The nature of O2 activation by the ethylene-forming enzyme 1-aminocyclopropane-1-carboxylic acid oxidase

Ethylene is a plant hormone important in many aspects of plant growth and development such as germination, fruit ripening, and senescence. 1-Aminocyclopropane-1-carboxylic acid (ACC) oxidase (ACCO), an O2-activating ascorbate-dependent nonheme iron enzyme, catalyzes the last step in ethylene biosynthesis. The O2 activation process by ACCO was investigated using steady-state kinetics, solvent isotope effects (SIEs), and competitive oxygen kinetic isotope effects (18O KIEs) to provide insights into the nature of the activated oxygen species formed at the active-site iron center and its dependence on ascorbic acid. The observed large 18O KIE of 1.0215 ± 0.0005 strongly supports a rate-determining step formation of an FeIVO species, which acts as the reactive intermediate in substrate oxidation. The large SIE on kcat/Km(O2) of 5.0 ± 0.9 suggests that formation of this FeIVO species is linked to a rate-limiting proton or hydrogen atom transfer step. Based on the observed decrease in SIE and 18O KIE values for ACCO at limiting ascorbate concentrations, ascorbate is proposed to bind in a random manner, depending on its concentration. We conclude that ascorbate is not essential for initial O2 binding and activation but is required for rapid FeIVO formation under catalytic turnover. Similar studies can be performed for other nonheme iron enzymes, with the 18O KIEs providing a kinetic probe into the chemical nature of Fe/O2 intermediates formed in the first irreversible step of the O2 activation.

studied extensively, and it generally is accepted that O 2 activation at the iron center is linked to oxidative decarboxylation of the ␣KG, ultimately forming an Fe IV AO species as the reactive intermediate (14)(15)(16)(17). This high-valent Fe/O 2 species functions as a ''generic'' oxidant for a wide range of oxidation/oxygenation reactions.
In contrast to the ␣KG-dependent enzymes, where ␣KG rather than substrate binds to the metal center, spectroscopic studies of ACCO have shown that ACC coordinates to the iron center via both its amino and carboxylate groups (10,18). † This finding implies a distinctly different mode for reductant interaction with ACCO than for most other nonheme iron enzymes. Magnetic circular dichroism (MCD) studies indicate that the active-site iron center is six-coordinate in the resting state Fe(II)-ACCO (19). On addition of ACC and ascorbate, the iron center becomes five-coordinate, allowing for O 2 binding and activation. These results are in line with steady-state kinetic analyses that suggest an ordered process where ACC binding to ACCO must precede O 2 binding, although it could not be distinguished whether ascorbate binds before ACC or after O 2 (12).
Single-turnover experiments showed that a substoichiometric amount of ethylene was formed in the absence of ascorbate, with the presence of either CO 2 or bicarbonate being essential for enzyme turnover (13). Although ascorbate was proposed to act mainly as a reductant for restoring the iron to the ferrous state, the rate of ethylene formation under single-turnover conditions was significantly lower than under steady-state conditions, suggesting that ascorbate is needed for O 2 activation under catalytic turnover (Fig. 2).
From these accumulated results, it is clear that the complexity of the ACCO chemistry presents a challenge for mechanistic enzymologists and requires further examination. In this study, steady-state kinetics, solvent isotope effects (SIEs), and competitive oxygen kinetic isotope effects ( 18 O KIEs) are used to provide insights into the rate-limiting steps contributing to the interaction of O 2 with ACCO, the nature of the Fe/O 2 intermediates, and the role of ascorbate in O 2 activation. Of particular interest is the investigation of the Fe IV AO species formation and its proposed role in substrate oxidation (20).
are similar to our previous results (12). Because both ACC and ascorbate have solvent-exchangeable protons, the only deuterium isotope effects that can be measured are kinetic SIEs. Comparison of the kinetic parameters measured in both H 2 O and D 2 O revealed the presence of SIEs ( Table 1). The measured values for D2O k cat , D2O k cat /K m (ACC), and D2O k cat /K m (Asc) are Ϸ2.3, identical within experimental error, suggesting changes in hydrogen bonding between the reactants and the transition state. None of the kinetic parameters has been found to vary between pH 6 and 8 (data not shown), indicating the lack of a pK a effect as the origin of the SIEs. Interestingly, the D2O k cat /K m (O 2 ) value is 5.0 Ϯ 0.9 (Table 1 and Fig. 3), an uncommonly large value for a SIE. Such a value suggests that the rate-determining step in O 2 activation involves either a proton-coupled electron transfer (PCET) or hydrogen atom transfer (HAT), as discussed below. 18 16 O isotopic ratio at f fractional conversion and R 0 is the isotopic ratio before the enzymatic reaction.
The data are well fitted by Eq. 1 to give an 18  values at lower ascorbate concentrations is not expected for an ordered terreactant enzymatic reaction, suggesting a possible change in the order of ascorbate binding to ACCO that has an effect on the O 2 activation steps (see below).

Discussion
The gene for the tomato ACCO has been expressed recently in this laboratory in high yield in E. coli and purified to homogeneity to yield a protein with approximately 10 times greater activity than previously reported (12). Some of the initial studies undertaken with this enzyme have included the kinetic order for binding of all substrates to enzyme, the relationship of O 2 activation to substrate activation, and the chemical steps that lead from ACC to its products. Steady-state kinetic studies indicated a sequential mechanism involving a quaternary complex (ACCO, ACC, O 2 , and ascorbate) and an equilibriumordered pattern with ACC binding before O 2 but were unable to distinguish whether ascorbate binds before ACC or after O 2 (12).
Activation of ACC by ACCO is likely to occur by one-electron oxidation or hydrogen atom abstraction from the amine group to form an amine radical cation or aminyl radical, respectively (23). Rapid radical rearrangement results in cleavage of the cyclopropane ring and eventually formation of ethylene, HCN, and CO 2 . In a recent study, the behavior of three cyclic and three acyclic substrate analogs was analyzed in regard to turnover rates, product distribution, and O 2 uncoupling (20). Although these analogs have different structures, the turnover rates (k cat ) are within a factor of four of ACC, suggesting that the ratedetermining step occurs before substrate oxidation (20). It also was proposed that the first committed step involved the formation of an Fe IV AO species, which acts as the oxidant in substrate activation. An Fe IV AO species had been implicated previously as the oxidant that activates the substrate in the ␣KG-dependent enzymes (14)(15)(16). However, as noted ACCO uses ascorbate as the reductant, which can serve as a one-or two-electron donor;  Lipscomb, Que, and colleagues suggested that ACC oxidation may involve either direct HAT to an Fe III OOOH or prior OOO bond heterolysis to generate Fe V AO as the oxidizing species (13 (Table 1). These SIEs are suggestive of proton transfers occurring along hydrogen bonds involved in general acid/base catalytic stabilization of the transition state (24). However, given the large number of protons in the active site that can exchange with solvent (water molecules coordinated to the iron center, amine protons of ACC, acidic protons of ascorbate, bicarbonate), these SIEs may represent the product of multiple effects and are referred to as background SIEs. A pK a origin of the SIE can be eliminated because all kinetic parameters are found to be invariant between pH 6 and 8. Interestingly, D2O k cat / K m (O 2 ) is 5.0 Ϯ 0.9 (Table 1 and Fig. 3). Such a large SIE indicates that either a PCET or HAT from a solventexchangeable site occurs in the rate-limiting step of k cat /K m (O 2 ) (24). Possible proton sources for a PCET process may include acidic side-chain residues, ascorbate, or bicarbonate, whereas a hydrogen atom may be removed from ascorbate or the ACC amine group.    (26,27). Equilibrium oxygen isotope effects ( 18 O EIEs) can be calculated from vibrational frequencies of reactants and products by using the Bigeleisen-Meyer equation (28). For the formation of the Fe IV AO species, the calculated 18 O EIE is 1.0287 (29). ‡ Additionally, 18 O EIEs have been reported for the O 2 -binding iron proteins myoglobin and hemerythrin (21), in which O 2 is reduced at an iron center to an Fe III OO 2 •Ϫ and Fe III OOOH species, respectively. The respective measured 18 O EIEs are 1.0054 Ϯ 0.0006 and 1.0113 Ϯ 0.0005 (Table 2).
Generally, the 18 O EIE values can be used as upper limits for measured 18 O KIEs based on transition state theory (22,25). This assumption is based on the approximation that, for innersphere electron transfer reactions, the heavy-atom isotope effect contribution from the reaction coordinate frequency is negligible (30), as supported by recent results (31). In our case, the calculated EIE for the formation of an Fe IV AO species (1.0287) is the only value larger than the 18 O KIE of ACCO (1.0215 Ϯ 0.0005). Given that the first irreversible step is the last step to be expressed in the 18 O KIE, the present result points toward Fe IV AO species formation as the first committed step in O 2 activation by ACCO. As an alternative, and by analogy to D␤M and PHM (26,27), it was conceivable that in ACCO the Fe III OO 2 •Ϫ species formed rapidly and reversibly, followed by formation of an Fe III OOOH species through an irreversible hydrogen atom abstraction from ascorbate or ACC; the former is favored as the reductant based on steady-state kinetics studies that implicate substrate oxidation occurs after the rate-limiting step in k cat (20). However, given the small measured 18 O EIEs of 1.0054 Ϯ 0.0006 and 1.0113 Ϯ 0.0005 for conversion of Fe(II) to Fe III OO 2 •Ϫ and Fe III OOOH in myoglobin/hemoglobin and hemerythrin, respectively (21), together with the expected ease with which ascorbate will be oxidized, it is very difficult to envisage the irreversible Fe III OOOH formation as the source of the large observed 18 O KIE. The large 18 O KIEs measured for the Cu II OOOH formation in the copper monooxygenases D␤M and PHM may be attributed to increased back-bonding interactions between copper and the O 2 -derived moiety (32). Turning to the Fe V AO species as a potential intermediate in ACCO (13), the absence of experimental frequency data precludes the estimation of an 18 (Fig. 4A). The lack of a SIE on the 18 O KIE suggests that solvent deuteration does not impact the relative energies of the transition states for the O 2 activation steps, which implicates Fe IV AO species formation through a PCET step that is fully rate-limiting for k cat /K m (O 2 ).
The Role of Ascorbate in O2 Activation. Given the ambiguous role of ascorbate during O 2 activation, the impact of ascorbate was investigated by measuring the 18 O KIE at a lower ascorbate concentration (2 mM), below its K m value. Assuming an ordered terreactant mechanism, if ascorbate binds first, followed by ACC, and O 2 binds last, then k cat /K m (O 2 ) and consequently the 18 O KIE will be independent of ascorbate concentration (33). If ascorbate binds last, after O 2 , then at higher ascorbate concentration the reaction will be fully committed, because high ascorbate concentration prevents the release of O 2 and its equilibration with excess O 2 . A large forward commitment factor translates into an observed 18 O KIE of 1.0 (33), and lowering the ascorbate concentration will increase the measured 18 In contrast to the predicted effect, the measurements reveal that the 18 O KIE decreases from 1.0215 Ϯ 0.0005 at saturating ascorbate concentration (20 mM) to 1.0157 Ϯ 0.0004 at low ascorbate (2 mM, Fig. 4B), which suggests that another step becomes rate-limiting such that a smaller change in the oxygen bond order has been measured. Additionally, D2O k cat /K m (O 2 ) decreases from 5.0 Ϯ 0.9 (at 20 mM ascorbate) to 2.7 Ϯ 0.7 (at 2 mM ascorbate), a value within experimental error of the background SIEs for the other kinetic parameters. The reduction of both 18 O KIE and D2O k cat /K m (O 2 ) is attributed to a partially rate-determining binding of ascorbate at the reduced ascorbate concentrations. In an earlier study (13), the possibility of an effector site for ascorbate was postulated. In light of the proposal of very tight binding of ascorbate to such a site (13), this explanation does not appear relevant to the conditions of the present experiments (2 and 20 mM ascorbate). Further, since the kinetic properties of ACCO can accommodate the available data, there is no need to invoke the presence of an additional ascorbate binding site.
Proposed Mechanism for O2 Activation. Considering all experimental observations, a mechanism for O 2 activation by ACCO is proposed (Fig. 5). Our working model is that, although an ordered mechanism is still in place for substrate and O 2 , ascorbate interaction with enzyme occurs in a random manner, depending on its concentration. At high, saturating concentrations, ascorbate is proposed to bind first, followed by ACC and O 2 . For such a kinetic mechanism, the fully rate-limiting step for k cat /K m (O 2 ) is proposed to be the formation of an Fe IV AO species in a process that involves a PCET, consistent with the large 18 O KIE and D2O k cat /K m (O 2 ). The oxidation of the Fe III OO 2 •Ϫ species to Fe III OOOH by ascorbate is proposed to be reversible, similar to the reversible O 2 binding observed in hemerythrin (21). When the ascorbate concentration is below its K m value, ascorbate is proposed to bind last, subsequent to O 2, with the actual binding being partially ratelimiting. Because ascorbate binding most probably does not involve a change in the oxygen bond order or a direct proton transfer, a reduction in both 18 O KIE and SIE values is expected, as observed experimentally. These findings suggest ‡ A similar value (1.0274) is obtained by taking the square root of the previously calculated 18 O EIE for the formation of an Fe IV AO species from doubly labeled 18,18 O2 (29). that, unlike substrate, which directly coordinates to the iron center, binding of ascorbate at the active site of ACCO is not an essential step for the conversion of the metal center from six-to five-coordinate and initial reduction of O 2 to form the Fe III OO 2 •Ϫ species. Although it is possible that product release or the reduction of Fe(III) to Fe(II) may be rate-limiting under certain conditions (13), the rate-limiting step on k cat /K m (O 2 ) is unlikely to depend on these steps. Because the ascorbate concentration has a pronounced effect on 18 (13) is attributed to an alternate, noncatalytic pathway involving electron transfer from another molecule of ACCO. These studies make ACCO a unique system, providing kinetic as opposed to the more conventional spectroscopic evidence (14 -17) for the presence of an Fe IV AO species. It is important to mention that, to our knowledge, 18 O KIE measurements have not been performed before for an O 2 -activating, nonheme iron enzyme. Although 18 O KIEs values were reported for the pterin-dependent tyrosine hydroxylase, in that case, the initial O 2 activation is proposed to occur at the pterin cofactor and not at the iron center (34). Additionally, the role of ascorbate on O 2 activation has been investigated, leading to a kinetic model in which ascorbate binds to enzyme in a random manner. Thus, although ascorbate is required for rapid Fe IV AO formation under catalytic turnover, the presence of this reductant is not required for initial O 2 binding (13).
Several recent studies have used 18

Materials and Methods
General. All reagents were purchased from commercial sources and used without further purification unless otherwise indicated.
Overexpression and Purification of ACCO. ACCO from Lycopersicon esculentum (ACO1) was produced in Escherichia coli strain BL21(DE3)pLysS and purified by a two-column purification procedure as previously described (12).
Steady-State Kinetics. Initial velocities were measured by the rate of oxygen consumption at 25°C, pH 7.2, using a Yellow Springs Instrument biological oxygen monitor (model 5300) as previously described (12). Temperature was maintained at 25 Ϯ 0.1°C with a Neslab circulating water bath. Standard reaction mixture (1 ml) contained 100 mM MOPS, pH 7.2, 20 mM NaHCO 3, 100 mM NaCl, and various amounts of ascorbate and ACC. When ACC and ascorbate were kept constant, concentrations were maintained at 1 and 20 mM, respectively. When the oxygen concentration was varied, the reaction mixture was equilibrated by stirring for at least 10 min with the appropriate premixed O 2/N2 gas mixture to obtain the desired oxygen concentration. Starting oxygen concentrations were determined by using the oxygen monitor that was calibrated with air-saturated water (258 M oxygen at 25°C). Reactions were initiated with 2 l of ACCO reconstituted with equimolar Fe(NH4)2(SO4)2. Because of the loss of activity on prolonged exposure to Fe(II) in the presence of oxygen (35,36), ACCO was reconstituted in small aliquots and used within 30 min. Concentration of ACCO is as indicated in figure legends. All initial rates were measured under conditions where Ͻ5% of any given substrate was consumed. All rates were calculated subtracting background oxygen consumption attributable to ascorbate and/or Fe(II) in the absence of enzyme. Data from initial velocity experiments with varying substrate concentrations were fitted to the Michaelis-Menten equation by using the program KaleidaGraph. The kinetic parameters are reported with errors of Ϯ1.
SIEs on Steady-State Parameters. Initial rates were measured as described above. The standard buffer was prepared in D2O (99.9%) by dissolving MOPS, NaHCO 3, and NaCl in D2O and then titrating to pD 7.2 with a KOD solution. A value of 0.4 was added to the reading on the pH meter to correct to pD (37). To avoid H 2O contamination in the D2O reactions, the electrode tip was soaked in D 2O before each reaction. In parallel, a buffer was prepared in H2O   Fig. 5. Mechanisms for O2 activation in ACCO involving ascorbate-dependent formation of the Fe IV AO species. Only the redox-active fragment of ascorbate is shown. Asc, ascorbate; DHAsc, dehydroascorbate; RDS, rate-determining step; PRDS, partially rate-determining steps.