Unless noted otherwise, enzymes (see list in Table 1) and chemicals were purchased from Sigma-Aldrich and used as received. Sodium phosphate dibasic (Na2HPO4, 99 %, Carl Roth), sodium phosphate monobasic dihydrate (NaH2PO4⋅2H2O, 99 %, Merck), sodium acetate (98.5 %, Carl Roth), 2-amino-2-(hydroxymethyl)-1,3-propanediol (Tris, 99 %), N-2-hydroxyethylpiperazine-N′-2-ethane-1-sulfonic acid sodium salt (HEPES, 99 %, Carl Roth), sodium hydroxide (NaOH, 98 %), and hydrochloric acid (HCl, 37 %, VWR) were used to make buffer solutions. Sodium chloride (NaCl, 99.5 %, Carl Roth), potassium chloride (KCl, 99 %), thiamine diphosphate (95 %), manganese sulfate (MnSO4, 99 %, Carl Roth), flavin adenine dinucleotide disodium salt hydrate (FAD, 95 %), DL-dithiothreitol (99 %), Thesit^® (non-ionic surfactant for membrane research), and isopropanol (HPLC grade, Carl Roth) were added to certain enzyme assays to increase enzymatic activity or substrate solubility. Methanol (99.9 %, Carl Roth), ethanol (99.8 %, Honeywell), L-ascorbic acid (98 %), bilirubin (98 %), cholesterol (99 %), choline chloride (98 %), cytochrome c from bovine heart (95 %), D-alanine (98 %), histamine dihydrochloride (99 %), D-(+)-glucose (99.5 %), D-(+)-mannose (99 %), L-kynurenine (98 %), β-nicotinamide adenine dinucleotide phosphate reduced tetrasodium salt hydrate (NADPH, 95 %), hydroquinone (99 %), 2,2^′-azino-bis(3-ethylbenzothiazoline-6-sulfonic acid) diammonium salt (ABTS, 98 %), L-(+)-lactic acid (98 %), L-lysine monohydrochloride (99.5 %), sodium pyruvate (99 %), and sarcosine (98 %) were used as (co)substrates. Hydrogen peroxide (H2O2, 30 %), formaldehyde (36 %), acetaldehyde (99.5 %), betaine (98 %), ammonium chloride (≥99%), p-benzoquinone (98 %), sodium bicarbonate (99.5 %), and glycine (98.5 %) were used to test product inhibition of enzymatic activities. Sodium sulfite (Na2SO3, 98 %) was used to calibrate optical oxygen sensors. All solutions were made in ultrapure water (18.2 MΩcm, ELGA LabWater). O2 (99.995 %), N2 (99.999 %), and He (99.999 %) gases were from Carbagas AG.

To measure (initial) O2 consumption rates, enzyme assays were performed in clear-glass, crimp-top vials with a volume of 9 mL when closed. These vials contained small magnetic stir bars, were filled headspace-free with assay solution, and closed with hollow butyl rubber stoppers and crimp caps. Assay solutions consisted of an air-equilibrated buffer, an organic substrate, cofactors and co-substrates if necessary, and the respective enzyme of interest (see Appendix A for details). Once filled and closed, vials were placed on a magnetic stirring plate at room temperature (23±1°C). Enzymatic reactions were initiated with the addition of small volumes of enzyme or substrate solution through the septum with a gas-tight glass syringe. Dissolved O2 concentrations were continuously monitored inside the closed vials with fiber-optic oxygen minisensors and a FireSting meter (PyroScience GmbH) with automated pressure, humidity, and temperature correction. The fiber-optic minisensors are housed in stainless-steel needles (1.1 mm o.d.), with which the crimp vial septa can be pierced. Optical oxygen sensors were calibrated for maximum and minimum dissolved O2 concentrations with air-equilibrated water and with a 300 mM Na2SO3 solution, respectively. Accurate temperature compensation was performed with optical temperature sensor spots (PyroScience GmbH) inside the vials. These sensor spots were regularly calibrated with the temperature probe of the FireSting meter.

With this type of enzyme assay, initial O2 consumption rates were measured to determine Km values for all enzymes with varying initial organic substrate concentrations, referred to as Km(S), except for cytochrome-c oxidase and kynurenine 3-monooxygenase (KMO) because of limited substrate availability. In addition, this type of enzyme assay was used to measure initial O2 consumption rates in the presence or absence of reaction products (see Appendix A for details). Inhibition of enzymatic activities due to the presence of reaction products (i.e., product inhibition) was tested for all enzymes but only detected for KMO and laccase from Trametes versicolor, with 2,2^′-azino-bis(3-ethylbenzothiazoline-6-sulfonic acid) diammonium salt (ABTS) as the substrate, at relevant product concentrations. Due to this observed product inhibition, Km values with varying initial O2 concentrations, referred to as Km(O2), were determined as described for Km(S) above for KMO and laccase from T. versicolor with ABTS as the substrate (see Appendix A for details). Varying initial O2 concentrations were achieved by mixing air-equilibrated buffer (270±10µM O2) with N2-purged buffer (approx. 0 µM O2) or O2-purged buffer (1200±100µM O2). For all other enzymes, Km(O2) values were determined from complete O2 consumption experiments performed with the same type of enzyme assay either in air-equilibrated or O2-purged buffer.

Enzyme assays to determine 18O KIEs and λ values were performed in air-equilibrated buffer solutions with saturating concentrations of all other substrates (see Appendix A for details). As saturating substrate concentrations, we considered either 10 times the Km(S) value or a sufficiently high substrate concentration to limit the difference between the initial and final reaction rates (ν, determined with Eq.  and the corresponding Km(S) value) in the experiment to below 5 %. These enzyme assays were typically conducted in a 50 mL gas-tight glass syringe equipped with an optical oxygen sensor spot (PyroScience GmbH), an optical temperature sensor spot, and a small magnetic stir bar. Optical sensor spots were placed on the inside wall of the syringe, as close to the Luer-Lock tip as possible, and calibrated as described above for the fiber-optic oxygen sensors. These sensor spots allowed for a continuous, temperature-corrected measurement of O2 concentrations through the glass wall via an optical fiber. The syringe was filled completely with a buffer solution containing all required substrates. To start the reaction, a small volume of enzyme solution was added through the Luer-Lock tip with a gas-tight glass syringe. Immediately after enzyme addition, a stainless-steel needle (0.8 mm o.d.) was attached to the Luer-Lock tip. To limit exchange of O2 with the atmosphere, the needle was flushed with a few drops of assay solution and then pushed into a 12 mm thick chlorobutyl stopper. For experiments with diamine oxidase, the reaction was initiated by adding a small volume of substrate solution to assay solutions already containing the enzyme. Except during sampling, the syringe was placed on a magnetic stirring plate. Six sampling time points (t1 to t6) were determined from the continuously monitored O2 concentrations, typically at 200, 150, 120, 90, 70, and 50 µM remaining O2, corresponding to approx. 25 %–80 % O2 consumption. To sample, the needle was removed from the stopper and the first milliliter assay solution was discarded. The next 3–7 mL (depending on O2 concentration) was injected into 12 mL Exetainers (Labco Limited). Before starting an enzyme assay, Exetainers were closed with chlorobutyl septa, purged with He gas for 1 h, and amended with 100–200 µL of 2 M NaOH or 2–3 M HCl to stop enzymatic reactions in the added sample. To ensure equal headspace pressure in the Exetainers despite different sample volumes, Exetainer septa were pierced with a stainless-steel needle (0.45 mm o.d.), connected with a T-piece to a slow He flow and an open outlet submerged under 10 cm of water during sample injection. After sample injection, Exetainers were shaken and stored upside down until isotope analysis (see Sect. 2.4). Procedural blanks were prepared by transferring 1–7 mL of N2-purged water with a 50 mL gas-tight glass syringe from closed, over-pressured serum bottles into He-purged Exetainers containing NaOH or HCl solution, as described above for enzyme assay samples. Similarly, quantification standards (see Sect. 2.4 for details) were prepared by transferring 1–5 mL of air-equilibrated water with a 50 mL gas-tight glass syringe into He-purged Exetainers. For each experiment, one or more control samples were prepared by transferring 3 mL leftover assay solution without enzyme with a 10 mL gas-tight glass syringe into a He-purged Exetainer containing NaOH or HCl solution. These control samples were used to determine the concentration and isotopic composition of O2 at the start of the experiments (t0).

Some enzyme assays with choline, diamine, and glucose oxidase were also performed in 4–10 identically prepared 12 mL crimp-top vials per assay, as described recently (de Carvalho et al., 2024). Reactions were initiated by injecting a small volume of enzyme or substrate solution through the septa into filled vials. Prior to sampling, a fiber-optic oxygen microsensor (PyroScience GmbH) housed in a stainless-steel needle (0.5 mm o.d.) was inserted through the septa into the vials to measure the remaining O2 concentration. The oxygen sensor was calibrated as described above. After initiating the reaction and before measuring O2 concentrations, vials were shaken vigorously. To stop reactions at the desired degrees of O2 consumption, 3–7 mL assay solution was transferred into He-purged Exetainers that had been amended with 100–200 µL of 2 M NaOH or 2–3 M HCl. Procedural blanks, control samples, and quantification standards were prepared as described above. Experiments with diamine, choline, and glucose oxidase performed with the two different setups resulted in equal 18O KIEs and λ values, respectively, within error.

All samples, blanks, quantification standards, and controls were placed upside down on an orbital shaker at 125 rpm (revolutions per minute) for 1 h, prior to analysis by gas chromatography coupled to isotope-ratio mass spectrometry (GC–IRMS).

δ18O and δ17O values of O2 were measured in the headspace of Exetainers with a GasBench II coupled via a Conflo IV to a Delta V Plus isotope-ratio mass spectrometer (Thermo Fisher Scientific) as described recently (de Carvalho et al., 2024) and reported as per mil (‰ ± one standard deviation) relative to the international measurement standard Vienna Standard Mean Ocean Water (VSMOW) according to Eq. (), 5δhO=(hO/lO)sample(hO/lO)VSMOW-1, where (hO/lO)sample is the ratio of heavy (18O or 17O) to light (16O) isotopes in O2 in a sample, and (hO/lO)VSMOW is the ratio of heavy to light O isotopes in VSMOW. Briefly, seven 100 µL injections were made from each Exetainer headspace onto a 60 m Rt-Molsieve 5 Å PLOT column (Restek from BGB Analytik, 0.32 mm i.d., 30 µm film thickness), kept at 25 °C. Each GC–IRMS sequence consisted of 5–14 samples from enzyme assays, 10–12 procedural blanks, 5 quantification standards, and 3 air standards. Half of the blanks were measured at the beginning of the sequence and the other half at the end. Air standards were evenly distributed across the sequence and consisted of 150 µL ambient air in 12 mL of He. Air standards were used to verify instrument drift (which was never observed) and to perform a one-point calibration of the δ values to the VSMOW scale. The δ18O and δ17O values of O2 in air were assumed to be 23.8 ‰ and 12.1 ‰, respectively (Luz and Barkan, 2011; Laskar et al., 2019; Wostbrock et al., 2020). We recently showed that for δ18O values a one-point calibration is sufficient, while for δ17O values an additional correction factor must be used (de Carvalho et al., 2024). Procedural blanks were used to correct the measured δ values for blank contributions (Pati et al., 2016). Quantification standards were used to relate IRMS peak amplitudes to dissolved O2 concentrations and to correct δ values for instrument linearity (change in δ values with signal size) (Werner and Brand, 2001).

Initial O2 consumption rates were determined through linear regressions of the continuously measured O2 concentrations versus time during the initial, linear phase of enzyme assays. 18O KIEs and λ values were obtained from a single linear regression of all O2 isotope and concentration data from duplicate or triplicate enzyme assays according to Eqs. () and (), respectively. 6ln⁡δ18O+1δ18O0+1=118OKIE-1⋅ln⁡[O2][O2]07ln⁡(δ17O+1)=λ⋅ln⁡(δ18O+1) where [O2]0 and δ18O0 are the initial concentration and δ18O value of O2, respectively, measured in the control sample (see Sect. 2.3), and [O2], δ18O, and δ17O are the values measured in each enzyme assay sample at the different time points. All linear regressions were performed with Microsoft Excel, and errors are reported as 95 % confidence intervals. Km values were determined with a non-linear least square regression according to Eq. (), 8νt=νmax⁡⋅[i]tKm(i)+[i]t, where νt is the O2 consumption rate at a given time point t, νmax is the maximum O2 consumption rate of an enzymatic reaction, [i]t is the initial (t=0) nominal concentration of an organic substrate (S) or the measured concentration of O2 at time t, and Km(i) is the Michaelis constant determined under constant initial O2 and variable initial substrate concentration (Km(S)) or under constant initial substrate and variable initial O2 concentration (Km(O2)). For all Km(S) values, as well as for Km(O2) values determined for KMO and laccase from T. versicolor with ABTS, regressions were performed with initial rates of O2 consumption (ν0) from different experiments against the nominal initial organic substrate concentrations ([S]0) or against the measured initial O2 concentrations ([O2]0), respectively. For all other enzymes, where product inhibition was not detected, we determined Km(O2) values from the continuous measurement of O2 concentration over time ([O2]t) in a single enzyme assay, as described previously (Pati et al., 2022). For each time point, νt was calculated as the derivative of the measured [O2]t vs. t (i.e., Δ[O2]t/Δt) with Igor Pro software (WaveMetrics, Inc.). Km values and corresponding 95 % confidence intervals were determined with R software (R Core Team, 2023) using the MASS package (Venables and Ripley, 2002).

The flavin-dependent KMO was studied as an example for flavin monooxygenases, for which O2 reduction mechanisms have been well-described. Km(S) values for the native substrate L-kynurenine (0.012±0.003mM) and the co-substrate NADPH (0.009±0.001mM) were obtained from the literature (Crozier and Moran, 2007). Km(O2) and 18O KIEs were determined in experiments at optimal pH (7.5) and room temperature (23±1°C), with saturating concentrations (see Sect. 2.3 for details) of L-kynurenine (1 mM) and NADPH (0.5 mM), as well as 2 mM dithiothreitol to prevent loss of KMO activity (Crozier and Moran, 2007). A Km(O2) of 6±4µM was determined from initial rates of O2 consumption measured in 10 separate experiments with different initial O2 concentrations (25–260 µM) as shown in Fig. 1a. The 18O KIE and λ values were determined from the change in concentration, δ18O, and δ17O of O2 over time, measured in duplicate experiments. Figure 1b and c illustrate typical δ18O data from one experiment. The combined data from both experiments (see Sect. 2.5 for details) resulted in an 18O KIE of 1.0304±0.0003 and a λ value of 0.545±0.005.

Nine flavin-dependent oxidases were investigated. All of them convert O2 to H2O2 in the oxidative half-reaction and oxidize an organic substrate in the reductive half-reaction. Pyruvate oxidase was the only flavin-dependent oxidase that required cofactors for activity, namely, thiamine diphosphate, MnSO4, and FAD. Experiments with cholesterol oxidase were performed with the surfactant Thesit^® and isopropanol due to the low water solubility of the native substrate cholesterol. Experiments with glucose and alcohol oxidase were each performed with their native and an alternative substrate: glucose and mannose, in the case of glucose oxidase, and methanol and ethanol, in the case of alcohol oxidase. Experiments to determine 18O KIEs for alcohol, choline, and L-lysine oxidase were performed at two different initial O2 concentrations (260±10 and 1200±100µM).

Five copper-dependent oxidases were investigated, namely, laccases, L-ascorbate oxidase, and bilirubin oxidase, which convert O2 to water in the oxidative half-reaction, and diamine oxidase, which converts O2 to H2O2. All experiments were performed with native substrates, except experiments with laccase (see below), in a buffered, air-equilibrated solution at optimal pH and room temperature.

Laccases are multicopper oxidases that can oxidize a wide variety of substrates and lack a specific native substrate (Strong and Claus, 2011). Hydroquinone and ABTS were selected as substrates in this study, because they displayed different substrate-to-O2 consumption stoichiometries. Four ABTS molecules are required to reduce one molecule of O2, while only two hydroquinone molecules are required to reduce one molecule of O2 (see Appendix C). Despite these differences, laccase from T. versicolor yielded similar values for Km(S), Km(O2), and 18O KIEs irrespective of the substrate oxidized. With hydroquinone as the substrate, Km(S), Km(O2), and 18O KIEs were 0.23±0.04mM, 72±4µM, and 1.0196±0.0007, respectively. With ABTS as the substrate, Km(S), Km(O2), and 18O KIEs were 0.12±0.07mM, 47±59µM, and 1.0194±0.0005, respectively. Laccase from Agaricus bisporus exhibited a 10-fold higher Km(S) and a 2-fold lower Km(O2) with hydroquinone as the substrate compared to laccase from T. versicolor under identical conditions (see Table 2). However, the 18O KIEs were identical within error (1.0190±0.0002), and λ values ranged from 0.530±0.007 to 0.54±0.01 (see Table 2).

The three remaining copper-dependent oxidases, L-ascorbate, bilirubin, and diamine oxidase, displayed low Km(S) values between 0.14±0.05mM and 0.018±0.007mM (see Table 2). Km(O2) values decreased from 144±11µM for L-ascorbate oxidase to 73±3µM for bilirubin oxidase and 9.4±0.5µM for diamine oxidase. L-Ascorbate and diamine oxidase exhibited the lowest observed 18O KIEs of all enzymes in this study with 1.0086±0.0006 and 1.0103±0.0007, respectively, while bilirubin oxidase had an 18O KIE of 1.0223±0.0005. λ values ranged from 0.51±0.03 to 0.54±0.01 (see Table 2). During experimental assays with diamine oxidase, O2 production due to catalase contamination in the lyophilized diamine oxidase powder was detected. Catalase catalyzes the oxidation of H2O2 to H2O and O2, which could lead to inaccurate measurements of O2 consumption by diamine oxidase. To address this potential interference, 18O KIEs for diamine oxidase were determined in the presence of the catalase contamination alone and with the addition of excess horseradish peroxidase and ascorbic acid. Horseradish peroxidase catalyzes the oxidation of H2O2 and ascorbic acid to H2O and dehydroascorbic acid. In the presence of excess horseradish peroxidase, H2O2 was converted to H2O faster than catalase could reduce H2O2 to H2O and O2. The 18O KIEs determined for diamine oxidase were found to be identical within error, regardless of catalase activity (data not shown).

Cytochrome-c oxidase is a heme-copper-dependent oxidase, in which the heme a3 subunit initially binds O2 (Yoshikawa and Shimada, 2015). The Km(S) was not determined, but all experiments were performed with 25 µM cytochrome c and 3 mM ascorbic acid, to continuously reduce the product ferricytochrome c back to the substrate ferrocytochrome c. Km(S) values for ferrocytochrome c are reported to be 1.48 µM or lower (Merle and Kadenbach, 1982). Therefore, the 25 µM of cytochrome c used is considered a saturating substrate concentration. The Km(O2) was 3.3±0.5µM, the 18O KIE was 1.0189±0.0005, and the λ value was 0.543±0.009 (see Table 2).

Flavin-dependent O2-consuming enzymes utilize derivatives of the vitamin riboflavin as cofactors in their active sites. The organic flavin cofactor can be present in three different redox states: fully oxidized flavin (FL), radical flavin intermediate (FLH⚫), and fully reduced flavin (FLH2 or FLH-). The oxidation of FLH2 to FL releases two electrons and two protons, which can be used for the reduction of O2 (Massey, 2002), as illustrated in a simplified catalytic cycle in Fig. 3. The reduction of O2 by both flavin-dependent monooxygenases and oxidases starts with an outer-sphere single-electron transfer from FLH2 (or FLH-) to O2 forming FLH⚫ and O2⚫-. A recombination of the two radical species then forms a peroxyflavin intermediate (FLOO-), which can be protonated to a hydroperoxyflavin intermediate (FLOOH). In all known flavin-dependent monooxygenases, the (hydro)peroxyflavin can be detected and is responsible for substrate hydroxylation with concomitant O–O bond cleavage to form a hydroxyflavin (FLOH) (Massey, 2002) (see blue arrows in Fig. 3). In a subsequent step, FLOH reacts to FL by releasing H2O (see blue arrows in Fig. 3). In flavin-dependent oxidases, FLOO(H) has not been observed directly, and its formation remains a matter of ongoing debate (Massey, 2002). In addition to FL formation similar to the monooxygenation pathway (FLH⚫ and O2⚫- recombination to FLOO(H) and subsequent release of hydrogen peroxide), FL can also be formed through a sequence of outer-sphere electron and proton transfer steps from FLH⚫ to O2⚫- without covalent-bond formation between the flavin and O2 (see green arrows in Fig. 3) (Massey, 2002; Mattevi, 2006; Chaiyen et al., 2012). The reduction of FL to FLH2 is coupled with substrate or co-substrate oxidation in oxidases and monooxygenases, respectively, to complete the catalytic cycle (see Fig. 3).

In this study, we determined the first 18O KIE for a flavin-dependent monooxygenase, namely, KMO, which was 1.0305±0.0003. The magnitude of this isotope effect indicates that changes in bond order of O2 occur during the rate-limiting step of the reaction between KMO and O2, excluding O2 binding and product release as possible rate-limiting steps. Hence, the rate-limiting reaction step of KMO is the formation of O2⚫-, FLOO-, FLOOH, or S–OH and FLOH (see blue arrows in Fig. 3). 18O EIEs have been calculated for the reversible formation of O2⚫-, HO2-, H2O2, and two H2O from O2 as 1.033, 1.034, 1.009, and 0.968, respectively (Roth and Klinman, 2005). When comparing experimental 18O KIEs to calculated 18O EIEs, it is generally assumed that a measured 18O KIE (i) reflects intrinsic 18O KIEs of all electron and proton transfer steps up to, and including, the rate-limiting (i.e., first irreversible) step and (ii) is similar to, but not larger than, the 18O EIE calculated for the formation of the product/intermediate after the rate-limiting step (Roth and Klinman, 2005; Roth, 2007). Based on these 18O EIEs, the reduction of O2 by KMO is thus likely characterized by a rate-limiting O2⚫- or FLOO- formation.

Similar magnitudes of 18O KIEs compared to KMO have been determined in this study for glucose oxidase: 1.029±0.001 and 1.0341±0.0005 with D-glucose and D-mannose as the substrate, respectively. These values agree with previous studies of the same enzyme (Su and Klinman, 1998). Based on 18O EIEs, solvent isotope effects, and viscosity effects, Roth and Klinman (2003) suggested the initial outer-sphere electron transfer from FLH- to O2 to be the rate-limiting step of O2 reduction by glucose oxidase. The 18O KIEs determined for cholesterol, choline, and alcohol oxidase in this study were similar to, or lower than, those determined for KMO and glucose oxidase (1.0191–1.028; see Table 2). Because these isotope effects were still larger than the calculated 18O EIE for H2O2 or H2O formation (1.009 and 0.968, respectively), these enzymes likely also have a rate-limiting O2⚫- or FLOO- formation. In contrast, D-amino acid, L-lactate, L-lysine, pyruvate, and sarcosine oxidase exhibited much larger 18O KIEs (1.044–1.0565; see Table 2). These distinctively high 18O KIEs clearly suggest a different rate-limiting step than previously discussed, even though the first outer-sphere electron transfer to O2 has also been proposed as the rate-limiting step for D-amino acid oxidase (Kiss and Ferenczy, 2019). To date, 18O KIEs of this magnitude have only been measured for L-amino acid (1.0478) and D-amino acid oxidase (1.053) by Cheah et al. (2014). The only 18O EIE of similar magnitude was calculated for the formation of O22-, a two-electron reduction product of O2, with a value of 1.050 (Roth and Klinman, 2003).

Considering the two possible reaction mechanisms described in Fig. 3 for flavin-dependent oxidases, we suggest that glucose, cholesterol, choline, and alcohol oxidase (like KMO) reduce O2 through the formation of FLOO(H) with a rate-limiting formation of either O2⚫- or FLOO-. The same applies to glycolate oxidase with an 18O KIE of 1.023 (Guy et al., 1993; Ribas-Carbo et al., 1995; Cheah et al., 2014). However, for D-amino acid, L-amino acid, L-lactate, L-lysine, pyruvate, and sarcosine oxidase, we suggest the alternative O2 reduction mechanism, where FL is formed directly from FLH⚫ and O2⚫- without the formation of FLOO(H) (see green arrows in Fig. 3). Still, the exact nature of the rate-limiting step (a second single electron transfer, a proton-coupled electron transfer, or a hydrogen atom transfer) in this alternative O2 reduction mechanism cannot be inferred from the current experimental evidence. It is also possible that the rate-limiting step differs among the six oxidases with 18O KIEs between 1.044 and 1.057 or that the first electron transfer to O2 is partially rate-limiting in some of these enzymes, which could explain the lower-than-expected 18O KIEs for such a rate-limiting step.

For KMO, cholesterol, choline, and glycolate oxidase, as well as glucose oxidase with three different substrates, which we consider to share a common reaction mechanism, we found a tentative correlation between 18O KIEs and the corresponding Km(O2) values (see Fig. 4). The Km(O2) values for glucose oxidase with the substrate 2-deoxy-D-glucose and for glycolate oxidase were reported to be 25±5 and 210 µM, respectively (Macheroux et al., 1991; Roth and Klinman, 2003). Based on the limited number of data points, we do not consider the correlation to be necessarily linear as shown in Fig. 4. However, the data clearly indicate that enzymes with lower Km(O2) values have higher 18O KIEs, ranging from choline oxidase with a Km(O2) of 298±20µM and an 18O KIE of 1.0194±0.0006 to glucose oxidase with D-mannose as the substrate with a Km(O2) of 3.9±0.6µM and an 18O KIE of 1.0341±0.0005. Since 18O KIEs reflect the ratios of reaction rates of the different O2 isotopologues, a correlation between 18O KIE and Km(O2) only makes sense when we consider the kinetic properties of the Michaelis constant (Northrop, 1998). In O2-consuming enzymes, O2 typically binds to the enzyme after binding of the organic substrate (oxygenases) or in a ping-pong mechanism (oxidases) (Malmstrom, 1982; Romero et al., 2018). Thus, we can describe the consumption of O2 by these enzymes kinetically with a two-step reaction, where O2 first binds reversibly to the enzyme, followed by an irreversible reduction step of O2. In such a case, the measured 18O KIE depends on the intrinsic 18O KIE and 18O EIE of the O2 binding step, the 18O KIE of the irreversible reduction step, and the forward commitment to catalysis. This commitment factor is the ratio of two elementary reaction rates, namely, the rate of the irreversible reduction step divided by the rate of the backward reaction of O2 binding (see Appendix D for details). In fact, as long as the reduction step is slower than the backward binding step, and thus the commitment factor is below 1, the measured 18O KIE will show an apparently linear trend with an increasing commitment factor, similar to the trend observed in Fig. 4. For this set of enzymes, it thus appears that Km(O2) is a proxy for the forward commitment to catalysis or the extent to which O2 binding contributes to the overall reaction rate. One can indeed mathematically relate Km(O2) to the commitment factor, as shown in Appendix D, and reconcile the observed decrease in 18O KIE with increasing Km(O2) values if (i) O2 binding and unbinding is faster than O2 reduction for all enzymes but with different degrees of forward commitment, (ii) the intrinsic 18O KIE for O2 reduction is larger than for O2 binding while all intrinsic isotope effects are close to identical for these enzymes, and (iii) the dissociation constant (the ratio of backward and forward reaction rates of O2 binding) of these enzymes varies much less than Km(O2). If O2 binding does not contribute to the overall rate, the apparent 18O KIE is expected to reflect the intrinsic 18O KIE of the rate-limiting O2 reduction step. Accordingly, the intrinsic 18O KIE for the rate-limiting step of O2⚫- or FLOO- formation is likely between 1.030 and 1.035, based on both calculated 18O EIEs for these reactions (1.033–1.034) (Roth and Klinman 2003) and on the maximum 18O KIEs observed for glucose oxidase (1.0341±0.0005) and KMO (1.0304±0.0003). The lower 18O KIEs (1.019–1.023), particularly for cholesterol, choline, and glycolate oxidase, can thus still arise from a rate-limiting O2⚫- or FLOO- formation but with increasing contributions from a relatively slower O2 binding to the overall reaction rate that is likely associated with an intrinsic isotope effect close to unity because, upon binding, no bond changes occur in O2.

Alcohol oxidase, with either methanol or ethanol as the substrate, was the only enzyme with 18O KIEs close to or below 1.03 that did not follow the observed trend between Km(O2) values and 18O KIEs discussed above. Km(O2) values of alcohol oxidase (1017±93 and 901±200) were substantially larger than Km(O2) values of all other flavin-dependent enzymes studied, except for L-lysine oxidase, which likely has a different O2 reduction mechanism (formation of O22-). Interestingly, alcohol oxidase was the only enzyme tested in this study that exhibited particularly low λ values between 0.483±0.007 and 0.488±0.009 (see Table 2). These values are not only lower than typical λ values (0.51–0.53) but also significantly lower than λ values observed for all other enzymes in this study, which ranged from 0.51±0.03 to 0.547±0.002 (see Table 2). We note that λ values determined for the majority of enzymes in this study are close to but slightly higher than previously determined λ values of 0.51–0.53 for biological O2 consumption (Young et al., 2002; Luz and Barkan, 2005; Ash et al., 2020; Hayles and Killingsworth, 2022). It is possible that the applied δ17O scale correction factor from de Carvalho et al. (2024) leads to a slight overestimation of λ values. Regardless of this uncertainty in the δ17O scale correction factor, the λ values determined for alcohol oxidase are clearly much lower than any previously determined λ values for biological O2 consumption and significantly lower than those for any other enzyme studied here. This difference in λ values suggests a unique O2 reduction mechanism for alcohol oxidase, differing from the mechanism proposed for enzymes that exhibit a correlation between 18O KIEs and Km(O2) values. However, this reduction mechanism cannot be further elucidated in this study.

Unlike flavin-dependent O2-consuming enzymes, which have a relatively conserved active site and catalytic mechanism, iron- and copper-dependent O2-consuming enzymes are known to employ a wide variety of different active-site structures and catalytic mechanisms (Costas et al., 2004; Blank et al., 2010; Liu et al., 2014; Solomon et al., 2014; Huang and Groves, 2018). For the five copper-dependent oxidases tested in this study and for six (out of seven) copper-dependent monooxygenases and oxidases examined in previous research, the 18O KIEs grouped closely around two main values. Namely, L-ascorbate and diamine oxidase from this study, as well as tyrosinase, bovine serum amine oxidase, and amine oxidase from Hansenula polymorpha, were characterized by 18O KIEs between 1.0086 and 1.011 (see Tables 2 and S1 in the Supplement) (Feldman et al., 1959; Su and Klinman, 1998; Welford et al., 2007). Conversely, the 18O KIEs of bilirubin oxidase and the two laccases from this study, as well as peptidylglycine monooxygenase, dopamine β-monooxygenase, and galactose oxidase, ranged between 1.0173 and 1.223 (see Tables 2 and S1) (Tian et al., 1994; Francisco et al., 2003; Humphreys et al., 2009). The only copper-dependent enzyme studied so far that fell in between these two clusters is pea-seedling amine oxidase with an 18O KIE of 1.014±0.001 (Mukherjee et al., 2008). The two groups of copper-dependent enzymes defined by the two groups of 18O KIE values both contain a mix of monooxygenases and oxidases (see Fig. 5). The monooxygenases peptidylglycine monooxygenase, dopamine β-monooxygenase, and tyrosinase catalyze the incorporation of one O atom from O2 into their substrate. Multicopper oxidases, including laccase, L-ascorbate oxidase, and bilirubin oxidase, reduce O2 to two H2O molecules. The cofactor-dependent mononuclear copper enzymes (copper amine oxidases including diamine oxidase and galactose oxidase) reduce O2 to H2O2 (Mure et al., 2002; Humphreys et al., 2009). Despite these differences, all copper-dependent O2-consuming enzymes form common copper–oxygen intermediates, namely, copper-superoxo (Cu(II)-OO⚫), copper-peroxo (Cu(II)-OO-), and copper-hydroperoxo (Cu(II)-OOH) species. Figure 6 shows the electron and proton transfer steps involved in the formation of these intermediates. 18O EIEs for the reversible formation of these three copper–oxygen species have been determined to be 1.009–1.010 for copper-superoxo, 1.018–1.031 for copper-peroxo, and 1.025–1.026 for copper-hydroperoxo intermediates (Mukherjee et al., 2008; Humphreys et al., 2009). The copper-dependent enzymes that exhibited 18O KIEs between 1.0173 and 1.223 are thus likely characterized by a rate-limiting step involving the formation of a copper-peroxo or copper-hydroperoxo intermediate. Accordingly, studies of peptidylglycine and dopamine β-monooxygenase, which exhibited 18O KIEs of 1.0173±0.0009 and 1.0197±0.0003, respectively, suggested a rate-limiting hydrogen atom abstraction by a copper-superoxo intermediate to form a copper-hydroperoxo species (Evans et al., 2003; Osborne and Klinman 2011). The 18O KIE of 1.019±0.001 determined for galactose oxidase by Humphreys et al. (2009) was also attributed to a rate-limiting hydrogen atom abstraction by a copper-superoxo intermediate. The rate-limiting steps of multicopper oxidases, such as bilirubin oxidase and laccase, have not been firmly established. However, based on the 18O KIEs determined in this study and the comparison with the three enzymes with similar 18O KIEs, a rate limiting copper-hydroperoxo formation by hydrogen atom abstraction seems likely. Similarly, the copper-dependent enzymes that displayed 18O KIEs between 1.0086 and 1.011 are likely characterized by a rate-limiting copper-superoxo formation, based on comparisons with 18O EIEs (1.009–1.010). Accordingly, copper-superoxo formation has been suggested as the rate-limiting step for bovine serum amine oxidase and amine oxidase from H. polymorpha (Su and Klinman, 1998; Mills et al., 2002). It can thus be assumed that tyrosinase, as well as L-ascorbate and diamine oxidase, also has a rate-limiting step involving the formation of a copper-superoxo intermediate. For pea-seedling amine oxidase, for which an 18O KIE of 1.014±0.001 was determined (Mukherjee et al., 2008), a rate-limiting step involving copper-peroxo formation has also been proposed. However, the preceding copper-superoxo formation is partially rate-limiting, which acts to lower the observed 18O KIE value from the expected 18O EIE range of 1.018–1.031 (Mukherjee et al., 2008).

The 18O KIE of 1.0189±0.0005 determined here for cytochrome-c oxidase is consistent with previous reports from the literature (Ribas-Carbo et al., 1995; Cheah et al., 2014). Cytochrome-c oxidase is a hetero-di-nuclear copper-heme oxidase, in which a copper and a heme-iron are involved in the O2-reduction mechanism (Yoshikawa and Shimada, 2015). Iron-dependent enzymes form similar reactive oxygen intermediates, as described above for copper-dependent enzymes, including iron-superoxo (Fe(III)-OO⚫) and iron-hydroperoxo (Fe(III)-OOH) intermediates (see Fig. 6). In addition, iron can be oxidized further in certain active-site structures to a high-valent iron-oxo (Fe(IV)=O) intermediate. Calculated or measured 18O EIEs are also similar in magnitude, with 1.004–1.009 for iron-superoxo formation, 1.011–1.017 for iron-hydroperoxo formation, and 1.029 for iron-oxo formation (Tian and Klinman, 1993; Mirica et al., 2008). Previous studies have determined 18O KIEs for 12 iron-dependent O2-consuming enzymes showing a continuous range from 1.009±0.001 for soybean lipoxygenase (Guy et al., 1992) to 1.0281±0.0004 for alternative oxidase (Cheah et al., 2014). Observed 18O KIEs for iron-dependent enzymes have consistently reflected the intrinsic 18O KIE of the rate-limiting step, with increasing 18O KIEs indicating a higher degree of O2 reduction. For example, the 18O KIE of soybean lipoxygenase (1.009–1.012) reflects a rate-limiting electron transfer to O2 to form an iron-superoxo species (Guy et al., 1992; Knapp and Klinman, 2003). The 18O KIE of 1.015±0.001 determined for hydroxyethyl phosphonate dioxygenase reflects a rate-limiting iron-hydroperoxo formation by hydrogen atom abstraction (Zhu et al., 2015). Finally, the 18O KIE of 1-aminocyclopropyl-1-carboxylic acid oxidase (1.0215±0.005) reflects a rate-limiting iron-oxo formation (Mirica et al., 2008). For cytochrome-c oxidase, a rate-limiting hydrogen atom abstraction by an iron-bound superoxo species with concomitant O–O bond cleavage and formation of a high-valent iron-oxo intermediate has been suggested (Yoshikawa and Shimada, 2015). The corresponding 18O KIE of 1.0189±0.0005 determined in this study is in agreement with both a hydrogen atom abstraction by a metal-superoxo species, as seen for many of the copper-dependent enzymes, as well as with the formation of a high-valent iron-oxo species as described for 1-aminocyclopropyl-1-carboxylic acid oxidase.