)-bound peroxide species and TPQ imq (Scheme 1). The necessary singlet to triplet spin transition is allowed by the coordination of O 2 to Cu(I). AGAO and P. savitum CAO are CAOs that have been suggested to follow this mechanism (37,38). The formation of the TPQ sq ⅐Cu(I) state at the beginning of the oxidative half-reaction is believed to be essential in this process. The other mechanism that has been proposed is an outer sphere process that is suggested to occur in bovine serum CAO (34) and HPAO-1 (36,39). In both of these enzymes, O 2 binds to a hydrophobic pocket close to the cofactor and is initially reduced by TPQ amr via a single electron transfer that does not change the oxidation state of the Cu(II) center. The resulting superoxide anion then coordinates to Cu(II), inducing the second electron transfer. The TPQ sq ⅐Cu(I) complex is not formed in this catalytic cycle and is assumed to be an off-pathway product generated only by the anaerobic reduction of the CAO by the amine substrate.
We have previously determined x-ray crystal structures of the intermediates, including TPQ ssb and TPQ psb , formed during the reductive half-reaction of AGAO with 2-phenylethylamine (2-PEA) (29), tyramine (40), and ethylamine (41). In each case, the TPQ cofactor in these structures had an off-copper conformation. Here, we present new high resolution structures of TPQ sq formed in the reductive half-reaction of AGAO with 2-PEA and histamine. The TPQ sq formed with these substrates exists exclusively in the on-copper conformation with the 4-OH group ligating axially to the copper center, which is probably in the Cu(I) oxidation state. Moreover, we found that the off-copper TPQ amr is formed when the crystals are soaked with substrate in the presence of halide ions that act as uncompetitive inhibitors with respect to the amine substrate and noncompetitive inhibitors with respect to dioxygen. Halide ions bind axially to the copper center, preventing TPQ amr from coordinating to copper. Combined with the results of spectrophotometric, steady-state, and transient kinetics analyses, the results presented herein provide unequivocal evidence for the occur-SCHEME 1. Presumed catalytic mechanism of AGAO. SEPTEMBER

Experimental Procedures
Materials-Recombinant AGAO was purified as its inactive precursor and converted to the copper-and TPQ-containing active form as reported previously (12,19). Protein and TPQ sq concentrations were determined spectrophotometrically using molar extinction coefficients of ⑀ 280 ϭ 93,200 M Ϫ1 cm Ϫ1 (12) and ⑀ 468 ϭ 4500 M Ϫ1 cm Ϫ1 , respectively (35). All amine substrates used for kinetic analyses and crystal soaking were neutralized with 1 M H 2 SO 4 .
Spectrophotometric Measurements-To achieve fully anaerobic conditions, the enzyme and substrate solutions were kept in a vacuum-type glove box (Iuchi, SGV-65V) filled with 99.999% (v/v) argon gas for at least 2 h, as described previously (18). The enzyme (final concentration, 100 M monomer) was anaerobically mixed with 1 mM 2-PEA in 50 mM HEPES, pH 6.8, in the presence or absence of various concentrations of sodium, potassium, or ammonium salts of halide ions. For measurements of pH dependence, the enzyme (100 M monomer) was reduced with substrate in 100 mM MES (pH 5.7, 5.8, 6.0, 6.3, 6.5, and 6.7), 100 mM HEPES (pH 7.0, 7.3, 7.5, 7.8, and 8.0), 100 mM TAPS (pH 8.5 and 9.0), or 100 mM CHES (pH 9.5 and 10.1) at a nearly constant ionic strength (I ϭ 0.35 Ϯ 0.03) after adjustment with 100 mM Na 2 SO 4 . The enzyme was mixed with its substrate under anaerobic conditions in a quartz cuvette with a gas-tight screw cap, and after 5 min, the absorption spectrum was measured at 25°C with an Agilent 8453 photodiode array spectrophotometer. An apparent pK a value for the absorbance change at 468 nm was determined by fitting the data to Equation 1, where y represents the absorbance at a particular pH, C 1 and C 2 are the pH-independent values of the absorbance, and K a1 and K a2 are the acid dissociation constants associated with the pH profile. Data fitting was performed by nonlinear regression using Kaleidagraph version 4.1 (Abelbeck Software).
Steady-state Kinetic Analysis-Steady-state kinetic analyses were conducted at 30°C with 2-PEA, histamine, or ethylamine (hydrochloride) as the substrate using the colorimetric assay protocol described previously (12). Inhibition by halide ions was studied using assay solutions containing 2-40 M 2-PEA and 0 -50 mM NaF, NaCl, NaBr, or NaI. Inhibition constants (K i ) were determined on the basis of uncompetitive inhibition (42) with respect to the amine substrate using Equation 2, where v 0 represents the initial reaction rate and [S] and [I] are the concentrations of substrate and inhibitor, respectively. Data fitting was performed by multiple regression analysis using R (available from the R Project Web site). Inhibition by halide ions was also studied using assay solutions containing 8.8 -1160 M dissolved oxygen and a fixed concentration (40 M) of 2-PEA in the presence of 0 -25 mM NaCl. Various concentrations of dissolved oxygen were established by mixing 99.99% (v/v) O 2 and 99.99% (v/v) N 2 gas with a gas mixer (KOFLOC, PMG-1) and bubbling the resulting mixture through a needle into the reaction mixture contained in a tightly sealed cuvette with a silicon rubber cap. The mixed gas line was also branched into the cell of a Clark-type oxygen electrode (YSI Inc., model 5300) for determination of the dissolved oxygen concentration. The reaction was initiated by adding a small amount of the anaerobic enzyme solution with a microsyringe. Inhibition constants (K i ) were determined on the basis of noncompetitive inhibition (42) with respect to dissolved oxygen using Equation 3, where v 0 represents the initial reaction rate, and [S] and [I] are the concentrations of dissolved oxygen and inhibitor, respectively. Data fitting was performed by multiple regression analysis using R.
The effect of solvent viscosity was investigated by measuring the enzyme's activity toward 2-PEA at 4°C in a solution of 0 -30% (w/v) glycerol or sucrose as a viscogen; the solvent's viscosity was determined relative to a buffer-only solution with an Ostwald viscometer.
Stopped-flow Measurements-Transient kinetic analyses were done at 4°C with an Applied Photophysics stopped-flow spectrophotometer (40,41). Typically, equal volumes (about 30 l each) of enzyme (200 M monomer in 50 mM HEPES buffer, pH 6.8) and substrate (1 mM 2-PEA) solutions were mixed in a 20-l mixing cell by triggering with an N 2 gas piston; the mixing dead time was generally 2.3 ms at an N 2 gas pressure of 500 kilopascals. To avoid spectral changes associated with the oxidative half-reaction, both enzyme and substrate solutions were maintained under fully anaerobic conditions as described above. The substrate solution was supplemented with 200 mM NaCl, 600 mM NaBr, or 600 mM Na 2 SO 4 , as appropriate. The effect of solvent viscosity on transient kinetics was also studied as described above by adding a viscogen (0 -30% (w/v) glycerol or sucrose) to both the enzyme and substrate solutions. UVvisible absorption spectra were recorded every 2.5 ms at wavelengths of 250 -800 nm. Spectral data were analyzed using Pro-Kineticist II (Applied Photophysics) to obtain the spectra of the reaction intermediates and to calculate the rate constants for each reaction step.
Single-crystal Microspectrophotometry-AGAO crystals prepared as described below were subjected to microspectrophotometry before x-ray diffraction as reported previously (29).
X-ray Crystallographic Analysis-AGAO was crystallized by microdialysis essentially according to the method described previously (29). Briefly, a 15 mg/ml protein solution was dialyzed in a 50-l dialysis button at 16°C against 1.05 M potassium-sodium tartrate in 25 mM HEPES buffer, pH 6.8. After 2 weeks of crystal growth, the dialysis buttons were transferred into a fresh reservoir solution supplemented with 45% (v/v) glycerol as a cryoprotectant, and the crystals were soaked at 16°C for 24 h, followed by further soaking in the fully anaerobic reservoir solution containing 45% (v/v) glycerol for 24 h. The crystals were then incubated in a solution (pH 6.8) containing 4 mM 2-PEA, 10 mM histamine, or 50 mM ethylamine (hydrochlo-ride) with or without 100 mM NaCl or 300 mM NaBr for about 1 h until their color faded. They were then mounted on thin nylon loops (, 0.5-0.7 mm) and frozen by flash cooling in liquid CF 4 . All procedures were done in the anaerobic box, and the frozen crystals were kept in liquid N 2 until x-ray diffraction analysis.
Diffraction data sets were collected at 100 K with synchrotron X-radiation using a DIP6040 imaging plate (Bruker AXS, Billerica, MA) in the BL44XU station or using a Quantum 210 CCD detector (ADSC) in the BL38B1 station at SPring-8 (Hyogo, Japan). X-rays with a wavelength of 0.919 Å were used to detect anomalous peaks derived from bromine; a wavelength of 0.9 Å was used otherwise. The collected data sets were processed and scaled using HKL2000 (43) or MOSFLM (44) and SCALA (45), respectively. The starting model was obtained by molecular replacement with Phaser (46). The search model was based on the coordinates of the AGAO monomer (Protein Data Bank code 1IU7) after removing all water molecules and a metal ion. Refinements, electron density map calculations, and assignment of solvent molecules were initially done using Refmac 5 (47) and later with Phenix (48). Manual rebuilding was performed using Coot (49), and water molecules and other ligands, such as metal ions, were added step by step to the model during the refinement process. The models of the catalytic intermediates of TPQ and phenylacetaldehyde were built using the Monomer library sketcher from the CCP4 package (50), and then the dictionary files used with Refmac 5, Phenix, and Coot were generated using the PRODRG server (51). PyMOL version 1.5 (Schrödinger, LLC) was used for figure drawings. Anomalous maps for bromine atoms were generated using fft (52) from the CCP4 package (50) based on anomalous difference and phase data for the final model. Details and statistics pertaining to the data collection and refinement are summarized in Table 1.

Effect of Halide Ions on the Absorption Spectrum of TPQ sq -It
is well established that the absorption spectrum of the bound cofactor TPQ changes rapidly during the reductive half-reaction of CAOs (1-3), as demonstrated by stopped-flow spectrophotometry of the reaction with AGAO (29,40,53). The final form of TPQ in the reductive half-reaction without turnover (measured under anaerobic conditions) is TPQ sq , which exhibits characteristic absorption maxima at about 440 and 470 nm with a shoulder at about 350 nm. We incidentally noted that the TPQ sq spectrum of AGAO was markedly depressed in the presence of NaCl, which was added to the reaction mixture to maintain a constant ionic strength. For example, in the presence of 10 mM NaCl, the intensities of the absorption bands at 352, 438, and 468 nm decreased to about 32% of those without NaCl (Fig.  1). When calculated using the molar extinction coefficient at 468 nm (⑀ 468 ϭ 4500 M Ϫ1 cm Ϫ1 ) reported for TPQ sq (35), the presence of NaCl reduced the calculated TPQ content relative to total AGAO from 38 to 9.8%. This was attributed to a shift of the equilibrium between TPQ amr and TPQ sq (see Scheme 1) toward the former, which has no absorption bands above 300 nm (54). To identify the chemical species that caused this equilibrium shift, we investigated the effects of adding 10 mM NH 4 Cl, KCl, Na 2 SO 4 , or NaH 2 PO 4 in the reductive half-reaction and found that comparable shifts were induced by NH 4 Cl and KCl but not Na 2 SO 4 or NaH 2 PO 4 ( Fig. 1), clearly indicating that the Cl Ϫ ion was preventing the formation of TPQ sq . Other halide ions (F Ϫ , Br Ϫ , and I Ϫ ) behaved similarly (Fig. 2), and their effects were roughly concentration-dependent, with an order of effectiveness of Cl Ϫ Ϸ Br Ϫ Ͼ F Ϫ Ͼ I Ϫ . A similar effect (bleaching of TPQ sq absorption) by a high concentration of Cl Ϫ (0.45 M KCl) was reported in an early study on methylamine oxidase from Arthrobacter P1 (55), but the mechanism of bleaching was not further pursued.
Effect of Halide Ions on Catalytic Activity-We have briefly reported the inhibition of AGAO activity by Cl Ϫ ion (56) in the past. In this work, more precise steady-state kinetic analyses were conducted by systematically varying the concentrations of the substrate amine (2-PEA) or dissolved oxygen in the presence of 0 -50 mM solutions of different halide ions. Double reciprocal plots (1/v versus 1/s) revealed that the Cl Ϫ ion was an uncompetitive inhibitor with respect to the substrate amine and a noncompetitive inhibitor with respect to dissolved oxygen (Fig. 3); other halide ions (F Ϫ , Br Ϫ , and I Ϫ ) exhibited similar inhibition patterns (data not shown). These results show that halide ions bind to a substrate⅐enzyme complex but not to the free enzyme and that they bind equally to the O 2 -bound and O 2 -unbound enzyme forms at a site distinct from the O 2 -binding site (41). The K i values calculated for uncompetitive inhibition were 97.2 Ϯ 6.5 mM for F Ϫ , 26.2 Ϯ 1.4 mM for Cl Ϫ , 58.2 Ϯ 2.8 mM for Br Ϫ , and 69.6 Ϯ 9.4 mM for I Ϫ , whereas that for noncompetitive inhibition by Cl Ϫ was 32.8 Ϯ 1.7 mM. The inhibition of catalytic activity by halide ions may be unique to AGAO; the Cl Ϫ ion reportedly has no inhibitory effect on the activities of human kidney diamine oxidase, P. pastoris amine oxidase, or P. savitum CAO (57).
Effects of Halide Ions on Transient Kinetics-To identify the substrate⅐enzyme complex to which the halides bind, we performed transient kinetic analyses of the reductive half-reaction in the presence of 100 mM NaCl, 300 mM NaBr, or 300 mM Na 2 SO 4 (as a control for the ionic strength). As reported previously (29,40,53), rapid spectral changes associated with changes in the redox and chemical state of TPQ (TPQ ox 3 TPQ ssb 3 TPQ psb 3 TPQ amr 3 TPQ sq ) were observed (Fig. 4). We noted that in the presence of 100 mM NaCl or 300 mM NaBr, the TPQ sq absorption band appeared transiently within ϳ117 ms and then gradually declined in intensity to about 30 -40% of the value achieved in the absence of halide ions or in the presence of 300 mM Na 2 SO 4 (Fig. 4, A and B). The final spectra (at 1023 ms) acquired in the presence of NaCl or NaBr (Fig. 4, C and D) mostly lacked the TPQ sq -characteristic peaks, suggesting a shift of the equilibrium toward TPQ amr , as discussed above (Fig. 2, B and C).
Initially, the multiwavelength data of all spectral changes were fitted to the four-step mechanism connecting TPQ ox , TPQ ssb , TPQ psb , TPQ amr , and TPQ sq (Scheme 1) by global analysis as reported previously (29). The spectral changes in the presence of 300 mM Na 2 SO 4 ( Fig. 4B) were solved to provide rate constants that were essentially identical to those obtained without the salt (Fig. 4A) (Table 2), and the deduced UV-visible absorption spectra (Fig. 4, E and F) of TPQ ox , TPQ ssb , TPQ psb , d R free is an R factor of the refinement evaluated for 5% of reflections that were excluded from the refinement. TPQ amr , and TPQ sq were similar to those reported previously (29). However, the spectral changes observed in the presence of 100 mM NaCl or 300 mM NaBr could not be fitted to the fourstep model. On the basis of the spectral changes and inhibition mechanism described above, it appeared that halide ions (X Ϫ ) bound to the TPQ amr state. We therefore proposed a branched model (with the branch connecting TPQ ox to TPQ amr ⅐X Ϫ ; Scheme 1), in which the TPQ amr ⅐X Ϫ complex accumulates. The new model provided a reasonable solution to the data fitting of the spectral changes in the presence of NaCl or NaBr. As shown in Table 2, the rate constants of the steps between TPQ ox and TPQ sq (k Ϯ1 , k Ϯ2 , k Ϯ3 , and k Ϯ4 ) were comparable with those observed without halide ions, although the k Ϫ3 value was approximately halved. The rate constant of the branching step from TPQ amr to TPQ amr ⅐X Ϫ (k ϩ5 ) (Scheme 1) was estimated to be half that for TPQ sq formation (k ϩ4 ). The magnitude of these parameters well explains the slow accumulation of TPQ amr ⅐X Ϫ ; TPQ sq is formed initially but gradually converted to TPQ amr ⅐X Ϫ via TPQ amr . The deduced absorption spectra of TPQ ox , TPQ ssb , TPQ psb , TPQ amr , and TPQ sq in the presence of NaCl (Fig. 4G) and NaBr (Fig. 4H) were essentially identical to those observed without halide ions (Fig. 4E) and in the presence of 300 mM Na 2 SO 4 (Fig. 4F). Altogether, these results show that halide ions bind to the TPQ amr form in the reductive halfreaction, thereby inhibiting the formation of TPQ sq from TPQ amr .
Effect of Solvent Viscosity on AGAO Activity-To further probe the equilibrium shift between TPQ amr and TPQ sq , we examined the dependence of catalytic activity on solvent viscosity, which can perturb diffusion-controlled steps, including substrate binding, product release, and conformational changes of the enzyme (58,59). Glycerol (M r ϭ 92.1) and sucrose (M r ϭ 342.3) were used as viscogens. Steady-state kinetic analyses in viscogenic solutions, in which the relative solvent viscosity (/ 0 , where and 0 denote viscosities in the presence and absence of the viscogen, respectively) was increased by adding glycerol or sucrose, yielded values of k cat /K m for 2-PEA that were basically identical to that observed in non-viscogenic solution, whereas the k cat values decreased. The ratio of the viscogen-free k cat value to that in the presence of viscogen (k cat 0 /k cat , where k cat 0 and k cat denote the rate constants in the absence and presence of the viscogen, respectively) was roughly proportional to the relative solvent viscosity (Fig.  5). This suggests that a diffusion-controlled step(s) is included in the overall reaction consisting of TPQ imq , TPQ ssb , TPQ psb , TPQ amr , TPQ sq , and TPQ imq under steady-state conditions (Scheme 1).
To evaluate the effect of solvent viscosity on each step of the reductive half-reaction, transient kinetic analyses were conducted (Fig. 6). Stopped-flow measurements in the absence and presence of 30% (w/v) glycerol generally produced similar spectral changes, but the formation of TPQ sq was clearly slower in the viscogen's presence (Fig. 6, A and B, insets). Slow TPQ sq formation was evident from the comparison of the traces of absorbance changes at 468 nm specific to TPQ sq (Fig. 6C). Further, the rate constants of each step (k ϩ1 , k ϩ2 , k ϩ3 , k Ϫ1 , k Ϫ2 , and k Ϫ3 ) in the reductive half-reaction were determined by global   analysis according to the four-step model starting from TPQ ox and ending at TPQ sq (Scheme 1). Of these rate constants, k ϩ1 , k ϩ2 , k ϩ3 , k Ϫ1 , k Ϫ2 , and k Ϫ3 were independent of the glycerol concentration, but the values of k ϩ4 and k Ϫ4 (which relate to the interconversion of TPQ amr and TPQ sq ) decreased in proportion to the glycerol concentration ( Table 2), showing that this step is diffusion-controlled. For detailed analysis, the ratios of k 0 /k were plotted against the relative solvent viscosity, / 0 (Fig. 6D), where k 0 and k are the rate constants (k ϩ2 , k Ϫ2 , k ϩ4 , and k Ϫ4 ) in the absence and presence of viscogen (glycerol), respectively. The ratios of k ϩ4 0 /k ϩ4 and k Ϫ4 0 /k Ϫ4 increased significantly in proportion to the relative solvent viscosity, although the slope of the fitted line (0.50) indicated a partial effect; an entirely diffusion-controlled reaction would give a slope of 1.0 (60) (Fig. 6D). The ratios of the other rate constants were almost independent of the relative solvent viscosity (e.g.   slope ϭ 0.003 for k ϩ2 0 /k ϩ2 ) (Fig. 6D). The reaction step connecting TPQ amr and TPQ sq involves neither substrate binding nor product release. Thus, these findings show that the interconversion of TPQ amr and TPQ sq is accompanied by a conformational change(s) of the enzyme. Reducing the viscogen's access to the region undergoing the conformational change could reduce its effective concentration, explaining the partial effect indicated by the k 0 /k slope of ϳ0.5. It is therefore likely that the conformational change(s) occurs somewhere in the protein interior, such as the buried active site (most likely in the TPQ cofactor itself). Supporting this, we were unable to observe any effect of solvent viscosity on the rate constants k ϩ4 and k Ϫ4 when using 30% (w/v) sucrose as the viscogen; sucrose's molecular size is 3.7-fold greater than that of glycerol, so it probably cannot penetrate into the active site cavity (Fig. 6D).
Reductive Half-reaction in Crystals-To obtain structural insights into the interconversion between TPQ amr and TPQ sq , the AGAO crystals were reacted anaerobically with substrates in the presence or absence of halide ions under six different conditions (Table 3), and the intermediates formed were freeze-trapped for structural determination. Before x-ray analysis, these crystals were subjected to single-crystal microspec-   SEPTEMBER 18, 2015 • VOLUME 290 • NUMBER 38 trophotometry to identify the reaction intermediates present within them (29,40,41). The absorption spectra of the frozen crystals anaerobically soaked with 4 mM 2-PEA or 10 mM histamine (these crystals were designated AGAO PEA and AGAO HTA , respectively) (Fig. 7, A and B) were very similar to that of TPQ sq for the enzyme in solution, showing specific absorption peaks around 365, 438, and 465 nm (Fig. 1). In contrast, the crystals anaerobically soaked with 50 mM ethylamine (hydrochloride) (AGAO ETA/HCl ) gave a rather peakless spectrum resembling that of TPQ amr (Fig. 7C), although small TPQ sq -like peaks remained. Furthermore, in the presence of 100 mM NaCl or 300 mM NaBr, the crystals prepared by anaerobic soaking with 4 mM 2-PEA (designated AGAO PEA/NaCl and AGAO PEA/NaBr , respectively) exhibited absorption spectra (Fig.  8, A and B) that were distinct from those of TPQ sq and TPQ amr but comparable with that of TPQ psb for the enzyme in solution (Fig. 4E). To identify the halide ion-binding site(s) in the unreacted enzyme, the crystal was aerobically soaked with 300 mM NaBr alone (AGAO NaBr ). This yielded a spectrum identical to that of TPQ ox because the reaction could not proceed (Fig. 8C). These crystal absorption spectra revealed that the reductive halfreaction occurred in all of the crystals except for AGAO NaBr and that the accumulation of the TPQ psb , TPQ amr , or TPQ sq intermediates could be induced by soaking the crystals with appropriate amine substrates in the presence or absence of halides ( Table 3). The overall structures determined for AGAO PEA , AGAO HTA , AGAO ETA/HCl , AGAO PEA/NaCl , AGAO PEA/NaBr , and AGAO NaBr were comparable with that of resting AGAO (Protein Data for residue 382 contoured at 3.5 . Active-site residues are represented by green stick models. Water molecules and copper centers are represented by brown and cyan spheres, respectively. All molecular drawings were generated using PyMOL. Bromine-anomalous maps contoured at 8 are represented by red meshes in the active sites of AGAO PEA/NaBr (D) and AGAO NaBr (F). The anomalous dispersion of the copper atom is pronounced although the wavelength of the used x-ray (0.919 Å) deviated from the peak wavelength of the copper atom (1.3808 Å), and an anomalous peak (about 18 ) was detected on the copper site of the active center as well as for the bromine atom. Active-site residues are represented by green stick models. Water molecules and copper atoms are represented by brown and cyan spheres, respectively. All molecular drawings were generated using PyMOL.

Conformational Change of Topa Quinone in Copper Amine Oxidase
Bank code 1IU7), with root mean square deviations for the main-chain atoms within ϳ0.4 -0.5 Å. In these crystal structures, we assigned the chemical structures of TPQ based on its absorption spectra (Table 3) and constructed the active-site structures using the assigned models of TPQ in different conformations, which were built to coincide with the F o Ϫ F c omit maps (Figs. 7 and 8). The two monomers (chains A and B) of the homodimer in the asymmetric unit of the crystals showed essentially identical active-site structures except for those in the AGAO ETA/HCl crystal, in which active-site residues, including TPQ and water molecules, had slightly different conformations and electron densities between the two monomers.
The most notable finding from the crystal structures was that the TPQ sq moieties in AGAO PEA and AGAO HTA had an oncopper conformation with ϳ100% occupancy, with the 4-OH group of TPQ sq projecting toward the copper center at a distance of 2.7-2.9 Å (Fig. 7, D and E) and the 5-NH 2 group positioned opposite to the catalytic base (Asp-298) in close proximity to Met-602. This is the first x-ray structure of CAO with TPQ sq being exclusively copper-ligating, although an on-copper TPQ sq structure with ϳ65-70% occupancy was recently reported for HPAO-1 crystals reduced with methylamine in a low oxygen environment at pH 8.5 (39). The copper center in AGAO PEA and AGAO HTA was tetrahedrally coordinated with the 4-OH group of TPQ at the "axial" position and the imidazole groups of three histidines (His-431, His-433, and His-592) at the "equatorial" positions. No water molecules were coordinated to the copper center. On the other hand, TPQ amr in AGAO ETA/HCl had an off-copper conformation with ϳ100% occupancy, in which the 5-NH 2 group of TPQ amr was positioned close to Asp-298 and the 4-OH group was hydrogenbonded to the Tyr-284 side chain rather than coordinating to the copper center (Fig. 7F). The off-copper conformation was the sole conformer in chain A. However, both the off-copper and on-copper conformers were present (at a relative abundance of about 6:4) in chain B. Probably, the minor on-copper conformation in chain B was partly responsible for the crystals' small TPQ sq -like absorption peaks (Fig. 7C). The spectrophotometrically assigned TPQ psb moiety of AGAO PEA/NaCl and AGAO PEA/NaBr also had an off-copper conformation, in which the C5 position of TPQ was connected to additional electron density corresponding to the phenylethyl moiety of the product phenylacetaldehyde (PAA) via a covalent linkage in the form of an imine bond (Fig. 8, D and E).
The binding of several halide ions was detected in the AGAO ETA/HCl , AGAO PEA/NaCl , AGAO PEA/NaBr , and AGAO NaBr crystals (2 and 5 Cl Ϫ ions/dimer in AGAO ETA/HCl and AGAO PEA/NaCl , respectively, and 8 Br Ϫ ions/dimer in AGAO PEA/NaBr and AGAO NaBr ), as judged by the anomalous peaks (over 7 ) generated by the Br Ϫ ion and the electron densities (over 8 ) of the Cl Ϫ ion, which were clearly greater than those of water molecules (less than 5 ). The Cl Ϫ ions identified in AGAO ETA/HCl derived from the substrate (ethylamine hydrochloride), which was present at a high concentration (50 mM). In the halide-bound complexes, a Br Ϫ or Cl Ϫ ion was found to occupy the axial coordination site of the active-site copper center in the AGAO ETA/HCl , AGAO-PEA/NaCl , and AGAO PEA/NaBr structures (Figs. 7F and 8 (D and E) and Table 3). These are the first x-ray crystal structures of CAO with an anionic inhibitor (57,61,62) bound to the active site copper center. In contrast, the axial position in AGAO NaBr was occupied by a water molecule rather than Br Ϫ (Fig. 8F and Table 3); Br Ϫ ions instead bound to the protein surface in a seemingly nonspecific fashion. Similar behavior was observed in AGAO PEA/NaBr . Specific binding of a halide ion at the axial position of the active-site copper center only occurred in crystals that had been anaerobically soaked with substrate, strongly suggesting that the uncompetitive inhibition of the steady-state reaction by halide ions with respect to 2-PEA is due to their ability to bind to the copper center in the reaction intermediates, in which TPQ is reduced with substrates, rather than to the copper center in the free enzyme. It also suggests that the axial ligand-binding position of the copper center exhibits a stronger preference for halide ions in the substrate-reduced form of AGAO than in the resting form.
Comparison of the on-copper TPQ sq structures of AGAO PEA and AGAO HTA with the off-copper TPQ ox structure in resting AGAO (Protein Data Bank code 1IU7) revealed that most active-site residues (except for Tyr-296 and Met-602), water molecules, and the copper center are retained in almost the same positions and conformations (Fig. 9). In the TPQ sq structures, the side-chain phenol ring of Tyr-296 rotates ϳ80°a round the C␣-C␤ bond to participate in a hydrogen-bonding network involving two water molecules (W1 and W2); the 2-OH, 4-OH, and 5-NH 2 groups of TPQ sq ; the 4-OH group of Tyr-284; and the S␦ atom of Met-602 (Fig. 9A). The on-copper conformation is probably stabilized by this hydrogen-bonding FIGURE 9. Interactions in the on-copper and off-copper conformations. The active sites of AGAO PEA (A) and the substrate-free and oxidative form of AGAO (Protein Data Bank code 1IU7) (B), in which the TPQ ring has on-copper and off-copper conformations, respectively, are drawn showing hydrogen bonds (dotted lines) and ligation to the copper centers (red lines). Estimated hydrogen bond lengths are shown in Å. The superposition of A and B is shown in C, in which the on-copper (A) and off-copper (B) structures are colored in purple and gray, respectively. All molecular drawings were generated using PyMOL.
network, allowing the direct coordination of the 4-OH group of TPQ to the copper center, which would facilitate rapid, ligandto-metal charge transfer-like electron transfer from TPQ amr to Cu(II) to form the TPQ sq ⅐Cu(I) state.
In keeping with the single-crystal microspectrophotometric observations, a TPQ psb -like intermediate was identified in the AGAO PEA/NaBr and AGAO PEA/NaCl structures, which was probably formed by a condensation reaction between the amino group of TPQ amr (in the off-copper form) and the aldehyde group of the reaction product PAA, which remains in the substrate-binding hydrophobic pocket (Fig. 8, D and E); PAA was indeed found to remain bound in the AGAO PEA structure (Fig. 7D), as reported previously for the 2-PEA-reduced E. coli CAO crystals (63). This assumption is also supported by the finding that the TPQ psb observed in AGAO PEA/NaBr and AGAO PEA/NaCl had a cis-configuration, whereas the TPQ psb formed during the reductive half-reaction of the D298A mutant of AGAO had a trans-configuration (29). The absence of the product aldehydes (imidazole-4-acetaldehyde and acetaldehyde) in the AGAO HTA and AGAO ETA/HCl structures is probably due to their low affinities for the substrate-binding pocket of AGAO; for comparative purposes, the K m values for histamine (1.2 mM) and ethylamine (170 mM) are 220-and 30,000fold higher, respectively, than that for 2-PEA (5.4 M) (41). Overall, these findings indicate that in the AGAO PEA/NaBr and AGAO PEA/NaCl crystals, the binding of halide ions (Br Ϫ and Cl Ϫ ) to the axial position of the copper center prevented TPQ amr from adopting the on-copper conformation, causing it to back-react with PAA to form a TPQ psb configuration distinct from that formed during the reductive half-reaction.
Effect of pH on the Equilibrium between TPQ amr and TPQ sq -The equilibrium between TPQ amr and TPQ sq in AGAO has been reported to shift toward TPQ sq under alkaline conditions (35), which suggests that an ionizable group(s) plays a role in triggering the conformational change of the cofactor. We therefore used spectrophotometry to investigate the effect of pH on the equilibrium between TPQ amr and TPQ sq . The reductive halfreaction was performed with 2-PEA as the substrate at pH values ranging from 5.7 to 10.1 with a constant ionic strength (Fig.  10A); AGAO is stable in this pH range. The TPQ sq -specific absorption peaks at about 440 and 470 nm, and the shoulder at about 350 nm increased in intensity as the pH rose from 5.7 to 8.5 but did not change further above pH 8.5, indicating that the equilibrium shifted toward TPQ sq above pH 8.5. By plotting the absorbance at 468 nm, to which TPQ amr makes no contribution, against the pH (Fig. 10B) and fitting the data to Equation 1, two ionizable groups with apparent pK a values of 5.96 Ϯ 0.05 (pK a1 ) and 7.74 Ϯ 0.19 (pK a2 ) were found to be involved in the equilibrium shift from TPQ amr to TPQ sq . Judging from the magnitude of pH-independent absorbance values (C 1 ϭ 0.164, C 2 ϭ 0.053), deprotonation of the ionizable group with pK a ϭ 5.96 contributes predominantly to the equilibrium shift.

Discussion
The results presented above demonstrate that the off-copper and on-copper conformations of TPQ are readily interconvertible during the reductive half-reaction with various amine substrates and in the presence of halide ions. In the steps prior to the formation of TPQ amr (Scheme 1), the catalytic reaction proceeds with TPQ always maintained in the off-copper conformation irrespective of its chemical state (29,40,41). In the TPQ ssb and TPQ psb states, the distal part of the substrate amine is anchored to the substrate-binding pocket, preventing the TPQ ring from adopting the on-copper conformation (29). However, once TPQ psb is hydrolyzed, the reduced TPQ gains the conformational flexibility that enables facile interconversion between the off-copper and on-copper conformations.
In the on-copper conformation of TPQ sq as observed in the AGAO PEA and AGAO HTA structures (Fig. 9, A and B), the active-site copper center is equatorially coordinated by three imidazole groups from His residues without the equatorial water ligand seen in the resting TPQ ox ⅐Cu(II) state (17) and other intermediates formed in the reductive half-reaction (Figs. 7 and 8). A similar decrease in the number of equatorial ligands at the copper center was observed in extended x-ray absorption fine structure studies on various dithionite-treated CAOs, in which Cu(II) is reduced to Cu(I) (64). It is therefore suggested that the tetrahedrally coordinated copper centers observed in the AGAO PEA and AGAO HTA crystals are probably in the Cu(I) oxidation state, as was proposed in a recent paper on the structure of methylamine-reduced HPAO-1 (39). Based on the full occupancy of the modeled TPQ sq in the electron density map, essentially all of their TPQ is assumed to be in the TPQ sq state. Taken together, these results show that the on-copper conformation is a consequence of 1e Ϫ transfer from TPQ amr to Cu(II) to form the TPQ sq ⅐Cu(I) state. In contrast, a water molecule (W eq ) was identified as an equatorial ligand of the copper atom in the AGAO ETA/HCl crystal, with observed electron densities of 3.7 and 2.3 in crystallographically distinguishable chains A and B, respectively. In addition, a Cl Ϫ ion was found to occupy an axial position in the copper complex (Fig. 7C), whose fivecoordinate square pyramidal structure suggests that the copper is in the Cu(II) oxidation state, especially in chain A, for which TPQ is mostly in the off-copper TPQ amr state based on its absorption spectrum and x-ray structure. We have so far observed neither the off-copper TPQ sq ⅐Cu(I) form nor the oncopper TPQ amr ⅐Cu(II) form, strongly suggesting that the 1e Ϫ transfer occurs exclusively in the on-copper conformation of FIGURE 10. pH dependence of TPQ sq formation. A, 100 M AGAO monomer was anaerobically reduced with 1 mM 2-PEA at various pH values in the presence of 100 mM Na 2 SO 4 , and UV-visual absorption spectra of AGAO were measured at 25°C after a 5-min preincubation. B, absorbance at 468 nm specific to TPQ sq was plotted against pH. A solid line indicates the theoretical line obtained by data fitting. The spectra at pH 5.67, 6.02, 6.53, 6.99, 7.52, 8.03, 8.54, 9.04, 9.55 TPQ amr . This also leads to the suggestion that before the formation of the TPQ sq ⅐Cu(I) state, the TPQ quinone ring moves from the off-copper to the on-copper conformation.
The spectroscopic data from the transient kinetics experiments showed that the formation of TPQ sq is dependent on solvent viscosity (Fig. 6D), further supporting the occurrence of a conformational change in TPQ during the last step of the reductive half-reaction. Because the rate constants of this step (k Ϯ4 ) were determined from the absorbance changes associated with TPQ sq , they represent the rates of both the conformational change of TPQ and the subsequent electron transfer from TPQ amr to Cu(II). The same presumption holds for the rate constants of the electron transfer (k ET ) determined previously by temperature-jump relaxation studies (33,35,65), which varied from 60 -75 s Ϫ1 (Arthrobacter P1 methylamine oxidase) (65) to 20,000 s Ϫ1 (P. savitum CAO) (33). The latter extremely large rate constant was attributed to the intrinsic k ET and interpreted to mean that the TPQ cofactor is in close proximity (ϳ3 Å) to the copper center (33) (cf. the distance of 2.6 Å between the TPQ 4-O atom and the copper center; Fig. 9). It is thus likely that the smaller rate constants determined previously (k ET ) and in this study (k ϩ4 ) mainly reflect the rate constants for the conformational change of the TPQ cofactor, in agreement with the suggestion that conformationally "gated" or controlled electron transfer is plausible (33). An extended x-ray absorption fine structure study (64) also raised the possibility that variations in the redox potentials or the effective electron transfer distance between TPQ amr and Cu(II) may control k ET . The difference between the rate constants for AGAO determined here ( Table  2; k ϩ4 ϭ 39 s Ϫ1 at 4°C, pH 6.8) and previously (k ET ϭ 73 s Ϫ1 at 5°C, pH 7.2) (37) is probably due to subtle differences in temperature and pH, both of which strongly affect the equilibrium between TPQ amr ⅐Cu(II) and TPQ sq ⅐Cu(I) (33,35,37,39,65) (Fig. 10); at higher pH values and temperatures, the equilibrium shifts toward TPQ sq ⅐Cu(I), which suggests a ⌬H value of Ͼ0 for the conformational change of TPQ. Finally, it should be noted that the rate constants of the conformational change are appreciably larger than the k cat value (17 s Ϫ1 ) determined by steadystate kinetics under the same conditions (at 4°C and pH 6.8), supporting the hypothesis that the conformational change of the cofactor can occur within the overall turnover reaction.
Based on the pH dependence of the equilibrium between TPQ amr and TPQ sq , it is suggested that deprotonation of two ionizable groups with pK a values of 5.96 Ϯ 0.05 and 7.74 Ϯ 0.19 (Fig. 10B) facilitates the equilibrium shift toward TPQ sq . Among several ionizable groups in the active site (5-NH 2 (estimated pK a , 5.88), 4-OH (9.59), and 2-OH (11.62) groups of TPQ amr (54); the carboxyl group of Asp-298 (7.5 Ϯ 0.20) (29); and the water axially coordinated to the copper atom (ϳ7.5) (66)), the 5-NH 2 group of TPQ amr is most likely assigned to the group with the lower pK a value (5.96) and mainly contributes to the equilibrium shift toward TPQ sq , with the neutral form of TPQ amr that is the major form at pH Ͼ7 being the direct precursor to TPQ sq (Scheme 1). Furthermore, the neutral form of TPQ amr is only weakly tethered in the active site, forming neither electrostatic interactions nor charge-assisted hydrogen bonds (67) with surrounding residues. It should thus be amenable to facile conformational change. The ionizable group with the higher pK a value (7.74), although contributing insignificantly to the equilibrium shift toward TPQ sq , may be ascribed to the 4-OH group of TPQ sq with a pK a value of 6.39 determined with a model compound (68), rather than the same group in TPQ amr (pK a 9.59) (54); deprotonation of the 4-OH group of TPQ sq is expected to stabilize the TPQ sq form (see Scheme 1). If this is the case, deprotonation of the 4-OH group would occur after the conformational change of the TPQ ring and the electron transfer from TPQ amr to Cu(II). A notable increase of the pK a value (from 6.39 to 7.74) may be conceivable to occur in the hydrophobic active site of CAOs, as observed for the carboxyl group of the catalytic base Asp-298 with a significantly high pK a value (7.5 Ϯ 0.20) (29).
The TPQ amr structures have been determined previously with the substrate-reduced forms of E. coli CAO and HPAO-1 (Protein Data Bank codes 1D6U and 4EV2, respectively), in which the axial positions of the copper atom are occupied by water and a dioxygen species, respectively. Comparison of these structures with the TPQ amr of AGAO bound with a chloride ion (AGAO ETA/HCl ) has revealed that the TPQ amr ring is tilted anticlockwise by about 20°(rigid body rotation around the C␤-C␥ bond) in AGAO (Fig. 11A). This ϳ20°tilting of the TPQ ring appears to result from the minuscule movement (by 0.4 Å) of the position of the 2-OH group of TPQ amr , probably due to the repulsion from the axially coordinated chloride ion that has a larger van der Waals radius than water and a dioxygen species. Consequently, the 5-NH 2 group approaches within hydrogen bond distance (2.8 Å) to the carboxyl group of Asp-298 (Fig.  11A), which is predominantly protonated at crystallization pH of 6.8, thereby lowering the nucleophilicity of the 5-NH 2 group. The hydrogen bond may also stabilize the tilted conformation of TPQ amr even after the halide ion is released (in the step of k Ϫ5 in Scheme 1). In addition, the aldehyde group of the product PAA that remains bound in AGAO PEA is located rather distant (3.2 Å) from the 5-NH 2 group of TPQ amr in the AGAO ETA/HCl structure to undergo the nucleophilic attack (Fig. 11B). These structural consequences well explain the significantly decreased rate constant (k Ϫ3 ) for the back-formation of TPQ psb in the presence of halide ions (Table 2). Moreover, the geometry of the 5-NH 2 group of TPQ amr relative to the aldehyde carbon atom of PAA strongly suggests that the nucleophilic attack occurs from the si face of the carbonyl carbon, leading to the formation of TPQ psb in cis-configuration (cis-TPQ psb ) (Fig.  11D), unlike the formation of TPQ psb in trans-configuration (trans-TPQ psb ) from TPQ ssb in the forward reductive half-reaction of the D298A mutant (Fig. 11C) (29).
Monovalent anions, such as cyanide and azide, were reported to be inhibitors of various CAOs, showing competitive, noncompetitive, uncompetitive, or mixed type inhibition with respect to the amine substrate and dioxygen (57,61,62). For AGAO, cyanide is an uncompetitive inhibitor with respect to the amine substrate, and azide is a noncompetitive inhibitor with respect to both amine and dioxygen (57,61). Although the inhibition patterns of halide ions (uncompetitive and noncompetitive with respect to amine substrate and dioxygen, respectively) (Fig. 3) for AGAO are similar to those of azide and cyanide, their effects on the equilibrium between TPQ amr and TPQ sq are very different. Halide ions inhibit TPQ sq formation ( Fig. 2) by axially coordinating to the copper center (Figs. 7F and 8 (D and E)), whereas azide converts TPQ sq into a ligandto-metal charge transfer complex (57), and cyanide facilitates TPQ sq formation (61), both of which were suggested to bind at an equatorial position of the copper center (69,70). It is assumed that azide is probably too large to bind axially to the copper atom and probably cyanide too. Although we cannot currently explain why halide ions, but not azide/cyanide, bind at the axial position of the copper center in the reduced form of TPQ and inhibit TPQ sq formation, it is reasonable to assume that halide ion binding at the axial position of the copper atom would block the coordination of the 4-OH group of TPQ amr to the copper center and prevent the subsequent electron transfer to Cu(II). The inability of halide ions to bind to the axial position of the copper center in the resting TPQ ox state (Fig. 8F) may be due to electrostatic repulsion from the delocalized negative charge through the 4-O Ϫ to 2-CϭO group of TPQ ox ; in the TPQ amr state, both the 2-OH and 4-OH groups (whose estimated pK a values are 11.62 and 9.59, respectively) (54) are neutral and therefore would not electrostatically repel halide ions. The uncompetitive inhibition of halide ions with respect to the amine substrate is consistent with their binding only to the reaction intermediate, whereas noncompetitive inhibition with respect to dioxygen indicates halide binding at a site distinct from the O 2 -binding site. The exclusive binding of halide ions to Cu(II) with TPQ amr in the off-copper conformation (which does not bind dioxygen) is challenging to reconcile with their activity as inhibitors of the oxidative half-reaction that are noncompetitive with respect to dioxygen. This issue can be resolved by supposing that dioxygen binds directly to Cu(I) with TPQ sq in the on-copper conformation during the oxidative half-reaction and undergoes 1 e Ϫ -reduction by Cu(I) through the inner sphere mechanism proposed for AGAO (37,38). Alternatively, halide ions may inhibit the oxidative halfreaction by binding to the O 2 -bound enzyme.
The off-copper to on-copper conformational change of TPQ amr involves three motions of the TPQ ring: sliding (ϳ53°r otation around the C␣-C␤ bond), tilting up (ϳ20°rigid body rotation centered at the C␣ carbon), and revolution (180°rotation around the C␤-C␥ bond) (Fig. 12). In the off-copper con- FIGURE 11. Effect of binding of chloride ion at the axial position of the copper center on the conformation of TPQ amr . A, conformation of the TPQ amr ring in the AGAO ETA/HCl structure (green) is compared with those of the substrate-reduced E. coli CAO (ECAO) (purple) and HPAO-1 (magenta). Cyan spheres, copper atoms. Residue numbers are referred to those of AGAO. van der Waals surfaces of the chloride ion and the oxygen atom of the 2-OH group of TPQ amr are represented with gray dots. B, comparison of cis-TPQ psb formed in AGAO PEA/NaCl (orange), cis-TPQ psb formed in the D298A mutant of AGAO (29) (gray), TPQ amr formed in AGAO ETA/HCl (green), and PAA formed in AGAO PEA (green). C, schematic drawing of the presumed mechanism of the formation of cis-TPQ psb in AGAO PEA/NaCl . FIGURE 12. Possible route for the conformational change of TPQ in the active site of AGAO. Stick models of the active-site residues in AGAO ETA/HCl (TPQ amr , cyan) and AGAO PEA (TPQ sq , yellow) are shown within the cavity with its surface drawn in half-transparent gray. The on-copper TPQ amr conformer predicted after the first combined sliding/tilting up motion is colored green. The following 180°rotation of the TPQ ring provides a conformation identical with that of the on-copper TPQ sq . The rotation direction and movement of the TPQ ring are shown with blue arrows. The figure was generated with PyMOL.
formation, the TPQ ring is sandwiched between the side chains of Asn-381 and Tyr-384/Val-282 in a narrow wedge-shaped space (29,53). The initial step of the conformational change is a simultaneous combination of sliding and tilting up in order to avoid a steric clash between the TPQ ring and the side chain of His-433, which would come within ϳ1 Å of the ring if it only slid. This combined sliding/tilting-up motion leads to the axial coordination of the 4-OH group of TPQ amr to the copper atom, where there is sufficient space for the TPQ ring to rotate by 180°a round the C␤-C␥ bond. The final 180°rotation of the TPQ ring can only occur in the clockwise direction because anticlockwise rotation would lead to a clash between the 5-NH 2 group and the His-433 side chain while permitting a minor movement of the Tyr-384 side chain (Fig. 12). It is unclear whether the electron transfer from TPQ amr to Cu(II) occurs immediately upon formation of the on-copper conformation of the TPQ ring or after the 180°rotation to the final conformation stabilized by the hydrogen-bonding network (Fig. 9A).
Finally, it is noteworthy that the on-copper TPQ sq structure stabilizes the conformation of the side chain of Met-602 by hydrogen bond formation between the 5-NH 2 group of TPQ sq and the S␦ atom of Met-602 (Fig. 9). Met-602 is located at the end of the predicted O 2 pathway from the O 2 -prebinding site to the copper center and has conformational flexibility with dual extreme conformers (19). Thus, it is tempting to speculate that the tethering of the Met-602 side chain could act as a gate to allow O 2 to enter into the copper center in the initial phase of the oxidative half-reaction.
In conclusion, the results presented herein show that TPQ undergoes a large conformational change during the reductive half-reaction of AGAO, which efficiently mediates between the acid/base chemistry conducted in the off-copper conformation of TPQ by the conserved catalytic base (Asp-298 in AGAO) and the redox chemistry conducted in the on-copper conformation of TPQ at the metal center.