A Long-Lived, Substrate-Hydroxylating Peroxodiiron(III/III) Intermediate in the Amine Oxygenase, AurF, from Streptomyces thioluteus

Abstract

The amine oxygenase AurF from Streptomyces thioluteus catalyzes the six-electron oxidation of p-aminobenzoate (pABA) to p-nitrobenzoate (pNBA). In this work, we have studied the reaction of its reduced Fe₂(II/II) cofactor with O₂, which results in generation of a peroxo-Fe₂(III/III) intermediate. In the absence of substrate, this intermediate is unusually stable (t_1/2 = 7 min at 20°C), allowing for its accumulation to almost stoichiometric amounts. Its decay is accelerated ~10⁵-fold by the substrate, pABA, implying that it is the complex that effects the two-electron oxidation of the amine to the hydroxylamine. The nearly quantitative conversion of pABA to pNBA by solutions containing an excess of the intermediate suggests that it may also be competent for the two subsequent two-electron oxidations leading to the product.

AurF from Streptomyces thioluteus utilizes O₂ as oxidant to effect the six-electron oxidation of p-aminobenzoate (pABA) to p-nitrobenzoate (pNBA), a building block of the antibiotic aureothin.¹ This conversion entails three successive two-electron oxidations via p-hydroxylaminobenzoate (pHABA) and p-nitrosobenzoate intermediates.^2,3 Sequence analysis and the crystal structure of AurF showed that it has a carboxylate-bridged dinuclear metal center, which was presumed to activate O₂ for the oxidation reactions. Following initial controversy as to the identities of the metal ions in the functional form(s),^2–7 Zhao and co-workers recently established that the diiron form is competent to convert pABA to pNBA in the presence of O₂ and a reducing system.⁸ Here we demonstrate that the Fe₂^II/II cluster in AurF reacts with O₂ in the absence of substrate to form a remarkably stable (t_1/2 ~ 7 min at 20 °C) adduct with spectroscopic properties characteristic of a peroxo-Fe₂^III/III complex. The intermediate complex decays rapidly (t_1/2 ~ 0.005 s at 20 °C) when mixed with stoichiometric pABA, implying that it is competent for at least the first oxidation in the three-step sequence. The nearly quantitative (> 80%) conversion of pABA to pNBA upon addition of < 0.3 equiv of the amine substrate to a solution of the intermediate suggests that the complex might effect all three steps in the sequence. The results establish a new reactivity of non-heme diiron-peroxide intermediates: arylamine oxygenation.

Diiron AurF was isolated from over-producing Escherichia coli cells as described in the Supporting Information. The UV/visible absorption spectrum of the protein (Fig. 1, blue trace) has a band at ~360 nm similar to those exhibited by other proteins with µ-oxo-Fe₂^III/III clusters, wherein it has been assigned to an oxo-to-iron charge transfer transition.⁹ The 4.2-K/53-mT Mössbauer spectrum of AurF isolated from cultures supplemented with ⁵⁷FeSO₄ (> 95% ⁵⁷Fe) can be simulated as two quadrupole doublets with parameters typical of high-spin Fe^III [δ₁ = 0.54 mm/s, ΔE_Q,1 = −1.86 mm/s; δ₂ = 0.48 mm/s ΔE_Q,2 = 0.80 mm/s] and relative contributions to the experimental spectrum of 78% (orange line) and 22% (green line), respectively (Fig. 2A). The large |ΔE_Q,1| of 1.86 mm/s is typical of µ-oxo-Fe₂^III/III clusters, consistent with the absorption spectrum and the crystal structure.⁸ The complex with smaller ΔE_Q could have one or more µ-hydroxo bridge(s). Mössbauer spectra collected in externally applied fields of 5T and 8T (Fig. 2E and 2F) can be simulated well with the same parameters and the assumption of a diamagnetic ground state. The S = 0 ground state arises from antiferromagnetic coupling between the two high-spin Fe^III sites.

Figure 1.

Absorption spectra of a sample of AurF subjected to DT reduction and O₂ re-oxidation: as-isolated protein (0.40 mM in 50 mM HEPES pH 7.5 and 10% glycerol buffer; blue line); after treatment with 1 equiv DT for 10 min at 20 °C (red line); after addition of 0.2 volumes of cold oxygenated buffer ([O₂]_final = 0.38 mM, green line), and after addition of pABA (0.92 mM final; black line). Inset: Kinetic traces after mixing DT-reduced AurF (0.64 mM) at 21 °C with an equal volume of O₂-saturated buffer, allowing the reaction to proceed for 0.1 s, and then mixing the resultant solution with an equal volume of anaerobic buffer either lacking pABA (squares) or containing 1 equiv pABA (circles). The solid lines are simulations with parameters described in the text.

Figure 2.

4.2-K Mössbauer spectra (hashed marks) of as-isolated AurF (A, E, F); DT-reduced AurF (B); DT-reduced AurF exposed to 2 atm O₂ for 2 min, and then either directly frozen for Mössbauer analysis (C) or further treated with 0.3 equiv pABA prior to being frozen for analysis (D), and DT-reduced AurF treated with a slight excess of O₂ from an O₂-saturated buffer solution (G, H). Spectra were collected in external fields of 53 mT (A–D), 5T (E, G), or 8T (F, H). The solid lines in A – C are simulations with parameters quoted in the text. The solid line in D is the experimental spectrum from C re-plotted to illustrate the spectral changes. The solid lines in E – H are spin Hamiltonian simulations with parameters in the text, asymmetry parameters η = -0.4 (orange), η = 1 (green), η = 1.4 (red), η = 0.3 (blue), and the assumption of S = 0 ground states for all species.

Treatment of the protein with 1 equiv sodium dithionite (DT) bleaches the band at 360 nm in the absorption spectrum (Fig. 1, red trace) and produces a new quadrupole doublet with parameters (δ = 1.24 mm/s, ΔE_Q = 3.06 mm/s) typical of high-spin Fe^II (Fig. 2B). These observations indicate that the Fe₂^III/III protein is reduced to the Fe₂^II/II form by DT treatment.

Reaction of Fe₂^II/II AurF with O₂ produces a new broad absorption band centered at ~500 nm (ε₅₀₀ ~ 0.5 mM⁻¹cm⁻¹; Fig. 1, green trace). Formation of the 500-nm-absorbing species is complete within 0.01 s at ~ 0.6 mM O₂ and 20 °C (Fig. S1). The species does not decay noticeably on the 100-s time scale of the SF absorption experiment. The 4.2-K/53-mT Mössbauer spectrum of the protein following vigorous stirring of the Fe₂^II/II complex for ~ 2 min under ~ 2 atm O₂ (Fig. 2C) shows that the diiron cluster is almost quantitatively oxidized. In addition to small amounts of the sub-spectra of Fe₂^II/II (5%, purple line) and µ-oxo-Fe₂^III/III (18%, orange line) complexes, the spectrum reveals several new sharp peaks, which can be simulated as two quadrupole doublets with similar parameters. The parameters [δ₁ = 0.54 mm/s, ΔE_Q,1 = −0.66 mm/s (49 %, blue line) and δ2 = 0.61 mm/s, ΔE_Q,2 = 0.35 mm/s (33 %, red line)] are typical of high-spin Fe(III).¹⁰ Spectra collected in external fields of 5T and 8T (Fig. 2G and 2H)¹¹ confirm that the new state has S = 0 ground state(s), due to antiferromagnetic coupling of two high-spin Fe^III sites. Because it is an Fe₂^III/III cluster derived from the Fe₂^II/II enzyme by treatment with O₂, we assign this state as peroxo-Fe₂^III/III. The ~1.5:1 intensity ratio of the two quadrupole doublets suggests the presence of two or more isomeric complexes, perhaps in equilibrium (as has been observed in the reactive intermediate states of other iron enzymes).^12–14 Mössbauer spectra of samples of the peroxide state frozen after subsequent incubation at 20 °C show that it is stable for minutes in the absence of substrate (t_1/2 ~ 7 min) and decays to the µ-oxo-Fe₂^III/III species (Fig. S2).

The reactivity of the peroxo-Fe₂^III/III state was examined by stopped-flow experiments in which the Fe₂^II/II protein was allowed to react with O₂ for 100 ms and the resultant peroxo-Fe₂^III/III state was then mixed with the substrate, pABA, or (as a control) buffer. The 500 nm band of the peroxo-Fe₂^III/III state, stable in the absence of pABA (inset to Figure 1, squares), decays rapidly (k_obs = 150 ± 20 s⁻¹) upon exposure to one equiv of the substrate (inset to Figure 1, circles).

The 4.2-K/53-mT Mössbauer spectra of samples prepared by oxygenation of Fe₂^II/II AurF (Figure 2C, replotted in Figure 2D as solid line for clarity) and subsequent treatment of the oxygenated sample with 0.3 equiv pABA for 1 min (Figure 2D, hashed marks) confirm the reactivity of the peroxo-Fe₂^III/III state toward the substrate. The features of the intermediate complex(es) decay, and those of the µ-oxo-Fe₂^III/III cluster develop. Analysis of the small-molecule components of reactions performed similarly (described in Supporting Information) showed that, at these low pABA/AurF ratios of ≤ 0.3, the substrate is converted to pNBA with a yield of > 80%.¹⁵

The combined absorption and Mössbauer spectroscopic results thus establish that addition of O₂ to the Fe₂^II/II cluster of AurF produces a long-lived state that has properties consistent with a peroxo-Fe₂^III/III formulation. The stopped-flow data show that the intermediate state reacts rapidly with pABA, implying that it is competent for at least the first oxidation in the three step AurF sequence. The possibility that it is competent for all three oxidations seems likely (on the basis of the efficient conversion of limiting pABA to pNBA) but requires more rigorous evaluation.

The spectroscopic properties of the AurF state are different from those of peroxo-Fe₂^III/III complexes detected in related diiron oxidases and oxygenases such as ribonucleotide reductase subunit R2,¹⁶ stearoyl acyl carrier protein Δ⁹ desaturase,¹⁷ and soluble methane monooxygenase hydroxylase (sMMOH).¹⁸ The intermediates in these proteins have µ-1,2-peroxo¹⁹ (or perhaps µ-(η²,η²)-peroxo, for the case of sMMOH20) diiron cores. The properties of the AurF complex are more similar to those of a peroxo-Fe₂^III/III intermediate recently detected in toluene/o-xylene monooxygenase (ToMO).²¹ In this enzyme, the peroxide complex is thought to be attacked as an electrophile by the π-system of the substrate. A similar mechanism for the first step in the AurF reaction, involving nucleophilic attack of the lone pair of the neutral arylamine on the peroxo electrophile (Scheme 1), seems plausible. An attractive possibility is that the peroxo complexes in ToMO and AurF differ structurally from the canonical µ-1,2-peroxides in ways that activate them as electrophiles. Protonation of the peroxide unit, perhaps with rearrangement to a µ-1,1-bridging geometry, seems likely to have this effect.

Scheme 1.

Proposed mechanism for hydroxylation of pABA by a putative µ-1,2 (left) or µ-1,1 (right) hydroperoxo-Fe₂^III/III intermediate.

The remarkable stability of the AurF intermediate, which permits its preparation in concentration, purity, and physical form (e.g., a transmitting glass) appropriate for most spectroscopic methods, should be a considerable asset in ongoing efforts to define the structure of the complex by a combination of experiments and density functional theory calculations.