C−H Bond Cleavage Is Rate-Limiting for Oxidative C−P Bond Cleavage by the Mixed Valence Diiron-Dependent Oxygenase PhnZ

Abstract

PhnZ utilizes a mixed valence diiron(II/III) cofactor and O₂ to oxidatively cleave the carbon–phosphorus bond of (R)-2-amino-1-hydroxyethylphosphonic acid to form glycine and orthophosphate. The active site residues Y24 and E27 are proposed to mediate induced-fit recognition of the substrate and access of O₂ to one of the active site Fe ions. H62 is proposed to deprotonate the C1-hydroxyl of the substrate during catalysis. Kinetic isotope effects (KIEs), pH–rate dependence, and site-directed mutagenesis were used to probe the rate-determining transition state and the roles of these three active site residues. Primary deuterium KIE values of 5.5 ± 0.3 for ^D(V) and 2.2 ± 0.4 for ^D(V/K) were measured with (R)-2-amino[1-²H₁]-1-hydroxyethylphosphonic acid, indicating that cleavage of the C1−H bond of the substrate is rate-limiting. This step is also rate-limiting for PhnZ Y24F, as shown by a significant deuterium KIE value of 2.3 ± 0.1 for ^D(V). In contrast, a different reaction step appears to be rate-limiting for the PhnZ E27A and H62A variants, which exhibited ^D(V) values near unity. A solvent KIE of 2.2 ± 0.3 for ^D₂O(V) is observed for PhnZ. Significant solvent KIE values are also observed for the PhnZ Y24F and E27A variants. In contrast, the PhnZ H62A variant does not show a significant solvent KIE, suggesting that H62 is mediating proton transfer in the transition state. A proton inventory study with PhnZ indicates that 1.5 ± 0.6 protons are in flight in the rate-determining step. Overall, the rate-determining transition state for oxidative C–P bond cleavage by PhnZ is proposed to involve C–H bond cleavage that is coupled to deprotonation of the substrate C1-hydroxyl by H62.

Inorganic phosphate (P_i) is an essential metabolic nutrient. In marine and aquatic environments, P_i concentrations can fall to low nanomolar, and occasionally picomolar, concentrations.^1,2 Such low P_i levels cannot adequately sustain microbial life, which has led to the evolution of enzyme pathways that enable the utilization of organophosphonic acids (Pns), phosphite, and hypophosphite as alternative sources of P_i. A number of simple Pns are actively synthesized by bacteria for incorporation into lipids, exopolysaccharides, and glycoproteins. Such commonly occurring Pns include 2-aminoethylphosphonic acid, methylphosphonic acid, 2-hydroxyethylphosphonic acid, hydroxymethylphosphonate, and phosphonoacetate.^3–5 These compounds are in turn scavenged by other bacteria for catabolism into P_i. To liberate P_i from Pns, microbes have developed three distinct mechanisms for cleaving the C–P bond.⁶ The heterolytic mechanism is represented by phosphonatase, phosphonoacetate hydrolase, and phosphonopyruvate hydrolase, where water or an enzyme nucleophile performs in-line attack on the phosphorus center to displace a carbanion. Such enzymes are specific for substrates that have a carbonyl group in the β position relative to the phosphorus center to allow resonance stabilization of the carbanion leaving group. A homolytic mechanism is utilized by the carbon–phosphorus lyase pathway. C–P lyase can process a wide array of Pn structures, including those in which the alkyl group is not activated for heterolytic cleavage. Although the name implies one enzyme, C–P lyase features multiple enzymatic steps and a multiprotein complex; in Escherichia coli, this pathway is encoded by the 14-cistron phn operon (phnCDEFGHIJKLMNOP). Radical-based cleavage of the C–P bond is performed by the PhnJ component, an outlying member of the radical SAM superfamily. The third mechanism of C–P bond cleavage is represented by the iron-dependent oxygenases HF130PhnY* and HF130PhnZ (herein termed simply PhnY* and PhnZ, respectively). PhnY* is an α-ketoglutarate/non-heme Fe(II)-dependent dioxygenase that catalyzes the stereospecific α-hydroxylation of 2-aminoethylphosphonic acid to form (R)-2-amino-1-hydroxyethylphosphonic acid [(R)-1]. The C–P bond of (R)-1 is then cleaved by the diiron-dependent oxygenase PhnZ (UniProtKB D0E8I5.1) to form P_i and glycine.⁷ Genes encoding homologues of PhnY* and PhnZ are commonly found in bacterioplankton that inhabit marine environments with low nanomolar P_i concentrations, often at frequencies that rival genes encoding hydrolytic or carbon–phosphorus lyase pathways.^8–10 Recent studies have shown that these other pathways allow utilization of different organophosphonates as P_i sources. For example, a PhnY*/PhnZ pathway from Gimesia maris is specific for oxidative catabolism of methylphosphonic acid,¹¹ while the TmpA/TmpB pathway from Leisingera caerulea is specific for 2-(trimethylammonio) ethylphosphonic acid.¹²

PhnZ is a member of a novel family of mixed valence diiron oxygenases (MVDOs).¹³ The founding member of this family is the mammalian enzyme myo-inositol dioxygenase, which oxidatively cleaves the C1−C6 bond of myo-inositol to form d-glucuronate.¹⁴ MVDOs comprise a subfamily within the histidine-aspartate (HD) motif superfamily of phosphohydrolases. However, instead of using metal ions to activate water for hydrolysis, MIOX and PhnZ use a pair of iron atoms in a mixed valence [Fe(II)/Fe(III)] state to activate O₂ for oxidation of C–C and C–P bonds. Unlike other diiron oxygenases, such as soluble methane monooxygenase, the mixed valence state is favored by MIOX and PhnZ over the diferrous state.¹³ Moreover, no external electrons are required for catalysis, as all necessary electrons for restoring the active Fe(II)/Fe(III) state are obtained from the substrate. The two active site Fe ions, denoted Fe1 and Fe2, are bound by a set of four histidines and two aspartates that are conserved within the HD superfamily. Both MIOX and PhnZ use Fe1 to activate O₂, while Fe2 is reserved for substrate binding. However, PhnZ deviates from MIOX in the presence of three distinct residues, Y24, E27, and H62, that are proposed to have roles in catalysis and substrate-induced activation of O₂.¹⁵ Y24 and E27 reside on a flexible loop that responds to binding of (R)-1 to Fe2 (Scheme 1). In the absence of a substrate, Y24 blocks the binding of O₂ to Fe1. When (R)-1 binds to Fe2, E27 flips into the active site to form an electrostatic interaction with the 2-amino group of the substrate. This results in Y24 being removed from the active site and creates a vacant ligand site on Fe1 for O₂. (R)-1 binds to Fe2 in a bidentate fashion using the C1-hydroxyl and one of the phosphonyl oxygens. The C1-hydroxyl makes an additional interaction with H62, which appears as a distinctive fifth conserved histidine in PhnZ homologues. Mutagenesis of H62 and E27 has a significant impact on PhnZ catalysis, while Y24 is largely dispensable for activity.¹⁵

Scheme 1.

The nature of the diiron oxo species formed by PhnZ and the relative free energy barriers for the bond breaking steps are of great interest due to the stabilities of the C–H and C–P bonds in (R)-1. The C–H bond of a secondary alcohol as found in (R)-1 is estimated to have a bond dissociation energy of 94 kcal mol⁻¹,¹⁶ while that of a phosphonate C–P bond involving a sp³ hybridized carbon is approximately 85 kcal mol⁻¹.¹⁷ Cleavage of unactivated C–H bonds was once thought to be the preserve of high-valence iron oxygen complexes, such as the Fe(IV)═O species formed by Fe(II)/2-oxoglutarate-dependent oxygenases¹⁸ or the dioxygen–Fe(IV)/Fe(IV) species (intermediate Q) used by soluble methane monooxygenase to cleave the C–H bond of methane.¹⁹ However, MVDOs such as MIOX²⁰ and a number of mononuclear non-heme oxygenases have now been shown or are proposed to cleave C–H bonds using ferric superoxo species.²¹ Both MIOX and PhnZ are proposed to initiate their respective reactions with C–H bond cleavage by a diiron(III/III)–superoxo species.¹³ Density functional theory (DFT)/quantum mechanics (QM) calculations predict C1−H bond cleavage to be the rate-limiting step for PhnZ.²² This is in contrast to MIOX, in which freeze-quench electron paramagnetic resonance (EPR) spectroscopic studies indicate that turnover of an advanced intermediate “H” is rate-limiting.²⁰ However, pre-steady-state accumulation and decay of the diiron(III/III)–superoxo species could be observed on MIOX using deuterium-labeled myo-inositol, implicating this species in C–H bond abstraction.

In this study, we examined the mechanism of PhnZ through a combination of deuterium kinetic isotope effects (D-KIEs), solvent kinetic isotope effects (SKIEs), site-directed mutagenesis, and pH–rate analysis. We show that C–H bond cleavage is rate-limiting for PhnZ during oxidative C–P bond cleavage of (R)-1. Moreover, we show that H62 mediates proton transfer in the rate-determining transition state.

Experimental Methods

General

All reagents were purchased from Sigma-Aldrich or BioShop Canada, Inc., unless otherwise specified, and used without further purification. The malachite green reagent was prepared as described by Lanzetta and co-workers.²³ Wild type PhnZ and the Y24F, E27A, and H62A variants were expressed and purified as described previously.¹⁵ The synthesis of (R)-1 has been described previously.¹⁵

¹H, ¹³C, and ³¹P nuclear magnetic resonance (NMR) spectra were recorded in CDCl₃ or D₂O on a Bruker Avance AV III 400 (¹H, 400.27 MHz; ¹³C, 100.65 MHz; ³¹P, 162.03 MHz) spectrometer. Proton NMR chemical shifts were referenced to residual CHCl₃ (δ_H 7.24) or HOD (δ_H 4.80). ¹³C NMR spectra were referenced to CDCl₃ (δ_C 77.23), and ³¹P NMR spectra to external H₃PO₄ (85%) (δ_P 0.0). All ¹³C NMR spectra were recorded j-modulated. Chemical shifts (δ) are given in parts per million, and coupling constants (J) in hertz. Optical rotations were measured at 20 °C on a PerkinElmer 341 polarimeter in a 1 dm cell. [α]_D values are given in 10⁻¹ deg cm² g⁻¹. Melting points were determined on a Leica Galen III Reichert Thermovar instrument and are uncorrected. Thin layer chromatography (TLC) was carried out on 0.25 mm thick MilliporeSigma plates coated with silica gel 60 F₂₅₄. Spots were visualized with ultraviolet light and/or by dipping the plate into a solution of (NH₄)₆Mo₇O₂₄·4H₂O (25.0 g) and Ce(SO₄)₂·4H₂O (1.0 g) in aqueous 10% H₂SO₄ (500 mL), followed by heating with a heat gun, if not stated differently. Flash (column) chromatography was performed with Macherey-Nagel silica gel 60 (230–400 mesh).

Synthesis of (R)-[1-²H₁]-2-Amino-1-hydroxyethylphosphonic Acid {(R)-[1-²H₁]-1}

An overview of the synthesis of (R)-[1-²H₁]-1 is shown in Scheme 2.

Scheme 2.

(±)-Diisopropyl [1-²H₁]-(2-Benzyloxy-1-hydroxyethyl)-phosphonate {(±)-[1-²H₁]-4}.^24,25

Dry dimethyl sulfoxide (3.125 g, 40.0 mmol, 2.84 mL), dissolved in dry CH₂Cl₂ (8 mL), was added dropwise to a solution of oxalyl chloride (2.539 g, 20.0 mmol, 1.72 mL) in dry CH₂Cl₂ (90 mL) at −78 °C under argon. The resulting clear solution was stirred for 10 min before the addition of [1,1-²H₂]-2-benzyloxy-1-ethanol {[1,1-²H₂]-2} (2.369 g, 15.4 mmol), dissolved in CH₂Cl₂ (11 mL). Stirring was continued for 20 min before the addition of Et₃N (4.048 g, 40.0 mmol, 5.54 mL). The reaction mixture was allowed to slowly warm to −35 °C in the cooling bath. Then, aqueous citric acid [50 mL, 39% (w/w)] was added, and the cooling bath was removed. The organic phase was separated and washed with aqueous citric acid (3 × 60 mL), water (2 × 60 mL), and a saturated aqueous saturated aqueous NaHCO₃ (3 × 10 mL). The organic phase was dried (Na₂SO₄) and concentrated in vacuo. Diisopropyl trimethylsilyl phosphite (3.789 g, 15.9 mmol) was added dropwise to a solution of the crude aldehyde [1-²H₁]-3 (1.842 g, 12.2 mmol) in dry CH₂Cl₂ (37 mL) under argon at room temperature. The reaction mixture was stirred for 16 h and then concentrated in vacuo without being heated. The crude silylated intermediate was purified by flash chromatography (1:1 hexanes/EtOAc; R_f = 0.50) to remove residual diisopropyl phosphite. The purified silylated phosphonate (2.941 g, 7.5 mmol, 62%) was dissolved in iPrOH (10 mL). HCl (12 M, 7 drops) was added, and the reaction mixture was stirred at room temperature for 1.5 h. Then, the solution was concentrated, and the residue was purified by flash chromatography (EtOAc; R_f = 0.39) to yield the desired racemic alcohol (±)-[1-²H₁]-4 (2.057 g, 6.5 mmol, 53%) as a colorless oil: ¹H NMR (CDCl₃, 400.27 MHz) δ 7.35–7.22 (m, CH_arom, 5H), 4.72 (m symm., 2 × ¹H, 2 × CH of iPr), 4.56 (AB system; A part, d, J = 11.9 Hz, ¹H; B part, d, J = 11.9 Hz, ¹H, CH₂ of Bn), 3.73 (ABP system; A part, dd, J = 10.4, 10.3 Hz, ¹H; B part, dd, J = 10.4, 5.5 Hz, ¹H; CH₂-CH-P), 3.24 (broadened d, J = 10.1 Hz, ¹H, OH), 1.30 (d, J = 6.0 Hz, 3H, CH₃), 1.29 (d, J = 6.2 Hz, 3H, CH₃), 1.28 (d, J = 6.0 Hz, 3H, CH₃), 1.26 (d, J = 6.2 Hz, 3H, CH₃); ³¹P NMR (CDCl₃, 162.03 MHz) δ 20.04 (s); ¹³C NMR (CDCl₃, 100.65 MHz) δ 137.9 (s, C_arom), 128.6 (s, 2 × CH_arom), 128.00 (s, 2 × CH_arom), 127.95 (s, CH_arom), 73.6 (s, CH₂ of Bn), 71.6 (d, J = 6.9 Hz, CH of iPr), 71.5 (d, J = 7.1 Hz, CH of iPr), 70.3 (d, J = 6.4 Hz, CH₂-CH-P), 68.0 (dt, J = 162.2, 21.6 Hz, CD-P), 24.31 (d, J = 2.8 Hz, CH₃), 24.28 (d, J = 3.1 Hz, CH₃), 24.11 (d, J = 2.4 Hz, CH₃), 24.06 (d, J = 2.5 Hz, CH₃).

(±)-Diisopropyl [1-²H₁]-(1,2-Dihydroxyethyl)phosphonate {(±)-[1-²H₁]-5}

HCl (12 M, 3 drops) and Pd/C (10% Pd, 0.242 g) were added to (±)-[1-²H₁]-4 (2.057 g, 6.5 mmol), dissolved in dry ethanol (30 mL). Then the flask was filled with hydrogen gas to a final pressure of 3.4 bar and shaken (using a Parr apparatus) at room temperature for 8 h. The catalyst was removed by filtration (Celite, moistened with ethanol), and the solvent was removed in vacuo. The crude product (±)-[1-²H₁]-5 (1.469 g, 6.5 mmol, quant.) was dissolved in toluene (15 mL) and concentrated again. It was used for the following step without further purification. An analytical sample was purified by flash chromatography (EtOAc; R_f = 0.10) for characterization: ¹H NMR (CDCl₃, 400.27 MHz) δ 4.97–4.63 (m, 3H, OH + 2 × CH of iPr), 3.93–3.76 (m, 3H, CH₂ + OH), 1.30 (d, J = 6.2 Hz, 6 H, 2 × CH₃), 1.29 (d, J = 6.2 Hz, 6H, 2 × CH₃); ³¹P NMR (CDCl₃, 162.03 MHz) δ 21.42 (s); ¹³C NMR (CDCl₃, 100.65 MHz) δ 71.9 (d, J = 7.0 Hz, CH of iPr), 71.8 (d, J = 7.1 Hz, CH of iPr), 69.2 (dt, J = 160.5, 21.8 Hz, CD-P), 62.9 (d, J = 4.9 Hz, CH₂), 24.3 (d, J = 3.5 Hz, 2 × CH₃), 24.1 (d, J = 4.9 Hz, 2 × CH₃).

(±)-Diisopropyl [1-²H₁]-(2-Azido-1-hydroxyethyl)-phosphonate {(±)-[1-²H₁]-6}

DIAD (0.748 g, 3.7 mmol, 0.73 mL) was added to a solution of Ph₃P (0.970 g, 3.7 mmol) in dry toluene (13 mL) at 0 °C under argon. The diol (±)-[1-²H₁]-5 (0.649 g, 2.9 mmol) was dissolved in a solution of hydrazoic acid (1.03 M in toluene, 3.59 mL) and added to the reaction mixture after being stirred for 5 min. The cooling bath was removed, and the reaction mixture was stirred at 40 °C for 2 h before the addition of another portion of HN₃ (1.03 M in toluene, 1.13 mL). Stirring at 40 °C was continued for 45 min before the addition of dry methanol (10 drops). The reaction mixture was concentrated, and the residue was purified by flash chromatography (EtOAc; R_f = 0.41) to give azide (±)-[1-²H₁]-6 (0.499 g, 1.9 mmol, 66%) as an oily liquid of acceptable purity: ³¹P NMR (CDCl₃, 162.03 MHz) δ 28.97 (s, O═PPh₃, 10 mol %), 19.08 (s, desired product, 85 mol %), 4.42 [s, HP(O)(OiPr)₂, 5 mol %]. The product was directly acetylated without any further purification.

(±)-Diisopropyl [1-²H₁]-(2-Azido-1-acetoxyethyl)-phosphonate {(±)-[1-²H₁]-7}

Azide (±)-[1-²H₁]-6 (1.10 g, 4.2 mmol) was dissolved in dry pyridine (2.088 g, 26.4 mmol, 2.13 mL). Acetic anhydride (0.898 g, 8.8 mmol, 0.83 mL) was added, and the reaction mixture was stirred at 50 °C for 1.5 h. The solvent was removed, and water (2 mL) was added to the residue. The resulting mixture was stirred for 10 min before the addition of more water (10 mL). The resulting slurry was extracted with CH₂Cl₂ (3 × 10 mL). The combined organic fractions were washed with HCl (2 M, 2 × 10 mL), followed by an aqueous NaHCO₃ solution (saturated, 3 × 10 mL), dried (Na₂SO₄), and concentrated in vacuo. The residue was purified by flash chromatography (1:1 hexanes/EtOAc; R_f = 0.29) to give the desired racemic acetoxyphosphonate (±)-[1-²H₁]-7 (1.153 g, 3.9 mmol, 93%) as a colorless oil of sufficient purity for chiral resolution.

Chiral Resolution of (±)-Diisopropyl [1-²H₁]-(2-Azido-1-acetoxyethyl)phosphonate {(±)-[1-²H₁]-7}.²⁶

Phosphate buffer (50 mM, pH 7, 30 mL) and lipase (from Rhizomucor miehei, ≥20000 units/g from Novozyme, 1.5 mL) were mixed, and the pH of the resulting solution was adjusted to 7.0 with acetic acid (0.1 M). Deuterated acetoxy-azidophosphonate (±)-[1-²H₁]-7 (1.094 g, 3.7 mmol), dissolved in a mixture of hexanes (7.5 mL) and tert-butylmethyl ether (2.5 mL), was added to the mixture. Acetic acid that was released during the hydrolysis reaction was neutralized with equimolar amounts of NaOH (0.5 M) using an autotitration unit. The reaction was stopped by addition of HCl (2 M) to a final pH of 3.0 at a conversion of 53% (25.5 h reaction time). The organic phase was separated, and the aqueous phase was extracted with EtOAc (3 × 10 mL). The combined organic phases were washed with brine (3 × 10 mL), dried (Na₂SO₄), and concentrated in vacuo. The two products were separated by flash chromatography [2:1 EtOAc/hexanes → EtOAc; R_f (acetate, 2:1 EtOAc/hexanes) = 0.42, R_f (alcohol, 2:1 EtOAc/hexanes) = 0.21] to give deuterated alcohol (S)-[1-²H₁]-6 (0.447 g, 1.8 mmol, 95%) and acetate (R)-[1-²H₁]-7 (0.478 g, 1.6 mmol, 88%). Acetate (R)-[1-²H₁]-7: ¹H NMR (CDCl₃, 400.27 MHz) δ 4.81–4.68 (m, 2H, 2 × CH of iPr), 3.58 (broadened d, J = 6.4 Hz, 2H, CH₂), 2.15 (s, 3H, CH₃ of acetyl), 1.35–1.29 (m, 12H, 4 × CH₃ of iPr); ³¹P NMR (CDCl₃, 162.03 MHz) δ 14.72 (s, desired product, 94 mol %), 14.08 (s, unknown impurity, 6 mol %); [α]_D²⁰ = −19.4 [c 1.165, acetone; lit. value¹⁵ of the nondeuterated compound, [α]_D²⁰ = −14.3 (c 1.000, acetone)].

(R)-Diisopropyl [1-²H₁]-(2-Azido-1-hydroxyethyl)-phosphonate {(R)-[1-²H₁]-6}

Water (0.8 mL) and Et₃N (2.191 g, 21.6 mmol, 3.0 mL) were added to acetate (R)-[1-²H₁]-7 (0.455 g, 1.5 mmol) in dry MeOH (15 mL). The resulting mixture was stirred at room temperature for 19 h and concentrated in vacuo, and the residue was purified by flash chromatography (2:1 EtOAc/hexanes; R_f = 0.24) to give (R)-[1-²H₁]-6 (0.382 g, 1.5 mmol, quant.) as a colorless oil: ¹H NMR (CDCl₃, 400.27 MHz) δ 4.83–4.68 (m, 2H, 2 × CH of iPr), 3.50 (ABP system; A part, dd, J = 13.1, 6.1 Hz, 1H; B part, dd, J = 13.1, 7.7 Hz, 1H; CH₂), 3.44–3.33 (m, 1H, OH), 1.36–1.30 (m, 12 H, 4 × CH₃); ³¹P NMR (CDCl₃, 162.03 MHz) δ 19.06 (s, desired product, 94 mol %), 17.93 (s, unknown impurity, 6 mol %).

(R)-[1-²H₁]-2-Amino-1-hydroxyethylphosphonic Acid {(R)-[1-²H₁]-1}

Azide (R)-[1-²H₁]-6 (0.361 g, 1.43 mmol) was dissolved in dry ethanol (20 mL), and HCl (12 M, 0.25 mL) and Pd/C (10% Pd, 0.103 g) were added. The reaction mixture was hydrogenated (3.5 bar, Parr apparatus) at room temperature for 2 h. The catalyst was filtered (Celite, moistened with ethanol), and the filtrate was concentrated in vacuo. The residue was dissolved in HCl (6 M, 20 mL) and refluxed (130 °C oil bath) for 7 h. The resulting solution was concentrated and applied to an ion-exchange resin (Dowex 50W H⁺ form, 50 mL, water as the eluent). Ninhydrin positive fractions were pooled and concentrated to give a colorless solid (0.168 g, 1.2 mmol, 83%); TLC 6:3:2 iPrOH/H₂O/NH₃ [25% (w/w)], R_f = 0.15. The degree of deuteration was determined from the ¹H NMR spectrum to be ≥99%. The product was crystallized from water/ethanol (from 60 to 4 °C) to give colorless needles (0.153 g, 1.1 mmol, 75%): mp 262–267 °C dec; [α]_D²⁰ = −33.5 [c 0.77, H₂O; lit. value¹⁵ of the nondeuterated compound, [α]_D²⁰ = −36.8 (c 0.987, H₂O)]; ¹H NMR (D₂O, 400.27 MHz) δ 3.29 (ABP system; A part, dd, J = 13.3, 6.4 Hz, 1H; B part, dd, J = 13.3, 6.3 Hz, 1H; CH₂); ³¹P NMR (D₂O, 162.03 MHz) δ 15.07 (s); ¹³C NMR (D₂O, 100.65 MHz) δ 64.87 (dt, J = 155.2, 21.3 Hz, CD-P), 41.36 (d, J = 8.9 Hz, CH₂).

Enzyme Assays

The rate of reaction of (R)-1 with PhnZ or its active site variants was measured by monitoring the production of P_i using the malachite green assay. A concentrated solution of PhnZ (5–20 μM) was preincubated with 1 mM of Fe(SO₄)₂(NH₄)₂, 5 molar equivalents of ascorbate, and 1 mM TCEP in reaction buffer [50 mM MOPS-HCl (pH 7.00), 150 mM NaCl, and 10% glycerol] for 20 min on ice. Reaction buffer containing 1 mg/mL bovine serum albumin (BSA) and (R)-1 in a total volume of 300 μL was preincubated at 25 °C for 2 min. The reaction was initiated by the addition of a freshly prepared Fe(II) solution [3 μL of 10 mM Fe(SO₄)₂(NH₄)₂ in 1 mM HCl] followed by the preincubated PhnZ solution (2–5 μL). The final reaction mixture consisted of 43 nM PhnZ, 100 μM Fe(SO₄)₂(NH₄)₂, 200 μM TCEP, 100 μM ascorbate, 1 mg/mL bovine serum albumin (BSA), and (R)-1 in a total volume of 300 μL. When the PhnZ variants were being assayed, the following final concentrations were used: PhnZ Y24F, 60 nM; PhnZ E27A, 600 nM; PhnZ H62A, 500 nM. Aliquots of 50 μL were removed from the reaction every 30 s and mixed with 50 μL of the malachite green reagent, and the absorbance of the resulting mixture was measured at 635 nm using a plate reader (SpectraMax M2, Molecular Devices Corp.). The concentration of (R)-1 was varied, at a minimum, from 1/5 to 5 times the final K_M value determined for PhnZ or its variants. The P_i concentration produced per unit time was determined from a P_i standard curve. The initial rate (V₀) at each concentration of (R)-1 was obtained by a linear regression of the measured [P_i] as a function of time. A plot of V₀/[E]₀ versus the concentration of (R)-1 was fit to the standard Michaelis–Menten equation to calculate kinetic parameters.

pH–Rate Studies

The pH dependence of k_cat/K_M for PhnZ and active site variants (Y24F, E27A, and H62A) was determined in the following buffers (50 mM) and pH ranges: sodium citrate at pH 4.75, MES at pH 5.5–6.5, MOPS at pH 7.0, BTP at pH 7.5–9.5, and CHES at pH 9.5–10. The activity assay was carried out as described above. A plot of V₀/[E]₀ versus the concentration of (R)-1 was fit to the standard Michaelis–Menten equation to calculate kinetic parameters at each pH value. k_cat/K_M values for (R)-1 were plotted against pH and fit to eq 1²⁷ to obtain maximal values of k_cat/K_M and the ionization constants pK_a1 and pK_a2.

$$Y=\frac{C({10}^{\text{pH}-\text{p}{K}_{\text{a}1}})({10}^{\text{p}{K}_{\text{a}2}-\text{pH}})}{({10}^{\text{pH}-\text{p}{K}_{\text{a}1}}+1)({10}^{\text{p}{K}_{\text{a}2}-\text{pH}}+1)}$$ (1)

where Y represents k_cat/K_M and C is the maximum value of Y. pK_a1 and pK_a2 are the acidic and basic limb ionization constants, respectively.

Deuterium Kinetic Isotope Effects

Kinetic parameters were determined for wild type PhnZ and active site variants H62A, Y24F, and E27A using (R)-1 and (R)-[1-²H₁]-1 as per the enzyme assay method described above. The concentration of (R)-1 and (R)-[1-²H₁]-1 was varied, at a minimum, from 1/5 to 5 times the final K_M value determined for PhnZ or its variants. The initial rates with (R)-1 and (R)-[1-²H₁]-1 were measured on the same day using the same PhnZ preparation to minimize differences arising from changes in the specific activity over time. A plot of V₀/[E]₀ versus the concentration of (R)-1 or (R)-[1-²H₁]-1 was fit to the standard Michaelis–Menten equation to calculate kinetic parameters. The kinetic isotope effect on k_cat [denoted ^D(V)] or k_cat/K_M [denoted ^D(V/K)] was calculated from the ratio of parameters obtained with (R)-1 to those obtained with (R)-[1-²H₁]-1.

Solvent Kinetic Isotope Effect (SKIEs)

Solvent KIEs on k_cat and k_cat/K_M were determined by measuring kinetic parameters of wild type PhnZ and its variants (H62A, Y24F, and E27A) in aqueous or D₂O buffers with the (R)-1 concentration varied typically from 1/5 to 5 times the final K_M value that was determined. The D₂O buffer was made by dissolving MOPS and NaCl in 100% D₂O, and the pD was adjusted by adding NaOD until the pD reached 7.0. The pD value was obtained by adding 0.4 to the pH meter reading.²⁸ All reagents were prepared in the aqueous buffer. The same preparation of PhnZ (or its variants) was used for rate measurements in H₂O and D₂O buffers. Kinetic parameters were determined by fitting the data to the Michaelis–Menten equation.

A proton inventory was performed by measuring the activity of wild type PhnZ in buffer with a variable mole fraction of D₂O [denoted n(D₂O)]. nD₂O is defined as the volume of D₂O-based buffer per volume of H₂O-based buffer. The relative rate v_n/v₀, where v_n = V₀/[E]₀ at n mole fraction of D₂O and v₀ = V₀/[E]₀ in the 100% H₂O-based buffer, was plotted against n(D₂O). The data were fit to eq 2²⁹

$$\frac{v_n}{v_0}={\left(1-n+\frac{n}{\text{SKIE}}\right)}^s$$ (2)

where n = n(D₂O) and s is the number of exchangeable protons responsible for an observed solvent KIE.

Viscosity Effects

The effect of diffusion-controlled substrate binding and/or product release on the rate of catalysis was examined by measuring the rate as a function of solvent viscosity at 1 mM (R)-1. Glycerol was used as the viscosigen to yield a final composition of 0%, 1%, 5%, 10%, 15%, or 20% (v/v) in reaction buffer. The relative rate was calculated as v₀/v_n, where v₀ = V₀/[E]₀ measured in buffer with no glycerol and v_n = V₀/[E]₀ measured at a given concentration of glycerol in reaction buffer. v₀/v_n was plotted against the percentage of glycerol.

Results

Kinetic Analysis

The rate of reaction of PhnZ with (R)-1 was measured in a stopped assay format using malachite green to quantify P_i release. The reaction of PhnZ with (R)-1 was also optimized, affording greater activity than what was reported in our earlier work.¹⁵ Key changes to the assay include preincubation of PhnZ on ice with freshly prepared iron(II) ammonium sulfate and the addition of ascorbate to the reaction mixture (Figure S2). Under these optimized assay conditions, the reaction of PhnZ with (R)-1 afforded typical Michaelis–Menten kinetics (Figure 1A, black curve). New kinetic parameters for wild type PhnZ and the active site variants Y24F, E27A, and H62A are listed in Table 1. The new PhnZ k_cat value is 36-fold greater than that reported previously.¹⁵ Substitution of the proposed general acid/base residue H62 with alanine has a major impact on PhnZ activity, resulting in a 20-fold decrease in k_cat, a 4.9-fold increase in K_M, and a 100-fold decrease in k_cat/K_M. An even greater impact is observed upon substituting the substrate recognition residue E27 with alanine, affording a 27-fold decrease in k_cat, a 37-fold increase in K_M, and a 1037-fold decrease in k_cat/K_M. However, as observed in our earlier studies,¹⁵ substitution of the Fe1 binding residue Y24 with phenylalanine has a modest impact on activity, affording a 1.5-fold decrease in k_cat, an 8.9-fold increase in K_M, and a 13.9-fold decrease in k_cat/K_M. Overall, these kinetic parameters are consistent with our previous proposals,¹⁵ where E27 and H62 provide substantial contributions to transition-state stabilization through direct contact with (R)-1, while Y24 is less important for catalysis due to its proposed role in controlling access of O₂ to Fe1.

Figure 1.

Primary deuterium and solvent kinetic isotope effect measurements with PhnZ. (A) Plot of V₀/[E]₀ vs the concentration of (R)-1 (black) or (R)-[1-²H₁]-1 (red). (B) Plot of V₀/[E]₀ vs the concentration of (R)-1 in H₂O buffer (blue) and 97% D₂O buffer (maroon).

Table 1. Kinetic Parameters and D-KIE Values for the Reaction of PhnZ and Active Site Variants with (R)-1.

	wild type	H62A	Y24F	E27A
Hkcat (s−1)a	6.8 ± 0.3	0.34 ± 0.01	4.5 ± 0.1	0.25 ± 0.02
HKM (μM)	63 ± 11	310 ± 37	566 ± 44	2354 ± 417
H[kcat/KM] (M−1 s−1)	(1.1 ± 0.2) × 105	(1.1 ± 0.1) × 103	(7.9 ± 0.7) × 103	106 ± 21
Dkcat (s−1)b	1.23 ± 0.03	0.37 ± 0.03	2.0 ± 0.1	0.28 ± 0.03
DKM (μM)	25 ± 2	3506 ± 602	327 ± 34	4196 ± 697
D[kcat/KM] (M−1 s−1)	(4.9 ± 0.5) × 104	104 ± 20	(6.1 ± 0.7) × 103	66 ± 13
D(V)	5.5 ± 0.3	0.9 ± 0.1	2.3 ± 0.1	0.9 ± 0.1
D(V/K)	2.2 ± 0.4	11 ± 2	1.3 ± 0.2	1.6 ± 0.4
D2O(V)c	2.2 ± 0.3	0.81 ± 0.05	1.9 ± 0.1	5.6 ± 1.7
D2O(V/K)	1.3 ± 0.3	0.98 ± 0.22	2.3 ± 0.5	1.5 ± 1.1

^aKinetic parameters for reaction with (R)-1.

^bKinetic parameters for reaction with (R)-[1-²H₁]-1.

^cSKIE values derived from kinetic parameters for (R)-1 in aqueous and 97% D₂O buffers (see Table S1). Error values for kinetic parameters are derived from a fit of the Michaelis–Menten equation to the corresponding V₀/[E]₀ vs [S] data. KIE error values are derived from propagation of the errors of each compared kinetic parameter.

Deuterium Kinetic Isotope Effects

To probe whether C–H bond cleavage is rate-limiting in the PhnZ reaction, a noncompetitive deuterium kinetic isotope effect (D-KIE) was measured by comparing the kinetic parameters for (R)-1 and (R)-[1-²H₁]-1 (Figure 1A). The kinetic parameters of PhnZ using (R)-[1-²H₁]-1 as a substrate are listed in Table 1. Large D-KIE values of 5.5 ± 0.3 for ^D(V) and 2.2 ± 0.4 for ^D(V/K) are observed with wild type PhnZ, indicating that C–H bond cleavage is rate-limiting. Kinetic parameters for (R)-[1-²H₁]-1 were also determined for the H62A, Y24F, and E27A PhnZ variants (Table 1 and Figure S1). Interestingly, a negligible ^D(V) value is observed for PhnZ H62A, but a large ^D(V/K) is still observed [^D(V) = 0.9 ± 0.1; ^D(V/K) = 11 ± 2]. In the case of PhnZ E27A, negligible values are observed for both ^D(V) and ^D(V/K). Finally, PhnZ Y24F exhibits a reduced but still significant ^D(V) value relative to that of the wild type enzyme [^D(V) = 2.3 ± 0.1] but a negligible ^D(V/K) value.

Solvent Kinetic Isotopic Effects and Proton Inventory

A solvent KIE was measured for wild type PhnZ by comparing the kinetic parameters for (R)-1 in water and 97% D₂O (pH 7.0) (Figure 1B), affording a ^D₂O(V) of 2.2 ± 0.3 and a ^D₂O(V/K) of 1.3 ± 0.3 (Table 1). The kinetic parameters for (R)-1 in water and 97% D₂O were also determined for the active site variants H62A, Y24F, and E27A (Figure S1). This afforded a ^D₂O(V) of 0.81 ± 0.05 and a ^D₂O(V/K) of 0.98 ± 0.22 for H62A, a ^D₂O(V) of 1.9 ± 0.1 and a ^D₂O(V/K) of 2.3 ± 0.05 for Y24F, and a ^D₂O(V) of 5.6 ± 1.7 and a ^D₂O(V/K) of 1.5 ± 1.1 for E27A (Table 1). The significant ^D₂O(V) values observed for wild type PhnZ, Y24F, and E27A reflect transfer of one or more protons in the rate-determining step, while the absence of a significant ^D₂O(V) value with PhnZ H62A suggests that H62 mediates transfer of one or more protons.

A proton inventory was performed with wild type PhnZ to provide greater insight into the observed solvent KIE. The proton inventory experiment was conducted on k_cat by measuring initial rates at one saturating concentration of (R)-1 (1 mM) in various mole fractions (n) of D₂O, which was derived by combining appropriate volumes of D₂O- and H₂O-based buffers (Figure 2A). The plot of v_n/v₀ versus the mole fraction of D₂O was fitted to eq 2,²⁹ affording a calculated solvent KIE value of 2.9 ± 1.4 and a number of protons in flight (s) of 1.5 ± 0.6. This calculated SKIE value is in good agreement with the ^D₂O(V) value measured for PhnZ from the Michaelis–Menten plots in H₂O and 97% D₂O. The calculated s of 1.5 ± 0.6 implies that one to two exchangeable protons are involved in the observed solvent KIE.

Figure 2.

Proton inventory and viscosity measurements with wild type PhnZ [1 mM (R)-1, 25 °C, pL = 7.00]. (A) Proton inventory measurements. A plot of v_n/v₀ vs the mole fraction of D₂O, where v_n and v₀ represent V₀/[E]₀ in a buffer with n mole fraction of D₂O and in a buffer with no D₂O, respectively. The data were fitted to eq 2 (—), affording a calculated solvent KIE of 2.9 ± 1.4 and a number of protons in flight (s) of 1.5 ± 0.6. (B) Solvent viscosity measurements. A plot of v₀/v_n vs glycerol concentration (%, v/v), where v₀ and v_n are V₀/[E]₀ values in a buffer with no glycerol and in a buffer with n% glycerol, respectively. The dashed lines are for reference, where a slope of zero represents a reaction that is not diffusion-limited and a slope of 1 represents a reaction that is diffusion-limited. A linear fit (—) produced a slope of 0.027 ± 0.007.

Dependence of the PhnZ Reaction Rate on Viscosity

The dependence of the PhnZ reaction rate on solvent viscosity was analyzed to determine whether diffusion-controlled substrate binding and/or product release limits the rate of catalysis. The effect of solvent viscosity on k_cat was tested by measuring initial rates at saturating concentrations of (R)-1 (1 mM) at different concentrations of glycerol. The ratio v₀/v_n was plotted against the glycerol concentration, where v₀ and v_n are initial velocities at 0% and n% glycerol, respectively (Figure 2B). A slope of zero is expected if catalysis is not limited by diffusion, whereas a slope of unity represents a reaction that is fully diffusion-limited by nonchemical steps.^30,31 For PhnZ, an essentially negligible dependence of k_cat on viscosity (m = 0.027 ± 0.007) is observed. This indicates that the PhnZ reaction is not limited by diffusion of the substrate or products from the active site. Additionally, the viscosity of 20% glycerol (1.76 cP at 20 °C),³² the maximal value tested with PhnZ, exceeds that of 100% D₂O (1.25 cP at 20 °C),³³ indicating that the observed ^D₂O(V) value observed with PhnZ is not an artifact of the greater viscosity of D₂O relative to that of water.

pH Rate Dependence of PhnZ and Active Site Variants

The pH dependence of k_cat, K_M, and k_cat/K_M for PhnZ and the active site variants Y24F and H62A was determined by Michaelis–Menten analyses at various pH values. The pH–rate dependence of PhnZ E27A could not be measured with confidence due to the low activity and high K_M value of this variant (Table 1). The pH dependence of k_cat/K_M for wild type PhnZ and PhnZ Y24F exhibited bell-shaped profiles with optima spanning pH 7 that could be fit to an equation describing two ionizations (Figure 3A,C). Essentially identical apparent pK_a values for wild type PhnZ (pK_a1 = 5.77 ± 0.05, and pK_a2 = 8.85 ± 0.04) and PhnZ Y24F (pK_a1 = 5.5 ± 0.2, and pK_a2 = 8.7 ± 0.3) are observed (Table 2). In the case of PhnZ H62A, the pH dependence of the kinetic parameters could be fit with confidence to only a single ionization [pK_a = 8.8 ± 0.2 (Figure 3B)] due to the low activity and instability of this variant at low pH values.

Figure 3.

pH–rate dependence of PhnZ and its variants with (R)-1. Plots of k_cat/K_M vs pH for (A) wild type PhnZ, (B) the H62A variant, and (C) the Y24F variant. The kinetic parameters for these experiments are summarized in Table 1.

Table 2. Parameters for the pH–Rate Dependence of PhnZ and Its Variants.

	wild type	H62A	Y24F
maximum kcat/KM (M−1 s−1)	(3.5 ± 0.1) × 104	(3.6 ± 0.4) × 102	(1.08 ± 0.14) × 104
pKa1	5.77 ± 0.05	not available	5.5 ± 0.2
pKa2	8.85 ± 0.04	8.8 ± 0.2	8.7 ± 0.3

Discussion

In this study, we show that the oxidative C–P bond cleavage reaction catalyzed by PhnZ is rate-limited by C–H bond cleavage. The large D-KIE values observed for PhnZ [^D(V) = 5.5 ± 0.3] using (R)-[1-²H₁]-1 as substrate indicate that cleavage of the C1−H bond of (R)-1 is the rate-limiting step. Additionally, the significant solvent KIE observed for PhnZ [^D₂O(V) = 2.2 ± 0.3] and a proton inventory indicates that one or two protons are in flight during C–H bond cleavage. One of the proton transfer events was assigned to H62 on the basis of the fact that PhnZ H62A does not show a significant solvent KIE. Interestingly, a negligible ^D(V) value accompanied by a large ^D(V/K) value is observed with the H62A variant. This implies that the rate-limiting step for PhnZ H62A is no longer C1−H bond cleavage but instead represents the first irreversible chemical step.³⁴

H62 is notable as a conserved fifth histidine residue that is unique to the PhnZ subclade of the HD superfamily of phosphohydrolases (the other four conserved histidine residues are dedicated to Fe ion binding). H62 is observed by X-ray crystallography to hydrogen bond to the C1-OH of (R)-1,^13,15 and deletion of this side chain has a large effect on k_cat and k_cat/K_M (Table 1), indicating an important role in the rate-determining transition state. Overall, these observations suggest H62 is acting as a general acid or base during C–H bond cleavage of (R)-1, possibly by mediating proton transfer to or from the C1-OH. Our results are consistent with the DFT/QM calculations on PhnZ by Zhao and co-workers,²² which predict that abstraction of the C1 hydrogen by a Fe1-bound superoxo species is rate-limiting and that H62 acts as a general base in this step by deprotonating the C1-OH (Scheme 1, III). Bollinger and co-workers have proposed that one advantage conferred by the mixed valent diiron(II/III) cofactor over mononuclear iron(II) or diiron(II/II) cofactors is that the ferric ion can be used to promote ionization of the substrate hydroxyl while the ferrous ion remains available to activate O₂.³⁵ The developing oxyanion character in the transition state can be expected to stabilize development of a radical at C1 analogous to resonance of a ketyl radical. Steigerwald and co-workers have estimated that such oxyanion character can reduce the neighboring C–H bond dissociation energy by 10–16.5 kcal mol⁻¹.³⁶ Similarly, by deprotonating the substrate C1-OH, H62 can assist in weakening the substrate C–H bond in the transition state.

As founding members of MVDO family, PhnZ and MIOX form an interesting comparison due to their application of a conserved active site architecture to cleavage of C–P and C–C bonds. There are notable differences in substrate interactions between the two enzymes. Although the substrates for PhnZ and MIOX both adopt a bidentate mode of binding to Fe2 in their active sites (Scheme 3), the ionization states of the C1-hydroxyls appear to be different at the onset of catalysis. Instead of histidine, MIOX uses a lysine residue (K127) to interact with C1-OH of myo-inositol.³⁷ Additionally, the C1-OH of myo-inositol is observed to bind to the diiron cofactor of MIOX as an alkoxide,^38,39 which is proposed to be stabilized by the Lewis acidity of Fe2 and the positive charge of K127.^22,37 Therefore, while PhnZ is proposed to perform C–H bond cleavage concurrently with ionization of the C1-hydroxyl in the transition state (Scheme 3A, I), MIOX appears to initiate C–H bond cleavage directly from a C1-alkoxide (Scheme 3B, I).^20,22

Scheme 3.

The rate-limiting step for MIOX also differs from that of PhnZ. Under steady-state conditions, the reaction of mouse MIOX with [²H₆]-myo-inositol does not show a significant D-KIE.²⁰ Instead, stopped-flow absorbance and freeze-quench EPR studies with MIOX have shown that turnover of an unknown intermediate “H” is rate-limiting.^20,35 Calculations by Hirao and co-workers⁴⁰ predict that O–O bond cleavage of a Criegee-like intermediate (Scheme 3B, III), possibly intermediate “H” mentioned above, is rate-limiting for MIOX, with an activation barrier 10 kcal mol⁻¹ greater than the C–H bond cleavage step. Using [²H₆]-myo-inositol as a substrate, freeze-quench EPR studies were successful in following the pre-steady-state formation and decay of the diiron Fe(III)/Fe(III)–superoxo complex that performs C1−H abstraction (Scheme 3B, I to II).²⁰ From this experiment, a D-KIE value of ≥5 was estimated for the C–H cleavage step of MIOX, which is comparable to the value measured for PhnZ. Interestingly, pig MIOX shows deuterium (k_H/k_D = 2.1) and tritium (k_H/k_T = 7.5) KIEs under steady-state conditions when assayed with 1-²H₁- and 1-³H₁-labeled myo-inositol, respectively,⁴¹ suggesting that under some reaction conditions the C–H bond cleavage step can be partly rate-limiting for MIOX.

The differing roles of H62 in PhnZ and K127 in MIOX are also predicted to impact the fragmentation mechanisms of the Criegee intermediates formed by each enzyme.²² Calculations predict that K127 maintains a positive charge throughout the reaction cycle of MIOX, whereas H62 of PhnZ alternately functions as a general base and acid.^22,40 In the case of MIOX, the inability of K127 to transfer a proton to the Criegee alkoxide promotes homolytic O–O bond cleavage, leading to a ferryl species on Fe1 and a gem-diolate alkoxyl radical (Scheme 3B, III to IV). With PhnZ, protonation of the Criegee alkoxide by H62 is predicted to promote an “inverse heterolytic” mode of O–O cleavage, leading to a diiron-bound orthoester intermediate (Scheme 3A, III to IV). The comparable O–O homolytic pathway in PhnZ is calculated to be significantly higher in energy (8.8 kcal/mol), likely due to the formation of a high-energy ferryl species. In this context, it is notable that C–H bond cleavage is not rate-limiting for the PhnZ H62A and E27A variants, as shown by the negligible ^D(V) values, but remains at least partly rate-limiting for the Y24F variant [^D(V) = 2.3 ± 0.1] (Table 1). This suggests that H62 and E27 contribute substantially to another step along the reaction coordinate, which becomes rate-limiting when these residues are absent. Intriguingly, the E27A variant still shows a substantial solvent KIE, while the H62A variant does not (Table 1). This indicates that the rate-limiting step for E27A also involves a proton transfer event. If the same step is rate-limiting for E27A and H62A, this suggests that H62 is also mediating proton transfer in this step.

Considering their respective roles in Fe1 ion binding, substrate binding, and general acid–base catalysis, the ionization states of Y24, E27, and H62 can be expected to have an impact on catalysis. PhnZ exhibits a bell-shaped profile for the dependence of k_cat/K_M with apparent pK_a values of 5.5 and 8.8 (Table 2). Y24F and H62A also have bell-shaped pH profiles that are defined by essentially identical apparent pK_a values. Unfortunately, an acidic pK_a value for H62 could not be unambiguously assigned. The H62A variant could be assigned only a single ionization with a pK_a of 8.8. However, due to the low activity of H62A and the instability of this variant at low pH, the absence of an ionization at low pH cannot be excluded. Overall, the pH dependence of the reaction of PhnZ with (R)-1 appears to be a composite of multiple ionization events. This would include the ionization state of the 2-amino group and phosphoryl oxygens of (R)-1, which would affect interactions with E27 and Fe2, as well as the residues that form a polar pocket (K108, S126, T129, Q133, and R158) that engage the phosphono group of (R)-1.¹⁵

The D-KIE values suggest that PhnZ has a high forward commitment to catalysis. This is reflected in the smaller ^D(V/K) value [^D(V/K) = 2.2 ± 0.4] relative to the ^D(V) value [^D(V) = 5.5 ± 0.3].⁴² The forward commitment of an enzyme to catalysis is reflected by the relative magnitudes of the isotopically sensitive rate constant (C–H bond cleavage in the case of PhnZ) relative to the net rate constant for dissociation of the reactive enzyme substrate (ES) complex into free enzyme and substrates.^42,43 Because V/K reflects all ES binding interactions up to the first irreversible chemical step³⁴ while V reflects the rate-limiting step, suppression of the ^D(V/K) value from the intrinsic value reflects a slow release of the substrate(s) from the active site.⁴² As illustrated in Scheme 1, V/K for PhnZ would include contributions from the following sequence of steps: (1) binding of (R)-1, (2) flipping of the Y24/E27 loop, (3) binding and reduction of O₂ by Fe1, and (4) C–H bond abstraction by the Fe1 superoxo species. With PhnZ, the steps involving movement of the Y24/E27 loop and subsequent reduction of O₂ to superoxide could contribute to a high commitment of catalysis by hindering dissociation of the reactive PhnZ substrate complex (Scheme 1, III) to the free enzyme, O₂, and (R)-1. It should be noted that binding of O₂ to MIOX is observed to occur reversibly,²⁰ although this enzyme does not use a direct induced-fit mechanism for substrate binding like PhnZ.

Beyond PhnZ and MIOX, there are relatively few characterized examples of C–H bond cleavage by ferric superoxo species. The mononuclear non-heme-dependent enzyme isopenicillin-N synthase has been shown to use a ferric superoxo species for abstraction of the C3 hydrogen of the cysteine moiety of γ-(α-aminoadipoyl)cysteinylvaline,⁴⁴ a step that is associated with a large D-KIE value ranging from 6⁴⁵ to 33.⁴⁴ The related mononuclear non-heme iron-dependent dioxygenases HEPD and MpnS perform two hydrogen abstractions during oxidative C–C bond cleavage of 2-hydroxyethylphosphonate, the first cleavage of a C–H bond by a ferric superoxo species, the second cleavage of an O–H bond by a ferryl species.⁴⁶ The C–H bond cleavage step is not rate-limiting for HEPD or MPnS as a negligible D-KIE is observed on V/K for 2-[²H₂]-2-hydroxyethylphosphonate; however, significant D-KIE and ¹⁸O KIE values were observed on V/K for O₂, indicating reversible formation of the ferric superoxo species and possibly hydrogen tunnelling during transfer.^46,47 The copper-dependent enzymes dopamine β-monooxygenase, tyramine β-monooxygenase, and peptidyglycine α-amidating enzyme are also proposed to reversibly form Cu(II)–superoxo species that perform rate-limiting C–H bond cleavage.⁴⁸ In this context, the large ^D(V) value observed with PhnZ presents an excellent opportunity to accumulate high levels of the diiron(III/III)–superoxo species for spectroscopic characterization.⁴⁹

Supplementary Material

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.biochem.9b00145.

Deuterium and solvent kinetic isotope analysis of PhnZ H62A, Y24F, and E27A (Figure S1 and Table S1), PhnZ rate dependence on iron and ascorbic acid concentrations (Figure S2), and ¹H, ¹³C, and ³¹P NMR spectra of all synthesized compounds (PDF)