Structure/function correlations among coupled binuclear copper proteins through spectroscopic and reactivity studies of NspF

Abstract

The terminal step of 4-hydroxy-3-nitrosobenzamide biosynthesis in Streptomyces murayamaensis is performed by NspF, a mono-oxygenase that converts o-aminophenols to the corresponding nitroso product (hydroxyanilinase activity). Previous biochemical characterization of the resting form of NspF suggested that this enzyme belonged to the coupled binuclear copper enzyme (CBC) family. Another member of this enzyme family, tyrosinase, is able to mono-oxygenate monophenols (monophenolase activity) but not o-aminophenols. To gain insight into the unique reactivity of NspF, we have generated and characterized the oxy form of its active site. The observation of spectral features identical to those of oxy-tyrosinase indicates that oxy-NspF contains a Cu₂O₂ core where peroxide is coordinated in a μ-η²∶ η² mode, confirming that NspF is a CBC enzyme. This oxy form is found to react with monophenols, indicating that, like tyrosinase, NspF also possesses monophenolase activity. A comparison of the two electrophilic mechanisms for the monophenolase and hydroxyanilinase activity indicates a large geometric change between their respective transition states. The potential for specific interactions between the protein pocket and the substrate in each transition state is discussed within the context of the differential reactivity of this family of enzymes with equivalent μ-η²∶η² peroxy bridged coupled binuclear copper active sites.

The coupled binuclear copper (CBC) protein family is characterized by a magnetically coupled binuclear copper active site that binds dioxygen in a symmetric, side-on (μ-η²∶η²) fashion (1). Three subclasses of proteins are classically differentiated within this family (Fig. 1): oxygen transport proteins in arthropods and mollusks (hemocyanins), o-diphenol:oxygen oxidoreductases (catechol oxidases), and monophenol oxidases (tyrosinases). The CBC family is unique in that while all members reversibly bind dioxygen (the hemocyanin function), only catechol oxidases and tyrosinases oxidize diphenols (the catechol oxidase function) and only tyrosinases oxygenate monophenols (the tyrosinase function). Thus, each class builds on the function of the previous class with hemocyanin < catecholoxidase < tyrosinase in order of increasing active site chemistry. The molecular basis for this additive function is largely unknown, and is intriguing to consider, given that the respective active sites bind dioxygen in an indistinguishable manner (2). Insight comes from a variety of experimental observations. Crystallography (3–7) and ligand displacement rates (8) indicate that the active sites of catechol oxidase and tyrosinase are more solvent accessible than hemocyanin, supporting substrate access as an important factor. However, structure/function correlations that direct oxidoreductase versus monophenolase reactivity are less well understood. It has been suggested on the basis of mutational studies that the protein pocket allows or precludes access to the copper proximal to the N-terminus, which modulates the observed reactivity (9). Additional insight comes from model chemistry, where dicopper-peroxide complexes have been shown to be competent for the monoxygenation of aryl C-H bonds (10–13). These experimental observations suggest that cores are intrinsically able to perform all of the CBC protein functions. Consistent with this hypothesis was the recent observation that partial denaturation or proteolytic digestion of hemocyanin induces monophenolase activity (14, 15).

Fig. 1.

Functions of the three subclasses of coupled binuclear copper enzymes.

Recently, an open reading frame (nspF) was identified in the nitrosobenzamide biosynthetic pathway of S. murayamaensis that was suggested to encode a tyrosinase-like monoxygenase (16). Recombinant NspF was expressed and purified to homogeneity, verified to contain two moles of copper by ICP-AES and mutant studies, and found to catalyze the monoxygenation of an o-amino phenol into an o-nitroso phenol (hydroxyanilinase activity). In contrast, the reaction of tyrosinase with o-amino phenols leads to o-imino quinones (16, 17) (Fig. 2). Resting NspF was further found to be competent for the oxidation of diphenols to quinones (catechol oxidase activity), but was found to not be competent for the monoxygenation of phenols (monophenolase activity). This interesting reactivity pattern implicates significant structure/function relationships within the active site of NspF, directing the reaction towards its nitroso arene product, and is the impetus for this study. Herein we report the generation and spectroscopic characterization of the oxy- adduct of NspF (oxy-NspF) that definitively identifies NspF as a CBC protein with a active site. Further, we have observed that oxy-NspF (but not resting NspF) is competent for monophenolase reactivity, which indicates that NspF belongs to a fourth subclass of the CBC protein family (Fig. 1). Density functional theory calculations of transition states indicate specific geometric requirements for the hydroxyanilinase vs monophenolase activity of the CBC class of enzymes.

Fig. 2.

The reaction of NspF and tyrosinase with 2-amino-4-carboxamidophenol.

Results

Generation and Spectroscopic Characterization of oxy-NspF.

The addition of twenty-fold excess hydroxylamine [a reductant that is unable to reduce oxy-tyrosinase (18)] to resting NspF under air-saturated conditions results in the formation of a distinct blue intermediate displaying two diagnostic features in its UV–Visible absorption spectrum at 348 nm and 635 nm (Fig. 3). These bands are comparable in energy and relative intensity to the and charge transfer transitions observed in the oxy forms of CBC proteins (dashed absorption spectrum in Fig. 3 is that of oxy-tyrosinase) (8), suggestive that oxy-NspF contains an analogous site. Under an atmosphere of pure dioxygen, the molar extinction coefficient of the 348 nm absorption band increased to 13 mM^-1 cm^-1 (at +4 °C), suggesting that not all of the reduced sites were fully saturated with dioxygen even at high dioxygen concentrations, and thus this value represents a lower limit of the molar absorptivity. Note that the extinction coefficient for oxy-tyrosinase and oxy-hemocyanin is 20 mM^-1 cm^-1 (8). To verify the assignment of the absorption bands as charge transfer (CT) transitions and to probe the geometric and electronic structure of the active site, resonance Raman (rR) studies were performed. Excitation into the dominant UV absorption band (λ_ex = 363 nm) revealed one isotope insensitive band at 271 cm^-1 and two oxygen isotope sensitive features at 749 cm^-1 and 1051 cm^-1, which downshift to 707 cm^-1 and 995 cm^-1 when oxy-NspF is prepared with (Fig. 3, inset). These features are comparable in energy and isotopic shifts to those found previously in the oxy forms of CBC proteins (2, 19) and the crystallographically characterized “side-on” dicopper peroxide model complex [(Tp^iPr)Cu]₂O₂ (20). On this basis, the 271 cm^-1, 749 cm^-1, and 1051 cm^-1 features are assigned as the ν_Cu⋯Cu, ν_O-O, and the second harmonic of the ν_Cu-O of a protein core and verifies the assignment of the UV absorption feature as a peroxo to copper CT transition. Taken together, the UV–Vis absorption and rR spectral features indicate that oxy-NspF possesses a active site analogous to that observed in the oxy forms of CBC proteins.

Fig. 3.

UV–Visible absorption spectra of oxy-NspF (43 μM, 10 mm cuvette) (—) and oxy-tyrosinase (- -) under ambient conditions. Inset: rR spectra (λ_ex = 363.8 nm, 233 K) of oxy-NspF prepared with natural abundance O₂ (—) and (- -).

Reactivity of oxy-NspF with Monophenol Substrates.

Resting NspF was previously demonstrated to be unreactive towards monophenolic substrates in the presence of dioxygen (16). However, the addition of 4-carboxamidophenol (Fig. 4) to a solution of oxy-NspF resulted in the rapid decay of the absorption features of oxy-NspF accompanied by the appearance of a band at approximately 430 nm. The later feature, which reflects the quinone product, indicates that in contrast to the resting enzyme, oxy-NspF reacts with monophenolic substrates to produce quinones (i.e., the tyrosinase function). The lack of an isosbestic point during this process precluded kinetic analysis under these conditions by optical spectroscopy. Therefore, dioxygen consumption was measured in a sealed chamber after the addition of catalytic amounts of NspF. The addition of resting NspF to monophenols such as 4-carboxamidophenol and 4-methylphenol (Fig. 4) does not result in the uptake of dioxygen, consistent with observations in (16) (SI Appendix, Fig. S1). Alternatively, dioxygen consumption is observed when hydroxylamine is added to reduce resting NspF to the deoxy form *. In the absence of substrate, no oxygen consumption is observed, indicating that hydroxylamine is unable to reduce oxy-NspF. Product analysis by HPLC (SI Appendix, Fig. S2) confirms that the product produced by NspF with 4-methylphenol was identical to the quinone that is produced by recombinant S. glaucescens tyrosinase with the same substrate. Kinetic parameters were measured under steady state conditions using the borate buffer system in (18) in the presence of excess hydroxylamine to overcome the reactivity of the enzymatically produced quinones and an initial lag phase. Under these conditions, Michaelis-Menten behavior is observed for the reaction of 4-methylphenol with NspF (Fig. 5 and Table 1). Comparison of the rate of oxygenation with that observed for A. bisporus (mushroom) tyrosinase (18) under identical conditions indicates that NspF has a 40× lower binding affinity and a 55× slower turnover rate for 4-methylphenol.

Fig. 4.

Names and structures of the tested NspF substrates.

Fig. 5.

Michaelis-Menten behavior of O₂ consumption (μM sec^-1) by NspF with 4-methylphenol in 500 mM borate buffer (pH 9.0) with excess hydroxylamine (10 mM) at 298 K. Inset: Hanes-Woolf plot.

Table 1.

Kinetic parameters for the coupled binuclear copper (CBC) enzyme family

Enzyme	Km (mM)	kcat (sec-1)	Ref.
Monophenolase Activity (4-methylphenol)
S. murayamaensis NspF	0.70 ± 0.04	2.22 ± 0.06	This work
A. bisporus Ty	0.0174	120	Yamazaki and Itoh (18)
O. vulgaris hemocyanin (Subunit g)*	6.5	0.099	Suzuki et al. (15)
Oxidoreductase Activity [2-amino-4-(carboxylic acid)-phenol]
S. murayamaensis NspF†	4.1 ± 0.6	3.5 ± 0.2	Noguchi et al. (16)
N. crassa tyrosinase‡	4.5 ± 0.1	0.83 ± 0.01	Toussaint and Lerch (17)
Hydroxyanilinase Activity [2-amino-4-(carboxylic acid)-phenol]
S. murayamaensis NspF†	9.8 ± 1.5	16 ± 2	Noguchi et al. (16)
N. crassa tyrosinase‡	N.D.§	N.D.§	Toussaint and Lerch (17)

^*6.0 M Urea;

^†30 °C;

^‡24 °C;

^§Not Detected.

While hydroxylamine is required for the monophenolase activity of NspF, it is not required for the production of nitrosophenols from o-aminophenols. In this case, the o-aminophenol can directly reduce resting NspF via a two-electron oxidation to produce the corresponding o-iminoquinone (16) and deoxy-NspF (Fig. 6), which can subsequently react with dioxygen to form oxy-NspF. Thus, the catalytic reaction with o-aminophenols by resting NspF can proceed without additional reductant.

Fig. 6.

Conversion of resting to oxy-NspF by an o-aminophenol (L = H₂O or OH^-). The axial His ligands have been omitted for clarity.

Computational Reaction Mechanism.

To gain insight into the geometric and electronic structure requirements for enzymatic hydroxyanilinase activity compared to monophenolase activity, density functional theory (DFT) calculations were performed on a truncated phenolate-bound dicopper peroxide model, {[(NH₃)₂Cu]-(O₂)-[Cu(X)(NH₃)₂]}²⁺ (X = 4-carboxamidophenolate or 2-amino-4-carboxamidophenolate). These calculations presume phenol substrate deprotonation upon coordination to the copper. Calculations specifically compared the electrophilic attack of a dioxygen derived oxygen atom on either the arene 2-position carbon (for the monophenolase reaction) or the 2-position amine (hydroxyanilinase reaction). The calculated reaction trajectories predict transition states consistent with electrophilic attacks, as observed in model studies of these reactions (10, 21). Insight into the geometric requirements for hydroxyanilinase versus monophenolase activity comes from superposition of the transition states from these two reaction pathways (Fig. 7). Specifically, while the Cu₂O₂ cores overlay with high fidelity (0.11 Å heavy atom r.m.s. deviation), the para substituent of the arene ring in the hydroxyanilinase substrate is distorted 3.6 Å away from the equivalent position on the monophenolase substrate. This distortion results from a reaction coordinate that is forming a larger metallacycle (five vs six membered ring for the monophenolase and hydroxyanilinase reaction, respectively) in the transition state. Thus, differences in protein–substrate interactions at the active site pocket would be expected to play a significant role in the stabilization (or destabilization) of the respective transition state.

Fig. 7.

Overlay of the transition states of the intramolecular {[(NH₃)₂Cu]-(O₂)-[Cu(X)(NH₃)₂]}⁺ reaction with X = 4-carboxamidophenolate (gray) and X = 2-amino-4-carboxamidophenolate (black). Since the two Cu₂O₂ cores overlay with high fidelity (see text), the second core has been omitted for clarity. See SI Appendix, Tables S1 and S2 for structural details.

Discussion

The generation and spectroscopic characterization of oxy-NspF indicates that dioxygen coordinates to the reduced site as peroxide in a μ-η²∶η² fashion. A comparison of oxy-NspF, oxy-tyrosinase, and oxy-hemocyanin indicates the presence of highly conserved spectral features, definitively assigning NspF as a CBC enzyme. The generation of oxy-NspF has also indicated that this intermediate possesses monophenolase activity, albeit with lower efficiency (both lower k_cat and higher K_m) than mushroom tyrosinase. This activity extends the linear reactivity pattern of the CBC protein family where oxy-NspF is capable of performing all of the tyrosinase functions (Fig. 1) while possessing the additional unique ability of oxygenating o-aminophenols. Thus enzymes with hydroxyanilinase activity, which currently include NspF and GriF (16, 22), constitute a fourth subclass of the CBC protein family (Fig. 1).

Mechanistic insight into this additional activity comes from the well characterized electrophilic reactivity of side-on binuclear copper peroxides in enzymes (1, 15, 18) and model complexes (10, 12, 23, 24). While Cu₂O₂ model complexes have also been shown to be competent for one-electron oxidations of substrate in a net-hydrogen atom abstraction mechanism (25–27), the oxidation of amino-phenols to iminoquinones is a two-electron process (28) and the oxidation of aryl-amines to the corresponding nitroso product by peroxide model complexes proceeds via an electrophilic mechanism (21). This suggests that the oxygenation of o-aminophenols by NspF proceeds via an electrophilic mechanism. In this mechanism (Fig. 8), deoxy-NspF is generated by the reduction of the resting enzyme by an o-aminophenol yielding the corresponding o-imminoquinone, analogous to the reduction of resting tyrosinase by o-diphenols to generate deoxy-tyrosinase (Fig. 8). Then, deoxy-NspF reacts with dioxygen to form an intermediate with peroxide bound in a side-on fashion, followed by coordination of the deprotonated o-aminophenol. While the illustrated mechanism shows the O─O bond remaining intact upon substrate coordination, O─O bond cleavage cannot be ruled out given results from model studies (12). The initial oxidation of the amine then proceeds via an electrophilic process resulting in the formation of an oxygenated intermediate (met-, Fig. 8). Finally, an intramolecular two electron oxidation of the coordinated o-(hydroxyamino)phenolate results in two Cu(I) ions that, upon product release, regenerate the deoxy form.

Fig. 8.

Proposed electrophilic mechanism for the hydroxyanilinase activity parallel the consensus monophenolase catalytic cycle (32, L = H₂O or OH^-). The axial His ligands have been omitted for clarity.

Within the framework of this mechanism, the experimental kinetic parameters (Table 1) provide insight into the unique reactivity of NspF. Since the measured oxidoreductase activities (o-imino quinone-forming activity) of NspF and tyrosinase with 2-amino-4-(carboxylic acid)-phenol (Fig. 4) are similar (16, 17), the difference in selectivity between the two enzymes for this substrate (i.e., the hydroxyanilinase activity observed for NspF but not tyrosinase) must originate from a difference in ΔG^‡ for the hydroxyanilinase activity (Table 1). From k_cat measurements on this substrate, the Eyring equation indicates that ΔΔG^‡ between NspF and N. crassa tyrosinase must be greater than 4 kcal mol^-1‡. The identical spectral features of the oxy form of both enzymes are inconsistent with differences in the Cu₂O₂ core being responsible for the differences in reactivity. Instead, our computational models (Fig. 7) suggest that this energy difference arises from substrate-protein interactions due to modified positioning of the phenolate rings in the transition states. A distortion of the substrate position is required to accommodate the formation of the six member metallacycle in the hydroxyanilinase reaction coordinate and suggests that unique substrate—protein interactions are necessary to achieve hydroxyanilinase function.

To develop insight into these interactions, a sequence alignment (SI Appendix, Fig. S3) of NspF with the two structurally characterized bacterial tyrosinases (7, 29) indicates a number of unique residues in NspF in the vicinity of the Cu₂O₂ core (SI Appendix, Fig. S4). Two of these substitutions involve the conversion of an aliphatic residue in both tyrosinase structures to a hydrogen bond acceptor (Asp46 and Met234) in NspF, which could potentially interact with the carboxamido group in NspF. Two regions in the vicinity of the active site (SI Appendix, Fig. S4) have lower sequence similarity: the eleven residues after the first His ligand (39 in NspF) and a loop (SI Appendix, Fig. S3) that contains a six residue insertion in NspF. Given these differences, the precise role of the residues in directing the unique function of NspF will require further study.

These results build on the growing body of evidence that the active sites of CBC enzymes are intrinsically reactive and that the specific function of each subclass results from substrate interaction in the protein pocket with specific second sphere residues rather than direct electronic tuning of the Cu₂O₂ core.

Materials and Methods

NspF Expression and Purification.

Recombinant S. murayamaensis NspF was expressed and purified as described previously (16) with minor modifications (SI Appendix). Protein concentrations were determined from the intensity of the 280 nm absorption feature (83.1 mM^-1 cm^-1 based on the amino acid sequence), and protein purity was confirmed by SDS-PAGE. Copper concentrations were determined spectrophotometrically with 2-2′-biquinoline (30), which indicated that more than 95% of the sites were loaded with copper.

UV–Vis.

Samples of oxy-NspF were prepared by addition of a hydroxylamine HCl solution to 20× excess to NspF (43 μM) in phosphate∶glycerol buffer (90∶10 v%, pH 7.4) while stirring under ambient conditions (298 K). Reactions with 4-carboxamidophenol monitored optically were initiated by addition of 4.0 μL of 9.0 mM solutions of the phenol in buffer.

Resonance Raman.

Samples of oxy-NspF (850 μM) in phosphate:glycerol buffer (90∶10 v%, pH 7.4) were prepared in 5 mm J-Young adapted NMR tubes by anaerobic addition of hydroxylamine HCl (20× excess) followed by exposure of the solution to either natural abundance or labeled dioxygen at room temperature. Spectra were recorded on an Andor Newton BR-DD CCD detector with a Spex 1877 CP triple monochromator that contained a holographic grating with 2,400 grooves per millimeter to yield a spectral resolution of 2 cm^-1. A Coherent Innova Sabre 25/7 Ar⁺ ion laser was used for laser excitation with 20 mW power at the sample. Data was collected for 10 min per spot while the samples were rotated at 228 K.

Product Analysis.

A catalytic amount of NspF (200 nM) was reacted with excess 4-methylphenol (1.02 mM). The mixture was loaded directly onto an HPLC for analysis (Agilent 1200 Infinity with an 4.6 × 150 mm Eclipse XDB-C18 column). Chromatographic separation was achieved with a water:acetonitrile solvent system. Ten microliters of sample was injected at 5% acetonitrile. The solvent was maintained at this ratio until 2.5 min, when the composition of acetonitrile was increased linearly to 70% at 8 min and remained isocratic until the run completed (10 min). Samples prepared using a catalytic amount of recombinant S. glaucescens tyrosinase (60 nM) were analyzed in an analogous manner for comparison to the NspF product mixture.

Oxygen Uptake.

Oxygen consumption was measured using a Clark-type oxygen electrode (Microsensor OX-MR from Unisense) in calibrated approximately 2 mL micro-respiration chambers in a temperature bath at 298 K while stirring at 300 rpms. Solutions containing hydroxylamine HCl (10 mM) and substrate (0.10–1.01 mM) were prepared in buffer (100 mM phosphate, pH 6.8 or 500 mM borate, pH 9.0 in DI) and pre-equilibrated in the temperature bath for 20 min. Oxygen consumption was initiated by adding 10 μL of enzyme (40 μM) for a final concentration of 200 nM. Rates of oxygen consumption were determined from the slope of the initial (linear) decrease in oxygen concentration at each substrate concentration. Triplicate measurements of each rate were averaged.

Calculations.

Both spin restricted and unrestricted density functional theory calculations were performed on small and large models of the Cu₂O₂ site. Restricted calculations on the {[(NH₃)₂Cu]-(O₂)-[Cu(X)(NH₃)₂]}²⁺ model were performed using the hybrid functional B3LYP with the Ahlrich triple-ξ basis set with polarization (TZVP) on an ultrafine integration grid as implemented in Gaussian 09 (31). Optimizations were performed solvated (default parameters with THF as solvent) to account for the hydrophobic protein environment and for comparison to the results on model complexes in this solvent. Analytical frequencies were calculated to verify that the appropriate stationary points and transition states had been obtained. Optimization of a coordinated phenolate to a μ-η²∶η² peroxo core resulted in the cleavage of the O─O bond in the side-on peroxo isomer to form a bis-μ-oxo isomer, in agreement with experimental observations (12). For the coordinated 2-amino-4-carboxamidophenolate, the spin-restricted formalism was utilized with this small model, as attempts to perform spin-unrestricted calculations lead to the transfer of an electron from the substrate to one Cu ion to generate a semiquinone adduct. Parallel calculations with the spin unrestricted formalism were also performed on a larger model system derived from the crystal structure of oxy-tyrosinase (PDB code 1WX2) (7). In this model, the six His ligands were replaced with methyl imidazoles and the α carbon atoms were constrained at their crystallographic positions during geometry optimizations. These calculations also give an electrophilic reaction coordinate but with no electron transfer from the substrate to the Cu in the reactant complex. An overlay of the transition states from the larger model indicated a similar distortion between the bound substrates as observed in the small model shown in Fig. 7. See SI Appendix for additional computational details.