Thiol dioxygenase turnover yields benzothiazole products from 2-mercaptoaniline and O₂-dependent oxidation of primary alcohols

Abstract

Thiol dioxygenases are non-heme mononuclear iron enzymes that catalyze the O₂-dependent oxidation of free thiols (-SH) to produce the corresponding sulfinic acid (-SO₂⁻). Previous chemical rescue studies identified a putative Fe^III-O₂^•− intermediate that precedes substrate oxidation in Mus musculus cysteine dioxygenase (Mm CDO). Given that a similar reactive intermediate has been identified in the extradiol dioxygenase 2, 3-HCPD, it is conceivable that these enzymes share other mechanistic features with regard to substrate oxidation. To explore this possibility, enzymatic reactions with Mm CDO (as well as the bacterial 3-mercaptopropionic acid dioxygenase, Av MDO) were performed using a substrate analogue (2-mercaptoaniline, 2ma). This aromatic thiol closely approximates the catecholic substrate of homoprotocatechuate of 2, 3-HPCD while maintaining the 2-carbon thiol-amine separation preferred by Mm CDO. Remarkably, both enzymes exhibit 2ma-gated O₂-consumption; however, none of the expected products for thiol dioxygenase or intra/extradiol dioxygenase reactions were observed. Instead, benzothiazoles are produced by the condensation of 2ma with aldehydes formed by an off-pathway oxidation of primary alcohols added to aqueous reactions to solubilize the substrate. The observed oxidation of 1°-alcohols in 2ma-reactions is consistent with the formation of a high-valent intermediate similar to what has been reported for cytochrome P450 and mononuclear iron model complexes.

Graphical Abstract

Introduction

Cysteine dioxygenase is a non-heme mononuclear iron enzyme that catalyzes the O₂-dependent oxidation of L-cysteine (cys) to produce cysteine sulfinic acid (csa). Only two mammalian thiol dioxygenases have been identified to date, cysteine dioxygenase (CDO) and cysteamine (2-aminoethanethiol) dioxygenase (ADO). Among these, CDO is the best characterized [1–5]. Intracellular cys concentration is limiting in glutathione synthesis, therefore the activity of CDO directly competes with cellular redox buffering under conditions of low cys availability and oxidative stress [6]. Abnormal cys metabolism has been observed in patients suffering from a variety of neurological diseases [5, 7, 8], suggesting a correlation between impaired sulfur metabolism, oxidative stress, and neurodegenerative disease [9, 10]. Consequently, enzymes involved in sulfur-oxidation and transfer are increasingly being recognized as potential drug targets for development of antimicrobials, therapies for cancer and inflammatory disease [11–14].

Shown in Figure 1A is the crystal structure for the Rattus norvegicus CDO (pdb code 4IEV) active site [15]. The iron first coordination sphere is comprised of iron ligated along one octahedral face by the N_ε-atoms of H86, H88, and H140. Prior to the addition of O₂, cys coordinates to the mononuclear ferrous site of CDO in a bidentate fashion through a thiolate and neutral amine [16–18]. As illustrated in Fig. 1B, ligation of the cys thiol-group within the trigonal plane opposite H86 and H88 positions the carboxylate group of cys favorably for charge stabilization by the R60 guanidinium group.

Figure 1.

Crystal structure of the resting and substrate-bound Rattus norvegicus CDO.

Crystal structure of the Rattus norvegicus CDO active site (A, pdb code 4IEV) as compared to the substrate-bound enzyme (B, pdb code 4IEZ) [15]. Selected distances A: 1 (2.63 Å); 2 (2.73 Å); 3 (2.72 Å); B: 1′ (2.78 Å); 2′ (2.82 Å); 3′ (2.91 Å); and 4′ (3.10 Å).

It was previously demonstrated that the catalytic cycle of CDO can be ‘primed’ by 1-electron through chemical oxidation to produce CDO with ferric iron in the active site [19]. While catalytically inactive, the substrate-bound form of Fe^III-CDO is more amenable to interrogation by UV-visible and EPR spectroscopy than the ‘as-isolated’ Fe^II-CDO enzyme. Chemical-rescue experiments were performed in which superoxide (O₂^•−) was introduced to the substrate-bound Fe^III-CDO to produce and characterize by EPR and UV-visible spectroscopy a transient intermediate kinetically matched to csa formation. This intermediate was tentatively assigned as a substrate-bound Fe^III-superoxide species similar to that spectroscopically observed for the extradiol cleaving catechol dioxygenase, homoprotocatechuate 2, 3-dioxygenase, (2, 3-HPCD) [20, 21]. While the decay of this intermediate was kinetically matched to the formation of product, the rate of csa produced was significantly slower than the steady-state turnover for Mm CDO. It is therefore not clear if this represents a catalytically relevant intermediate. More recently, stopped-flow UV-visible studies identified a transient species upon rapid mixing of the substrate-bound CDO with saturating oxygen. Supporting computational modeling of the transient UV-visible spectra suggest that this transient species is consistent with a cyclic [Fe-O-O-S_Cys]-structure with a quintet (S = 2) spin state. However, rapid freeze-quench Mössbauer experiments were unable to corroborate this assignment [22]. Therefore, direct spectroscopic validation of the transient intermediate produced during native catalysis is absent.

Catechol dioxygenases are key metabolic enzymes found within soil bacteria having the capacity to degrade aromatic compounds, which can be utilized as a carbon source [23, 24]. These enzymes are classified into two primary families based on the position of substrate ring cleavage and oxidation state of the catalytically essential iron within the non-heme active site. For instance, intradiol dioxygenase enzymes such as catechol 1, 2-dioxygenase (CTD) and protocatechuate 3, 4-dioxygenase (3, 4-PCD) catalyze the O₂-dependent carbon-carbon bond cleavage between the substrate phenolic hydroxyl groups to yield a dicarboxylic acid muconate product. This class of enzymes stabilizes a ferric iron (Fe^III) resting state, which activates the substrate for electrophilic attack by oxygen. It has been proposed that the high activation barrier for direct O₂-attack on the substrate is overcome by resonance delocalization of unpaired spin-density onto the substrate from the substrate-bound high-spin ferric site to facilitate a spin-allowed reaction with triplet O₂ [20, 23]. Alternatively, extradiol dioxygenases such as the aforementioned 2, 3-HCPD cleave the carbon-carbon bond adjacent to the ortho-hydroxyl groups to yield 5-carboxymethyl-2-hydroxymuconic semialdehyde. These enzymes stabilize a resting ferrous iron (Fe^II), or in rare instances Mn^II, in the active site to activate molecular oxygen for subsequent attack on the substrate [25]. For these enzymes, reversible binding of the catechol substrate precedes O₂-binding at the substrate-bound ferrous active site. Oxidative addition of O₂ to the ferrous iron yields a short-lived ferric-superoxo [Fe^III-O₂^•−] species ultimately responsible for nucleophilic attack on the aromatic substrate [26].

Assuming both 2, 3-HPCD and CDO produce a Fe^III-superoxo intermediate prior to substrate oxidation, it is possible that additional mechanistic parallels exist in reactions utilizing aromatic substrates. Steady-state specificity experiments with the Mus musculus CDO (Mm CDO) demonstrated that the minimal substrate for CDO must contain both amine and thiol functional groups, and that optimal activity and efficiency is observed when these groups are separated by two carbon units [27]. With this in mind, 2-mercaptoaniline (2ma) could provide an interesting comparison to the aforementioned catechol dioxygenase class of non-heme mononuclear iron enzymes. It should be noted that the bonds separating the thiol and amine functional groups in the aromatic 2ma substrate are sp² hybridized. As these bonds cannot be rotated to the extent of aliphatic thiols, the 2ma-bound Fe-site geometry is expected to be significantly perturbed relative to the native ES-complex. Another factor to consider is the differential reactivity of aromatic versus aliphatic thiols. Potentially, these deviations may result in unique reactivity.

Neglecting potential hydrolysis, cyclization, or tautomerization, Scheme 1 illustrates the expected reaction products for analogous thiol dioxygenase, extradiol-, or intradiol cleaving activity.

Scheme 1.

Potential reaction products for O₂-dependent 2ma oxidation by Mm CDO

Potential reaction products for O₂-dependent 2ma oxidation catalyzed by Mm CDO assuming (1) thiol dioxygenase [aminobenzenesulfinic acid], (2) extradiol dioxygenase- [2-mercapto-6-oxohexa-2, 4-dienamide or 2-amino-6-oxohexa-2, 4-dienethioic S-acid], or (3) intradiol dioxygenase cleaving activity [6-amino-6-oxohexa-2, 4-dienethioic S-acid].

Presented here are the results of O₂-dependent 2ma reaction using two different thiol dioxygenase enzymes [Mm CDO and 3-mercaptopropionic acid dioxygenase isolated from Azotobacter vinelandii (Av MDO)]. These enzymes bind substrates bidentate [16–18] and monodentate [28, 29], respectively and thus may exhibit differential reactivity.

Materials and Methods

Enzyme purification

Recombinant Mus musculus cysteine dioxygenase (Mm CDO) and Azotobacter vinelandii 3-mercaptopropionic acid dioxygenase (Av MDO) were expressed, purified, and activity verified as described previously [29, 30]. For all batches of enzyme used in these experiments, spectrophotometric determination of ferrous and ferric iron content was measured using 2, 4, 6-tripyridyl-s-triazine (TPTZ). A detailed description of this method is presented elsewhere [18, 29]. For clarity, the concentrations reported in enzymatic assays reflect the concentration of ferrous iron within enzyme samples. The protein content in each sample was determined by Bio-Rad protein assay. Steady-state kinetic parameters observed at 20 ° C for Mm CDO and Av MDO were consistent with previously published values (Mm CDO; k_cat, 0.7 ± 0.1 s⁻¹; K_M, 2.1 ± 0.3 mM; k_cat/K_M, 330 ± 50 M⁻¹ s⁻¹; Av MDO; k_cat, 0.5 ± 0.1 s⁻¹; K_M, 18 ± 2 μM; k_cat/K_M, 28,000 ± 3000 M⁻¹ s⁻¹) [28, 29].

Reagents and stock solutions

2-mercaptoaniline (2ma) (99%) was purchased from Sigma-Aldrich (cat. no. 274240) and stored under argon within an anaerobic chamber to prevent disulfide formation. Stock solutions of 2ma (0.75 mM) were prepared anaerobically in ethanol (95%, ACROS) prior to subsequent dilution in aqueous buffer (10mM HEPES, 50 mM NaCl, and pH 8.5) for enzymatic assays. The benzothiazole (bt) [96%, cat. no. 101338] and 2-methylbenzothiazole (2m-bt) [99%, cat. no. 112143] standards used for spike assays and LC-MS/MS (MRM) verification of products were purchased from Sigma-Aldrich and used without further purification. However, 2-phenybenzothiazole (2p-bt) was synthesized as described below. All standards were prepared by serial dilution of a ca. 5 mM stock solution in 80:20 (buffer:ethanol). Spike assays were performed by serial dilution of the stock solution to 1–5 μM prior to HPLC analysis.

Synthesis of 2-phenybenzothiazole (2p-bt)

Synthesis of 2-phenylbenzothiazole was performed as described elsewhere [31]. Briefly, 2-mercaptoaniline (250 mg, 2 mmol), benzaldehyde (212 mg, 2 mmol) and excess acetic acid were heated for 4 hours at reflux temperature. After cooling it to room temperature, the reaction mixture was added into an ice water mixture. The solid precipitated out was collected by suction filtration. The crude solid was crystallized from ethanol/water to give 345 mg of 2p-bt in 82% yield as colorless solid.

Characterization of 2p-bt product was performed by 500 MHz ¹H and 125 MHz ¹³C NMR experiments performed on a JOEL Eclipse Plus 500 NMR spectrometer. Chemical shifts were recorded in reference to residual solvent peaks (CDCl₃ = 7.26 ppm). FT-IR spectra were recorded on a Bruker Alpha-P FT-IR Spectrometer by attenuated total reflectance on a diamond sample plate. ¹H NMR (500 MHz, CDCl₃) δ 8.13 – 8.09 (m, 3H), 7.89 (ddd, J = 8.0, 1.2, 0.6 Hz, 1H), 7.53 – 7.47 (m, 4H), 7.40 – 7.36 ppm (m, 1H); ¹³C NMR (126 MHz, CDCl₃) δ 168.1, 154.2, 135.1, 133.7, 131.0, 129.1, 127.6, 126.4, 125.3, 123.3, 121.7; IR (Neat): 3064, 3018, 1477, 1432, 1312, 1224, 962, 757, 729. Melting point (113–114 °C) was obtained in capillary tubes on a Mel-Temp II apparatus, and the thermometer was uncorrected. These values are in agreement with those reported in the literature [32].

UV-visible experiments

UV-visible measurements were performed on an Agilent 8453 photo diode array spectrometer (Santa Clara, CA). Sample temperature was held constant by a 13L circulating water bath and an Agilent thermostatable cell holder (89054A) with magnetic stirrer. All measurements were made in ES Quartz cuvettes (NSG Precision Cells, Farmingdale, NY). For time dependent assays, a baseline correction at 900 nm was used. Spectra collected in kinetic mode used a 4-second interval between measurements for the first 3-minutes. Beyond this point, time intervals were increased by 5% up to 15 minutes.

In a typical UV-visible experiment, an O₂-saturated buffer [10mM HEPES, pH 8.5 at 20 °C] was diluted with anaerobic buffer and ethanol to obtain the desired O₂ concentration and 20% (v/v) ethanol (1.0 mL final volume). This solution was used to blank the UV-visible spectrometer. The stock 2ma solution was then added to obtain substrate concentrations ranging from ca. 30–130 μM. Reactions were initiated by injection of enzyme (final concentration, 2 μM).

Oxygen Electrode

Consumption of molecular oxygen was verified polarographically using a standard Clark electrode (Hansatech Instruments, Norfolk, England) in a jacketed 2.5 mL cell. Calibration of O₂-electrode is described in detail elsewhere [30, 33]. All reactions were initiated by addition of 2.0 μM enzyme (Mm CDO or Av MDO) under identical buffer conditions as described for UV-visible experiments. Reaction temperatures were maintained at 20 ± 2 °C by circulating water bath (ThermoFlex 900, Thermo Scientific).

HPLC sample preparation

Samples were prepared for select reaction time points ranging from 1–30 minutes by removing an aliquot of the reaction (180 μL) and quenching by addition of 5 μL 2N HCl. Samples were purified by 0.22 μM cellulose acetate membrane filtration (Corning, Spin-X) prior to analysis. As a control, a zero point sample was prepared by substituting heat-denatured enzyme in the above reaction. Unless otherwise noted, the concentration of 2ma, oxygen, and alcohol in reaction buffers was fixed at 150 μM, 420 μM (35% saturated), and 20% (v/v), respectively.

HPLC instrumentation and analysis

HPLC samples were analyzed using a Dionex UltiMate 3000 HPLC equipped with a 3000 variable wavelength detector. Column, Phenomex Luna (5 μm) C18 100 Å, column [250 x 4.6 mm] with accompanying C18 [4 x 3.0 mm] guard column; mobile phase, 35:65 HPLC grade H₂O:methanol; injection volume, 20 μL; isocratic flow rate, 0.5 mL/min; column temperature, 25 °C. UV-visible detection was monitored at 270 nm. Calibration curves for 2ma, bt, 2m-bt, and 2p-bt were prepared for determination of product concentration. Serial dilution of benzothiazole standards exhibited a linear response in concentrations ranging from 10–1100 μM.

LC-MS/MS (multiple reaction monitoring)

Verification of enzymatic product was performed by multiple reaction monitoring (MRM) using a triple quadrupole LC-MS/MS [Shimadzu Scientific Instruments, LCMS 8040] [34–36]. For each experiment, 10 μL of the benzothiazole standard (described above) was injected to determine the product ions and relative distribution for each ion. The molecular ion [M−H]⁺ for each specific benzothiazole standard [bt, 136 m/z; 2m-bt, 150 m/z; and 2p-bt, 212 m/z] was selected for secondary fragmentation in positive ion mode [37]. MRM optimization was then employed to maximize transition intensity and sensitivity for each fragment allowing for quantitation of product ions. The optimized MRM method was used to verify the formation of specific benzothiazole product ions and their relative intensities in enzymatic reactions by direct injection of assays.

GC-MS/MS sample preparation

A standard solution of benzaldehyde was prepared by dissolving 2 μL of benzaldehyde in 500 μL of dichloromethane. The same procedure was used for making standard solution of acetaldehyde. Detection of benzaldehyde in analyte samples was performed after extraction with 500 μL dichloromethane. Detection of acetaldehyde in analyte samples was performed with direct injection of the sample, without extracting with dichloromethane. Note: Acetaldehyde and benzaldehyde are highly volatile and therefore all extractions were performed in the sample vial using a Hamilton gas-tight syringe to prevent loss of volatiles.

GC-MS/MS Instrumentation and Method

GC-MS experiments were conducted in Shimadzu Scientific Instruments, GCMS-QP2010, using electron impact ionization (EI) at 150 eV and a mass selective detector. The method used for detecting acetaldehyde in the GC experiment represents 0.5 min @ 50 °C – 30 °C/min – 7.5 min @ 260 °C with injection volume of 1μL. For benzaldehyde, the GC experiment represents 1.0 min @ 50 °C – 30 °C/min – 8.0 min @ 260 °C with injection volume of 1 μL.

Results and Discussion

CDO catalyzes the O₂-dependent oxidation of a variety of substrates which contain both amine and thiol functional groups separated by two carbon atoms. Understandably, the catalytic efficiency (k_cat/K_M) for substrates other than cys is poor, nevertheless product formation is easily observed at concentrations amenable to HPLC, NMR, and LC-MS/MS analysis [27]. Kinetic experiments were performed on an Agilent photodiode array spectrometer as formation of the 2ma-sulfinic acid product (2-aminobenzensulfinate, 2abs) is expected to be red-shifted relative to the starting substrate. Given the low solubility of 2ma, ethanol was added to the enzymatic reactions to increase the upper limit of substrate concentration in solution. As a control, equivalent Mm CDO assays with cys were carried in the same buffer to verify that the addition of ethanol did not abolish enzymatic activity. In assays performed in 20% (v/v) ethanol, only a modest loss (10–20%) of Mm CDO specific activity was observed within the time scale of the experiments (10–20 minutes).

As illustrated in Figure 2, introduction of enzyme results in rapid change in the UV-visible spectrum ranging from 250 to 450 nm. A single isosbestic point at 318 nm confirms the presence of at least 2 species in reaction mixtures. As illustrated in Fig. S1A, kinetic flooding experiments at fixed Mm CDO (2 μM) and O₂ (420 μM) exhibit a linear increase in the observed rate (v_obs) with increasing 2ma concentration. Similarly, Fig. S1B demonstrates that v_obs also increases linearly with increasing enzyme concentration at fixed 2ma (125 μM) and O₂ (420 μM) concentration. Finally, a linear increase in v_obs is also observed with increasing O₂-concentration (Fig. 2, inset). These experiments verify that the reaction is first-order with respect to all reactants (Mm CDO, O₂, and 2ma). However, full enzymatic saturation is not observed within the 2ma (30–150 μM) or oxygen (50–900 μM) concentration range utilized in these experiments. Crucially, no change in the UV-visible data is observed upon addition of heat inactivated enzyme thus verifying that this reaction is not initiated by adventitious iron within enzyme samples.

Figure 2.

UV-visible spectra of 2ma-reactions and pseudo first order kinetic plot.

Representative UV-visible spectra illustrating the O₂-dependent reaction of 2-mercaptoaniline (2ma) catalyzed by Mm CDO. Selected time points: 1.4 s (purple trace, ○), 50 s, 120 s (green, Δ), 300 s and 600 s (red, □). Assay conditions: 2 μM Mm CDO, 420 μM O₂, 20% ethanol (v/v), 10 mM HEPES, pH 8.5 and 25 ± 2 °C. Inset. Pseudo first-order kinetics of initial rate (v_obs) as a function of oxygen concentration. The progress of the reaction at fixed and saturating 2ma (130 μM) was monitored spectrophotometrically at 305 nm.

Consumption of oxygen was verified using a standard Clarke-type electrode. Representative O₂ electrode results are shown in Supplementary Information (Figure S2). In these studies, equivalent reactions as described above were initiated by injection of enzyme. As with UV-visible experiments, the rate of O₂-consumption increased linearly with increasing enzyme and 2ma concentration. As an additional control, reactions were performed using benzenethiol as a substrate. However, neither CDO nor MDO exhibited any significant consumption of oxygen upon initiation of benzenethiol-reactions (Supplemental Information, Figure S2).

A translational benefit of this reaction is that 2ma can be used to optically screen for thiol dioxygenase activity. As shown in Fig. 2, enzymatic 2ma-reactions result in the formation of absorption features ~350 nm and thus a yellow color is apparent as the reaction progresses. While spectrophotometric and O₂-electrode results are useful for interrogation of the relative timing of chemical events, product analysis and assignment was performed using HPLC and LC-MS/MS as described in Materials and Methods. Figure 3 shows selected HPLC traces (spectrophotometric detection at 305 nm) for reactions of Mm CDO with 2ma in 20% (v/v) ethanol. Within this window, two peaks observed at retention times of 11.6 and 16.2 min show a clear time-dependent behavior. The 11.6 min peak decreases in intensity with time and can easily be identified as the starting 2ma by the matching retention times observed in calibration curves. LC-MS (and LC-MS/MS) was run in positive detection mode using the same mobile phase, flow rate, and column to observed the molecular ion [M−H]⁺ (150 m/z) of the species eluted with a retention time of 16.2 min. Since the expected mass for all dioxygenase products utilizing 2ma is 157 Da, the 150 m/z [M−H]⁺ species cannot be attributed to any of the reactions predicted in Scheme 1.

Figure 3.

Reverse phase HPLC chromatogram for 2ma-reactions at selected time points.

Representative HPLC chromatograms illustrating the change in analyte composition following introduction of enzyme. The initial composition of the reaction mixture is indicated by the bottom trace. The above chromatograms were quenched following 1- and 5-minutes reaction with Mm CDO. The peaks associated with 2ma (retention time, 11.6 min) and 2m-bt (retention time, 16.2 min) were verified by matching retention time with analytical standards, spike assays, and LC-MS/MS.

A fortuitous observation was made in equivalent reactions carried out where methanol was substituted for ethanol. In these reactions, the 16.2 min peak shifted to a lower retention time and the observed molecular ion [M−H]⁺ decreased to 136 m/z. The -14 amu shift in the molecular ion clearly suggests the absence of a (-CH₂-) fragment in methanol reactions as compared to ethanol. Similarly, if benzyl alcohol is substituted for ethanol, a positive molecular ion peak of 212 m/z is observed (+76 amu shift relative to the product formed in methanol). In all cases, the size of the molecular ion [M−H]⁺ can be calculated by the sum of 2ma (125 amu) and primary alcohol molecular weight (methanol, 32 amu; ethanol, 46 amu; and benzyl alcohol, 108 amu) followed by subtraction of water (18 amu) and 3 hydrogen atoms. This observation indicates that the alcohol utilized in this reaction becomes directly incorporated into the oxidized product.

Based on similar reactivity reported for flavin mimics [38], cytochrome P450 [39], and mononuclear iron model complexes [40–42], we hypothesize that O₂-dependent 2e⁻ oxidation of the solvent primary alcohol results in formation of aldehydes. Rapid and near quantitative condensation of an aldehyde with the 2ma starting material is anticipated based on the well-established nucleophilic addition of amines. Following this step, it is reasonable to expect formation of a benzothiazole product as illustrated in Scheme 2.

Scheme 2.

Proposed mechanism for benzothiazole formation.

Proposed reaction mechanism for benzothiazole formation.

Many synthetic methods describe the oxidative formation of 2-substituted benzothiazoles from 2-aminothiophenol (Scheme 2, compound 1) and aldehydes under various aerobic and dehydrogenative conditions [38, 43–45]. Generally, this transformation can be envisioned from the initial, reversible condensation of an aldehyde, e.g. acetaldehyde, with the amine to form iminium intermediate 2. This species undergoes rapid intramolecular cyclization with the nucleophilic thiol. At the working pH and with the electron withdrawing iminium substituent, it is likely that the sulfur nucleophile can form a significant concentration of the thiolate intermediate, 3, further favoring the cyclization reaction. After cyclization to benzothiazoline, 4, dehydrogenative formation of the benzothiazole product (similar to NADPH oxidation) is quite facile with a variety of mild oxidants (designated [Ox] in Scheme 2).

This mechanistic model correctly predicts the resulting molecular ion size observed in the aforementioned reactions. Further corroboration for the proposed reaction can be taken tandem LC-MS/MS [multiple reaction monitoring (MRM)] results obtained for reaction mixtures. In these experiments, standard solutions (1 μM each) of benzothiazole (bt), 2-methylbenzothiazole (2m-bt), and 2-phenybenzothiazole (2p-bt) were prepared as described in Materials and Methods for direct injection. The [M+H]⁺ molecular ion for each standard (bt, 136 m/z; 2m-bt, 150 m/z; and 2p-bt, 212 m/z) was selected for secondary fragmentation and MRM optimization was then employed to maximize transition intensity and sensitivity for each fragment allowing for quantitative analysis of product ions. The optimized MRM method was used to compare the product of each enzymatic reaction. For simplicity, all enzymatic reactions were quenched at 5 minutes after injection of enzyme (2 μL).

Figure 4 (panel 1) shows the MRM spectra for a standard solution of 2m-bt in the same assay buffer used for kinetic experiments. In addition to the 150 m/z parent [M+H]⁺ ion, two additional ions are observed at 65 and 109 m/z. For comparison, the direct injection of the Mm CDO catalyzed reaction with 2ma shown in Fig. 4 (panel 2) yields an identical fragmentation pattern. Both the matching fragmentation pattern and relative intensities verify that product generated in O₂-dependent assays with Mm CDO is indeed 2-methylbenzothiazole. Final corroboration was obtained by spiking known quantities of 2m-bt standards into enzymatic assays. As expected, 2m-bt coelutes exactly with the 16.2 min peak observed in HPLC chromatograms. These observations identify 2m-bt as the product of enzymatic reactions with 2ma in the presence of ethanol. For comparison, Figure 5 illustrates the same analysis used to verify formation of bt and 2p-bt in reactions utilizing 20% (v/v) methanol and benzyl alcohol, respectively.

Figure 4.

MRM spectrum of 2m-bt compared to 2ma-reactions in ethanol.

LC-MS/MS spectra of MRM [M−H]⁺ ions observed for the 2-methylbenzothiazole (Mol. Wt. mass, 149.21 Da) standard (panel 1). The lower panels represent direct injection of Mm CDO (2) and Av MDO (3) assays following 10 minutes reaction time. Assay conditions are identical as those described for Fig. 3.

Figure 5.

MRM spectrum of bt and 2p-bt compared to 2ma-reactions in methanol (A) and benzyl alcohol (B).

LC-MS/MS spectra of MRM [M−H]⁺ ions observed for the benzothiazole (A, Mol. Wt., 135.19 Da) and phenyl-benzothiazole (B, Mol. Wt., 211.28 Da) standards are shown in panel 1. Direct injection of the enzymatic assays carried out with Mm CDO in methanol (A) and benzyl alcohol (B) are shown in panel 2. Equivalent Av MDO reactions are shown in panel 3.

To this point, the evidence for aldehyde formation by enzymatic oxidation of primary alcohols has been largely inferred based on the products observed. Therefore, GC-MS analysis was performed on extracts of reactions mixtures to directly verify their presence and corroborate the proposed mechanism shown in Scheme 2. For this experiment, benzyl alcohol was selected as the co-solvent as its lower volatility (relative to formaldehyde and acetaldehyde) would decrease the possibility for loss during organic extraction. As shown in Figure 6A (trace 1), beyond the void volume peak at 1.94 min, a single peak is observed at a retention time of 2.94 minutes upon injection of a benzaldehyde standard. Enzymatic assays were performed identically to those described previously for LC-MS experiments. Reactions were initiated by addition of enzyme and allowed to turnover for 20 minutes prior to inactivation by spin-filtration (Corning, Spin-X) and extraction of organic components with dichloromethane. GC-MS chromatograms of reaction extracts exhibit a major peak at 3.33 minutes which corresponds to the benzyl alcohol solvent. While small relative to the benzyl alcohol peak, an additional peak matching the retention time of benzaldehyde (2.94 min) is readily observed in extracts of enzymatic reactions.

Figure 6.

GC-MS/MS chromatogram and MRM spectrum of benzaldehyde as compared to 2ma-reaction performed in benzyl alcohol.

GC-MS/MS chromatogram (A) and MRM [M−H]⁺ spectra (B) of enzymatic 2ma-reaction extracts. Panel 1 shows the secondary fragmentation pattern of the parent [M−H]⁺ benzaldehyde ion (107 m/z) observed in standard reactions (retention time of 2.94 min). Panel 2 shows the comparative MRM spectra for the enzymatic (Mm CDO) 2ma-reaction dichloromethane extract.

Shown in Figure 6B is the tandem mass spectrum (MRM) for the standard benzaldehyde peak compared to that observed for enzymatic reaction extracts. The molecular ion for benzaldehyde (106 m/z) was selected for secondary fragmentation and the optimized MRM analysis was run in positive detection mode. As illustrated in Fig. 6B (trace 1), beyond the molecular ion, a number of additional fragmentation ions are observed for the benzaldehyde standard (39, 51, 77, 78, 105, 106, and 107 m/z). For comparison, the mass fragmentation and relative intensities of ions observed in reaction extracts (Fig. 6B, trace 2) matches that of the benzaldehyde standard confirming the presence of benzaldehyde in reaction mixtures.

Similarly, MRM spectra obtained by direct injection of reactions carried out in 20% (v/v) ethanol verify the formation of acetaldehyde in enzymatic 2ma-reactions (Supporting Information, Figure S3). As with reactions performed in benzyl alcohol, the secondary fragmentation pattern of reaction extracts carried out in ethanol match acetaldehyde standards. Most important, control reactions performed in ethanol containing enzyme but lacking the 2ma substrate (Fig. S3, trace 3) exhibit a markedly different fragmentation pattern as either standards or enzymatic 2ma-reactions indicating that acetaldehyde is not present in the starting alcohol solvent and is only produced during enzymatic turnover with 2ma.

The scope of this chemistry was further investigated by substituting 20% (v/v) dimethyl sulfoxide (DMSO) rather than a primary alcohol as this solvent cannot form an aldehyde upon oxidation. As predicted by the proposed mechanism shown in Scheme 2, no benzothiazole products are observed in DMSO reactions. Moreover, no evidence for the formation of 3abs [157 Da] was observed by LC-MS either. Similar results were obtained in reactions with dimethylformamide (DMF) as a co-solvent and in reactions lacking any primary alcohol co-solvent. Reactions were performed utilizing the secondary alcohol 2-propanol as well but these failed to produce the expected 2, 2-dimethylbenzothiazoline product. This results of this experiment are discussed in greater detail within the Supplemental Information, Scheme 1 and Fig. S4. Collectively, these results imply that the formation of benzothiazoles is limited to the use of primary alcohols.

With the identity of reaction products bt, 2m-bt, 2p-bt confirmed by overlapping HPLC retention time and LC-MS/MS MRM, analytical determination of product yield was conducted by HPLC by comparison to primary standards. Given the proposed reaction shown in Scheme 2 it is not surprising that the product yield for these reactions is fairly low. For example, in reactions yielding 2m-bt, only 43 ± 8 μM product is observed within the first 5 minutes of the reaction. As enzyme concentration is 2 μM, this suggests 2m-bt is produced over multiple enzymatic turnovers. Beyond 10 minutes, no appreciable increase in 2m-bt concentration is observed. In fact, 2m-bt concentration appears to decrease slightly suggesting either precipitation of the benzothiazole product or further decay. Likely the rapid inactivation of the enzyme can be explained by either the release of ROS or protein denaturation precipitated by the solvent alcohol. Reactions yielding bt and 2p-bt follow a similar behavior but with attenuated product yields at 5 minutes (28 ± 6 and 3.5 ± 1.2 μM, respectively).

This reaction is first order with respect to oxygen concentration. However, from the perspective of mass balance, the absence of O-atom incorporation into the benzothiazole product (or oxidized 1°-alcohol solvents) indicates that this reaction is completely uncoupled with respect to oxygen; suggesting that all O₂ consumed over the course of the reaction is released as reactive oxygen species (ROS; H₂O₂, O₂^•−, ^•OH). Therefore, an obvious question is whether transient Fe-oxo intermediates produced during enzymatic turnover or ROS released by uncoupled catalysis are responsible for direct oxidation of the primary alcohols in solution. However, no decrease in 2m-bt generated, nor its rate of formation was observed in reactions carried out in the presence of catalase, superoxide dismutase, or both (1 mg/mL each); in fact, 2m-bt yields are slightly increased (~20%). This suggests that, if ROS are produced in O₂-dependent 2ma-reactions, they do not freely disassociate from the enzymatic active site. It is therefore more likely that a transient Fe-oxo species is responsible for the oxidation of 1°-alcohols. Potentially, ROS generated during the oxidation of the 1°-alcohols could act as an oxidant [Ox] for the dehydrogenation of benzothiazoline (4) to form the final benzothiazole product (5) shown in Scheme 2.

This uncoupled reactivity with oxygen reminiscent to what is observed for Mm CDO in reactions utilizing L-selenocysteine (sec) as a substrate. Despite modest spectroscopic perturbations from the increased polorizability of the Se-atom, both EPR and MCD spectroscopy indicate that the substrate-bound Mm CDO complex with cys and sec exhibit near equivalent coordination geometry and symmetry around the Fe-site [19, 46]. However, rapid mixing of O₂ with the sec-bound enzyme produces no detectable dioxygenase products. Based on spectroscopically validated computational studies, it was proposed that the subtle electronic differences in the Fe-bound O₂-adduct produced in sec-reactions differ substantially from the native reactions with cys resulting in an ‘off-pathway’ reaction coordinate [47]. The results presented here suggest that the structural constraints and/or differential reactivity of the aromatic 2ma substrate produce a similar effect on CDO and MDO.

The mononuclear non-heme iron site of Mm CDO binds substrate bidentate via neutral amine and thiolate coordination. By contrast, it has been demonstrated in both steady-state kinetic and EPR spectroscopic studies that the homologous 3-mercaptopropionic dioxygenase isolated from the soil bacteria Azotobacter vinelandii (Av MDO) binds substrate via thiolate coordination only [28, 29]. Given the change within the first coordination sphere of the enzyme-substrate complex for both thiol dioxygenase enzymes, it is possible that their reactivity with 2ma could differ. However, as shown in Fig. 4 and Fig. 5 (panel 3), identical benzothiazole products are produced in reactions utilizing Av MDO. Surprisingly, despite the monodentate substrate-coordination of MDO, no reactivity was observed toward benzenethiol (Supplemental Information, Fig. S2).

Regardless, these experiments clearly demonstrate that 2ma is capable of initiating O₂-consumption for both Mm CDO and Av MDO. However, rather than the expected sulfinic acid, a benzothiazole product is generated following oxidation of 1°-alcohols in solution. Vastly decreased activities, specificity, and O₂-coupling efficiencies have been reported for CDO and active site variants utilizing non-native substrates [27]. Extensive x-ray crystallographic studies have also been performed highlighting crucial outer Fe-coordination sphere interactions which facilitate efficient catalysis [48]. In some instances, relatively minor or imperceptible structural perturbations within the active site can produce significant shifts in enzymatic activity [49]. While MDO substrate specificity is significantly more relaxed as compared to CDO, both activity and O₂-coupling efficiency are attenuated in reactions with cysteamine (ca) and cysteine (cys) [29, 50]. However, to our knowledge, no crystallographic studies have been reported for the substrate-bound MDO and thus outer Fe-coordination sphere interactions facilitating efficient catalysis are poorly understood.

Although the dioxygenase product (2abs) is never observed in 2ma-reactions, direct consumption of molecular oxygen is observed with pseudo-first order kinetics dependence on O₂, 2ma, and enzyme concentration. It is therefore reasonable that an oxidized iron-oxo species is produced following substrate-gated O₂-activaiton. Based on a similar reports for a variety of model Fe-oxo compounds [40–42] and Compound I of CytP450 [39], it is reasonable to conclude that the substrate-oxidizing transient intermediate is similarly responsible for the oxidation of the 1°-alcohol in solution in 2ma-reactions. Among similar O₂-dependent oxidation of alcohols, H-atom abstraction is typically the rate limiting chemical step. This frequently results in kinetic isotope effects beyond the classical limit (k_H/k_D ≥ 7) [40–42, 51, 52]. By contrast, in Mm CDO O₂-saturated reactions with 2ma, a kinetic isotope effect of 1.54 ± 0.13 (n = 3) was observed in equivalent reactions where the 20% (v/v) ethanol buffer was prepared using 99.5% ethanol-d₆ [Sigma-Aldrich, 186414]. This suggests that H-atom abstraction from the alcohol is not the rate limiting step. However, saturation kinetics for 2ma is not observed due to its limited aqueous solubility and therefore it is likely that non-chemical events such as substrate-binding limit the rate of this reaction. Finally, while the formal oxidation state of the catalytically relevant iron-oxo intermediate [Fe^III-superoxo or Fe^IV-oxo] remains unclear, it is expected that the structure and/or reactivity of the (ES:O₂) ternary complex is different for aromatic versus aliphatic thiols. Therefore, the transient Fe-oxo intermediate responsible for oxidizing 1°-alcohols may not be relevant to native catalysis.

Concluding Remarks

Based on extensive mechanistic, spectroscopic, and crystallographic characterization a general mechanism for catalysis has been developed for non-heme mononuclear iron enzymes [53–55]. Nearly universally the monoanionic Fe-site contains a 5/6-coordinate ferrous iron with solvent molecules serving as the non-protein ligands. As isolated, the reduced Fe-site is largely unreactive toward molecular oxygen until the substrate and/or cofactor are bound within the active site [55]. The obligate-ordered binding of thiol-substrate to the Fe-site prior to O₂-activation has been reported in previous Mm CDO and Av MDO studies [18, 29]. It has also been noted that both CDO and MDO can catalyze thiol dioxygenase reactions for a number of aliphatic amine/thiol substrates and that optimal activity is observed when these groups are separated by two carbon units [27]. However, the bonds separating the thiol and amine functional groups in the aromatic 2ma substrate are sp² hybridized and thus cannot be rotated. Therefore the 2ma-bound Fe-site geometry of Mm CDO must be very different as compared to the native substrate ES-complex. Presumably, this is also true for Av MDO. Despite this obvious distinction, addition of 2ma to CDO and/or MDO triggers the immediate consumption of oxygen (Supplementary Information, Fig. S2). This suggests that the steps leading up to O₂-activation are largely equivalent for reactions utilizing either aliphatic or aromatic thiols. Therefore, this ‘off-pathway’ reactivity toward 1°-alcohols may be the result of geometric distortions within the substrate/O₂-bound (ES:O₂) ternary complex which prevent thiol-oxidation. Another factor to consider is the differential reactivity of aromatic versus aliphatic thiols. In either case, what can be concluded is that this reactivity is inconsistent with the fixed Fe(II)-oxidation state postulated for ‘substrate-activating’ mechanisms [56]. The unusual reactivity described here further illustrates the remarkable versatility in chemical reactivity exhibited by non-heme iron oxygenase/oxidases.

Highlights.

• CDO and MDO form benzothiazoles in reactions with 2-mercaptoaniline (2ma), 1°-alcohols, and O₂

• Aldehydes are produced by oxidation of 1°-alcohols in enzymatic 2ma-reactions

• Benzothiazoles are formed by condensation of 2ma with aldehydes followed by 2-electron oxidation

Abbreviations

CDO

cysteine dioxygenase

MDO

3-mercaptopropionic acid dioxygenase

ROS

reactive oxygen species

Mm

Mus musculus

Av

Azotobacter vinelandii

2ma

2-mercaptoaniline [2-aminobenzene-1-thiol]

3mpa

3-mercaptopropionic acid

cys

L-cysteine

2abs

2-aminobenzenesulfinic acid

3spa

3-sulfinopropionic acid

csa

cysteine sulfinic acid

HPLC

high performance liquid chromatography

LC-MS

liquid chromatography-mass spectrometry

MRM

multiple reaction monitoring