Tryptophan oxidation in proteins exposed to thiocyanate-derived oxidants

Archives of Biochemistry and Biophysics 564 (2014) 1–11

Contents lists available at ScienceDirect

Archives of Biochemistry and Biophysics journal homepage: www.elsevier.com/locate/yabbi

Tryptophan oxidation in proteins exposed to thiocyanate-derived oxidants Vincent Bonifay a,1, Tessa J. Barrett b,c,1, David I. Pattison b,c, Michael J. Davies b,c, Clare L. Hawkins b,c, Michael T. Ashby a,⇑ a b c

Department of Chemistry and Biochemistry, University of Oklahoma, Norman, OK 73019, USA Heart Research Institute, 7 Eliza St, Newtown, NSW 2042, Australia Sydney Medical School, University of Sydney, Sydney, NSW 2006, Australia

a r t i c l e

i n f o

Article history: Received 17 June 2014 and in revised form 2 August 2014 Available online 27 August 2014 Keywords: Tryptophan Hypothiocyanite Lactoperoxidase Myeloperoxidase Oxidation Biomarker

a b s t r a c t Human defensive peroxidases, including lactoperoxidase (LPO) and myeloperoxidase (MPO), are capable of catalyzing the oxidation of halides (X) by H2O2 to give hypohalous acids (HOX) for the purpose of cellular defense. Substrate selectivity depends upon the relative abundance of the halides, but the pseudohalide thiocyanate (SCN) is a major substrate, and sometimes the exclusive substrate, of all defensive peroxidases in most physiologic fluids. The resulting hypothiocyanous acid (HOSCN) has been implicated in cellular damage via thiol oxidation. While thiols are believed to be the primary target of HOSCN in vivo, Trp residues have also been implicated as targets for HOSCN. However, the mechanism involved in HOSCN-mediated Trp oxidation was not established. Trp residues in proteins appeared to be susceptible to oxidation by HOSCN, whereas free Trp and Trp residues in small peptides were found to be unreactive. We show that HOSCN-induced Trp oxidation is dependent on pH, with oxidation of free Trp, and Trpcontaining peptides observed when the pH is below 2. These conditions mimic those employed previously to precipitate proteins after treatment with HOSCN, which accounts for the discrepancy in the results reported for proteins versus free Trp and small peptides. The reactant in these cases may be thiocyanogen ((SCN)2), which is produced by comproportionation of HOSCN and SCN at low pH. Reaction of thiocyanate-derived oxidants with protein Trp residues at low pH results in the formation of a number of oxidation products, including mono- and di-oxygenated derivatives, which are also formed with other hypohalous acids. Our data suggest that significant modification of Trp by HOSCN in vivo is likely to have limited biological relevance. Ó 2014 Elsevier Inc. All rights reserved.

Introduction Peroxidases play an important role in human innate defense by generating antimicrobial agents by catalyzing the oxidation of halides (X) by H2O2 to give hypohalous acids (HOX).2 Such peroxidases include lactoperoxidase, [1] salivary peroxidase, [2]

⇑ Corresponding author. E-mail address: [email protected] (M.T. Ashby). These authors contributed equally. Abbreviations used: BSA, bovine serum albumin; DTT, dithiothreitol; HOBr, hypobromous acid; HOCl, hypochlorous acid; HOSCN; hypothiocyanous acid; HOX, hypohalous acid; HPLC-FD, High Performance Liquid Chromatography with fluorescence detection; HPAEC-iPAD, High Performance Anion Exchange Chromatograph with a pulsed amperometric detector; HSA, human serum albumin; LDL, low-density lipoprotein; LPO, lactoperoxidase; Mb; myoglobin; MPO, myeloperoxidase; MSA, methanesulfonic acid; NFK, N-formylkynurenine; OCN, cyanate; SCN, thiocyanate; (SCN)2, thiocyanogen; SRM, selective reaction monitoring; TCA, trichloroacetic acid. 1 2

http://dx.doi.org/10.1016/j.abb.2014.08.014 0003-9861/Ó 2014 Elsevier Inc. All rights reserved.

myeloperoxidase, [3] and eosinophil peroxidase [4]. Substrate selectivity of the peroxidases depends upon the redox potential of the enzyme and the bioavailability of the halides; however, the pseudo-halide thiocyanate (SCN) is the major substrate of all of the defensive peroxidases in most physiologic fluids. The resulting hypothiocyanous acid (HOSCN) [5] has been implicated in host cellular damage via thiol oxidation [6–8]. The pKa of HOSCN is 5.3, thus hypothiocyanite (OSCN) is largely unprotonated at physiologic pH. Nonetheless, HOSCN is the chemically active form of OSCN and herein we use ‘‘HOSCN’’ to represent the mixture of reactive species that are derived from OSCN. Although HOSCN has been implicated in the induction of cellular damage, there is only indirect evidence to link this oxidant to the development of disease [9]. Thus, smokers with high SCN levels have greater macrophage foam cell populations and deposits of oxidized low-density lipoprotein (LDL), which are early markers of atherosclerosis, compared to non-smokers [10,11]. Protein carbamylation (homocitrulline formation) mediated

2

V. Bonifay et al. / Archives of Biochemistry and Biophysics 564 (2014) 1–11

by the HOSCN decomposition product cyanate (OCN), has been used to implicate MPO-derived HOSCN in coronary events [12]. However, use of homocitrulline as a biomarker has limitations, particularly in the context of cardiovascular disease, as OCN is elevated under uremic conditions and hence elevated protein carbamylation does not unequivocally implicate HOSCN in tissue damage [13]. Furthermore, there are a number of reports that indicate that SCN may afford protection from oxidative damage [14–17]. The isolation and identification of a HOSCN biomarker is complicated by the relatively low reactivity that this oxidant exhibits for most biological targets [9]. While thiols are believed to be the primary target of HOSCN, it has been recently suggested that Trp residues in proteins may also be favorable targets of oxidation by HOSCN [18,19]. This is significant given that oxidized Trp residues have been detected previously in diseased tissues including, cardiac proteins and skeletal muscle of diabetic rats [20,21] and human atheroma [22,23], suggesting that the products resulting from the modification of Trp residues may have potential for the development of novel biomarkers for assessing the role of HOSCN and related oxidants in disease development. Significantly, Trp residues in proteins were previously found to be susceptible to oxidation by HOSCN, whereas free Trp and Trp residues in small peptides were unreactive [18]. In addition, although evidence was obtained for the formation of both monoand di-oxygenated Trp products, a detailed characterization of the resulting products was not performed. In this study we characterize the Trp oxidation products that are produced from thiocyanate-derived oxidation products and we provide an explanation for the difference in reactivity of Trp in small molecules versus in proteins. Materials and methods Reagents Unless stated otherwise, all chemicals were purchase from Sigma–Aldrich and were ACS certified grade or better. Chromatography mobile phases were all HPLC grade. Nanopure water was filtered in a four-stage Milli-Q system (Millipore). Control of the pH was achieved using 0.1 M sodium phosphate buffer (pH 7.4). Bovine serum albumin (BSA), human serum albumin (HSA) and myoglobin (Mb, from horse heart) were obtained from Sigma–Aldrich. Hydrogen peroxide [30% (v/v) solution; Sigma–Aldrich] was quantified by UV absorbance at 240 nm (e = 39.4 M1 cm1) [24]. Hypothiocyanite was quantified at 376 nm (e = 26.1 M1 cm1) [25]. Instrumentation HPLC with fluorescence detection (HPLC-FD): Analyses were performed using a Shimadzu SCL-10 Avp HPLC system, equipped with a solenoid valve, robotic autosampler SIL 10 Av and a fluorescence detector RIF-10. Chromatography conditions were in accordance with the Agilent method [16]. Briefly, the hydrolyzed samples were automatically derivatized with OPA by programming the robotic autosampler. After derivatization, 2.5 lL of each sample was injected on a Zorbax Eclipse-AAA column, 5 lm, 150  4.6 mm (Agilent), at 40 °C, with detection at k = 338 nm. Mobile phase A was 40 mM NaH2PO4, adjusted to pH 7.8 with NaOH and mobile phase B was acetonitrile/methanol/water (45/45/10 v/v/v). High Performance Anion Exchange Chromatography with pulsed amperometric detection (HPAEC-iPAD): Separations of 25 lL samples were performed at 30 °C with an AminoPac PA10 guard (Dionex, 50  2 mm) and an AminoPac PA10 analytical column (Dionex, 250  2 mm) at a flow rate of 0.25 mL/min. Chromeleon Version 6.70 software (Dionex Benelux) was used for

chromatographic system control, data acquisition and data analysis. The gradient comprised of four eluents: E1 (HPLC grade water), E2 (250 mM NaOH), E3 (1 M sodium acetate), and E4 (100 mM acetic acid). The gradient was as follows: For 2 min initially, 76% E1 and 24% E2, then for 6 min with a curve of 8 the ratio change to 64% E1 and 36% E2, which was maintained for 3 min. With a curve of 8 during 7 min the gradient was changed to 40% E1, 20% E2 and 40% E3, and then to 44% E1, 16% E2 and 40% E3 during the following 3 min (curve 5). For 2 min and a curve of 8 the ratio was changed to 14% E1, 16% E2 and 70% E3, which was then maintained for 22 min. All the strongly retained species were then removed from the column by applying for 2 min 100% of E4. After the acid wash, the column was washed for 2 min with 20% E1 and 80% E2. The column was then equilibrated to starting conditions during 15 min with 76% E1 and 24% E2. The detection waveform with integrated pulsed amperometric detection is adapted from Ding et al. [26]. UV/Vis spectroscopy: Electronic spectra were measured with a HP 8452A diode array spectrophotometer and conventional cuvettes with path lengths of 1–10 cm. For solutions of OSCN less than 1 mM, a WPI 100 cm Liquid Waveguide Capillary Flow Cell (LWCC-3100) fitted with an Ocean Optics USB2000 spectrophotometer and an AIS Model D1000 CE UV light source was used to make measurements. Mass spectrometry: Peptides were analyzed by LC–MS/MS with either (1) Agilent 1290 HPLC system and an Agilent 6538 Q-ToF MS or (2) Thermo Finnigan LCQ Deca XP Max ion trap mass spectrometer coupled to a Thermo Finnigan Surveyor HPLC system (Thermo Electron Corp., Rydalmere, NSW, Australia). The peptides were separated on a Zorbax column [(1) 300 SB C18, 4.5  150 mm, 3 lm; (2) ODS C18, 3  250 mm, 5 lm]. In each case, mobile phase A contains 95/5 v/v of water/acetonitrile with 0.1% formic acid and mobile phase B contains 100% acetonitrile with 0.1% formic acid. With instrument (1) a flow of 0.5 mL/min was used with a gradient of 0–60% of B over 60 min, with data treated with Masshunter, qualitative analysis version B.05.00 with Bioconfirm version B.05.00 (Agilent). With instrument (2) tryptic peptides were separated using a linear gradient from 10% to 30% B over 15 min, from 30% to 80% B over 20 min, followed by a 5 min wash at 100% B and re-equilibration to 10% B over 10 min. Nitrogen, the sheath gas, was held at 80 units, the sweep gas at 10 units, the collision energy was set at 35% with an injection volume of 25 lL. The amino acids from Pronase digestion were analyzed on the LCQ Deca XP Max and separated using 0.1% v/v TFA in H2O as solvent A and 0.1% v/v TFA in 50% v/v methanol as solvent B with a linear gradient from 10% to 60% solvent B over 40 min, followed by isocratic separation with 60% B for 30 min, before returning to 10% B over 5 min and reequilibration for 5 min as described previously [27]. In this case, the sheath gas, was held at 70 units, the sweep gas at 30 units, the collision energy was set at 25% with an injection volume of 50 lL. Procedures Generation of OSCN: LPO was incubated with NaSCN (15 mM) at 20 °C in potassium phosphate buffer (10 mL of 10 mM at pH 7.4). The reaction was initiated by the addition of 10 mL of H2O2 (7.5 mM in the same buffer) to the LPO solution while gently vortexing. Catalase (150 units) was added to remove unreacted H2O2 before filtration by centrifugation (10,000g for 6 min) using Pall Life Sciences Nanosep devices (10 kDa molecular-mass cut-off) to remove the catalase and LPO. Before use, the Nanosep was rinsed once with 300 lL 0.1 M NaOH and twice with 300 lL milliQ water. The formation of OSCN was quantified by electronic spectroscopy using a 100 cm path length by employing blank solutions that comprised the initial concentrations of buffer, LPO and NaSCN and/or by measuring the loss of 5-thio-2-nitrobenzoic acid (TNB)

V. Bonifay et al. / Archives of Biochemistry and Biophysics 564 (2014) 1–11

at 412 nm [18] using an extinction coefficient of 14,150 M1 cm1 [28]. Preparation of HOCl and HOBr: HOCl/OCl was prepared by dilution of a concentrated stock solution of NaOCl (BDH, Poole, Dorset, UK) with the concentration determined by the optical absorbance at 292 nm and pH 11 using a molar extinction coefficient of 350 M1 cm1 [29]. HOBr was prepared by mixing equal volumes of 45 mM KBr, and 40 mM HOCl for 60 min, to ensure the complete conversion of HOCl to HOBr [30]. HOBr was prepared fresh prior to each experiment, and was diluted directly into sodium phosphate buffer (0.1 M, pH 7.4) prior to use. Reaction of proteins with HOSCN: Protein (10 lM or 50 lM in 10 mM phosphate buffer at pH 7.4) was treated with HOSCN (final concentrations 0–1250 lM) for 30 min. Two protocols were used to isolate the protein. Protocol 1: Acid precipitation. The protein was subsequently precipitated using 5% (w/v) trichloroacetic acid (TCA) and pelleted by centrifugation (8000g for 3 min) at 20 °C. The pellets were washed twice with ice-cold acetone. Protocol 2: Microfiltration. The protein solution was filtered through an YM10 microcon centrifugal device (which retains molecules larger than 10 kDa; e.g., BSA and Mb) for 30 min 12,000g. Buffer (50 lL) was added to the sample reservoir prior the recovery at 1000g for 3 min. Protein hydrolysis using methanesulfonic acid (MSA) for amino acid analysis: The recovered HOSCN-treated proteins were hydrolyzed with 150 lL of 4 M methanesulfonic acid containing 0.2% tryptamine at 115 °C for 24 h according to the method of Lee and Drescher [31,32]. After hydrolysis, the samples were partially neutralized with 150 lL of 3.5 N NaOH. The hydrolysates were analyzed by HPLC-FD. Protein digestion using Pronase: Following incubation, proteins were precipitated from solution by the addition of TCA (5% w/v), pelleted by centrifugation (8000g, 5 min, 4 °C) and subsequently washed by the addition of 500 lL of ice-cold acetone. Samples were incubated overnight at 37 °C with Pronase E (Roche Diagnostics, Castle Hill, NSW, Australia) at 20% w/w of the target protein. Digestion was stopped by lowering the temperature to 4 °C, and filtering through Nanosep 10 kDa molecular mass cut-off filters (12,000g, 5 min). The internal standard (L-tryptophan-20 ,40 ,50 ,60 ,70 (indoled5); 25 lM) was added to all samples to quantify the loss of Trp and the formation of Trp oxidation products, assuming near identical ionization efficiencies. Tryptic digestion of protein for mass spectrometry analysis: The HOSCN-treated proteins were reconstituted in approximately 100 lL of 6.0 M urea in a 1.5 mL plastic microfuge tube. The reducing reagents (5 lL of 200 mM DTT in 25 mM ammonium bicarbonate) was added and mixed by gentle vortexing. The mixture was reacted for 1 h at 37 °C. The alkylating reagent (20 lL of 200 mM iodoacetamide in 25 mM ammonium bicarbonate) was added to the samples and then left it for 1 h at room temperature in the dark. To consume any leftover alkylating agent (so the trypsin is not alkylated), 20 lL of the reducing reagent was added. The samples were diluted in 900 lL of ammonium bicarbonate solution and trypsin was added with the approximate ratio of 1:50 trypsin to protein by weight. The digestions proceeded for 16 h at 37 °C.

Statistical analyses Statistical analyses to compare the effect of HOSCN treatment with the untreated control were carried out using one-way ANOVA with Dunnett’s post hoc test. For all of the measurements for which estimated errors are given, n = 6. All statistical analyses were performed using either GraphPad Prism 4 or 5 (GraphPad Software; http://www.graphpad.com), with P < 0.05 (or q > 2.7 for the Dunnett’s post hoc test) taken as significant.

3

Results Oxidation of Trp residues on exposure of proteins to HOSCN requires acidic conditions Four methods were employed to investigate the oxidation of Trp by the HOSCN system: HPLC-FD, HPAEC-iPAD, UV–vis, and LC–MS/ MS. The HOSCN-treated proteins were investigated initially by amino acid analysis using HPLC equipped with a fluorescence detector (HPLC-FD) and by a High Performance Anion Exchange Chromatograph (HPAEC) equipped with a pulsed amperometric detector (HPAEC-iPAD). Both methods of analysis produced essentially the same results for all samples. Myoglobin (Mb) and bovine serum albumin (BSA) (both 10 lM) were treated with various concentrations of HOSCN (0–1250 lM, i.e., equimolar to 125-fold molar excess) for 30 min. Following treatment with HOSCN, the samples were divided in half and one part was processed by precipitation of the protein with TCA while microfiltration to remove lowmolecular mass species was used for processing the other part. The samples were subsequently hydrolyzed with MSA, derivatized with OPA, and subsequently analyzed by either HPAEC-iPAD or HPLC-FD. A dose-dependent loss of Trp was observed on treatment of Mb with HOSCN when the samples were precipitated with TCA (for the purification) before analysis. A similar loss of Trp was observed on separation and detection of Trp with HPLC-FD (Fig. 1A) and HPAECiPAD (Fig. 1B). No evidence was obtained for any loss of other amino acid residues. Analogous results were obtained for BSA (Fig. 2). These data agree well with previously published studies [18]. In contrast, relatively little loss in Trp residues was observed on purification of the protein using the microfiltration protocol prior to digestion or hydrolysis (Figs. 1 and 2). Similarly, a dosedependent decrease in UV absorption (k = 279 nm) of the protein, consistent with Trp loss, was observed on increasing the concentration of HOSCN, when the protein was treated with TCA, but there was no significant loss of Trp when the BSA was isolated by filtration prior to hydrolysis (Fig. 3). When free Trp was treated with HOSCN in phosphate buffer at pH 7.4, consistent with previous observations [18], no oxidation was observed by HPAEC-iPAD and HPLC-FD (Fig. 4). However, with addition of 5% (w/v) TCA following the 30 min of treatment with HOSCN, significant oxidation of free Trp was observed (Fig. 4). Note that Trp was significantly oxidized when treated with equimolar amounts of HOSCN.

Characterization of Trp-derived products formed on HOSCN-treated Mb The oxidation of Trp residues in Mb by HOSCN was investigated by peptide mass fingerprint studies. HOSCN-treated Mb was digested with trypsin and analyzed with LC–MS/MS, giving 100% coverage for both proteins as determined by the Bioconfirm software. For the Mb samples that were isolated by TCA precipitation, the loss of the peptide that contains both Trp residues (GLSDGEWQQVLNVWGK, m/z 1814.9) was observed, together with the formation of two new peptides that corresponded to the mass of the ions [M+64+H]+ and [M+32+H]+ (Fig. 5). Tandem MS/MS experiments confirmed that the new peptides corresponded to oxidation of both Trp residues (data not shown), in agreement with previous studies [18]. The HOSCN dose-dependent decrease of the signal for the peptide GLSDGEWQQVLNVWGK was closely related to the increase of the signal of the peptide GLSDGEW(+32)QQVLNVW(+32)GK (Fig. 5). In contrast, the corresponding samples obtained after isolation of the Mb by filtration, did not exhibit a loss of the m/z 1814.9 fragment (Fig. 5). Further characterization of these Trp-derived oxidation products was performed using Mb (50 lM) exposed to HOSCN

4

V. Bonifay et al. / Archives of Biochemistry and Biophysics 564 (2014) 1–11

(a)

(a)

*

**

**

**

**

**

80 60 40 20

100

1

5

(b)

10 40 50 100 Excess of HOSCN

125

120 100

*

**

**

**

**

80 60 40 20

(b)

**

1

5

10 40 50 100 Excess of HOSCN

125

Fig. 1. Loss of Trp on treatment of Mb with increasing concentrations of HOSCN. HPAEC-iPAD (a) and HPLC-FD (b) analysis of Mb (10 lM) treated with HOSCN (10– 1250 lM) for 30 min, at pH 7.4, before protein precipitation with TCA (5% w/v) and hydrolysis (white) or before protein microfiltration and hydrolysis (stripes). For the measurements made on protein that had been not been precipitated (Protocol 2), a Dunnett’s post hoc test was performed to determine if the measurements with added HOSCN produced a statistically significant change in percent Trp relative to no added HOSCN: ⁄ = 2.7 < q < 3.2 and ⁄⁄ = q > 3.2. Note that if q > 2.7, P < 0.05.

(25–500 lM) for 30 min at 22 °C, with subsequent purification by TCA precipitation prior to enzymatic protein digestion with Pronase and LC–MS/MS analysis. A dose-dependent loss in Trp residues (m/z 205.1), assessed by selective reaction monitoring (SRM) of the specific Trp fragment m/z 188.1, was observed (Fig. S1). Exposure of the protein to decomposed HOSCN, did not lead to a reduction in Mb-associated Trp residues (Fig. S1, bar detonated ‘‘D’’). Similarly, no loss in Mb Trp residues was observed on purification of the protein by filtration rather than TCA precipitation (data not shown). Normalized Trp peak areas obtained for control Mb samples using Trp-20 ,40 ,50 ,60 ,70 (indole-d5), indicated that ca. 50% of the Mb Trp residues are recovered using this Pronase digestion approach. Treatment of Mb with HOSCN and TCA precipitation also resulted in the detection of multiple new products including seven products with m/z 221.1, eight products with m/z 237.1, three products with m/z 320.1, two products with m/z 336.1, and two products with m/z 261.1. In each case, the presence of the oxidation products was dependent on the presence of HOSCN, with no significant product formation observed in the absence of HOSCN, with decomposed HOSCN, in the absence of TCA or with Pronase in the absence of Mb. Fig. 6 shows the formation of new products with m/z 221.1 (labeled a–g) corresponding to the addition of +16 mass units to Trp, consistent with the addition of an oxygen atom, which were not apparent in the control, non-treated Mb. Products a, c and d appear to be produced in a dose-dependent manner, whereas b, e, f and g are seen at similar levels regardless of the amount of oxidant added (Fig. 6). The assignments of products a–g, made on the

**

**

10 40 50 100 Excess of HOSCN

125

60 40 20 1

5

120 100

**

**

**

**

**

10 40 50 100 Excess of HOSCN

125

80 60 40 20 0 1

0

*

80

0

0

% Trp detected by HPAEC-iPAD

% Trp detected by HPLC-FD

100

120

% Trp detected by HPAEC-iPAD

% Trp detected by HPLC-FD

120

5

Fig. 2. Loss of Trp on treatment of BSA with increasing concentrations of HOSCN. HPAEC-iPAD (a) and HPLC-FD (b) analysis of BSA (10 lM) treated with HOSCN (10– 1250 lM) for 30 min, at pH 7.4, before protein precipitation and hydrolysis (white) or before protein microfiltration and hydrolysis (no precipitation) (stripes). See the caption for Fig. 1 for the significance of * and **.

basis of detailed comparison of the MS/MS fragmentation spectra with published data, are collected in Table 1. The major fragmentation ion (m/z 175.1) observed for all the species with m/z 221.1 corresponds to a loss of 46 mass units (Fig. 7); this is attributed to loss of H2O from the protonated carboxyl group, followed by the subsequent loss of CO (Scheme 1). The other major fragmentation ion observed in the spectra of these peaks has m/z 203.1 (loss of 18 mass units, Fig. 7), which is assigned to the loss of H2O (Scheme 1), a common fragment observed with amino acids [33]. A unique fragment ion was identified in the spectra of peak b, with m/z 149.9. This fragment arises due to a loss of 71 mass units from the m/z 221.1 parent. Different proportions of the fragment ions were observed with some products, consistent with the formation of different stereoisomers, in accord with previous studies [27,34]. Evidence was also obtained for the generation of at least eight unique products with a mass of m/z 237.1 on treatment of Mb with HOSCN (Table S1); this is consistent with the addition of two oxygen atoms to the Trp residues (product peaks labeled a–h, Fig. S2). Peaks a, b, and c are not seen in the non-treated protein, and appear to be generated in dose-dependent manner upon addition of HOSCN, whereas products e, f, g and h are also seen in non-treated Mb, though the intensity increases in the presence of HOSCN (Fig. S2). The MS/MS spectra of products a–d are similar with respect to both the fragment ions produced, and their relative intensities (Fig. S3). In all cases, the two major fragment ions detected have m/z 219.1 and m/z 146.2, corresponding to a loss of 18 and 91 mass units from the parent product, respectively. These mass changes are attributed to loss of H2O and fragmentation of the m/z 237.1 product in each case (Scheme S1). The fragmentation pattern, and elution time of peak g is similar to that previously documented for N-formylkynurenine (NFK), with major fragment ions m/z 220.1 (loss of 17 mass units) and m/z 202.1 (loss

5

120

(a) 120

100

*

*

% Area normalized

% UV signal compared to control

V. Bonifay et al. / Archives of Biochemistry and Biophysics 564 (2014) 1–11

80 60 40 20 0

100 80 60 40 20 0

1

5

10 40 50 100 Excess of HOSCN

0

20

40 60 80 100 Excess of HOSCN

120

0

20

40 60 80 100 Excess of HOSCN

120

125

(b)

120

% Area normalized

120 Fig. 3. Percent of UV signal (279 nm) for BSA compared to control (no HOSCN). Analysis of BSA (10 lM) treated with HOSCN (10–1250 lM) for 30 min, at pH 7.4, before protein precipitation and hydrolysis (white) or before protein microfiltration and hydrolysis (no precipitation) (stripes). See the caption for Fig. 1 for the significance of * and **.

100 80 60 40 20 0

% Trp remaining

100 80

Fig. 5. LC–MS of Mb. (a) Percent of the peptide GLSDGEWQQVLNVWGK, m/z 1814.9 compare to the control for Protocol 1 (circles with solid line) and Protocol 2 (triangles with dashed line). (b) Area of [M+64+H]+ peak compared to the corresponding area of the peptide when using 125 fold of HOSCN. The areas are standardized to the area of the peak corresponding of the peptide HGTVVLTALGGILK (64 ? 77).

60 40 20 0

1

5 20 40 Excess of HOSCN

100

Fig. 4. Percent of UV signal (279 nm) for free Trp compared to control (no HOSCN). Analysis of free Trp (2 lM) treated with HOSCN (2–200 lM) for 30 min, at pH 7.4 (stripes) and with addition of TCA after the 30 min (white). Note that there was no statistical difference in the measurement of Trp that was treated with HOSCN without TCA (see the caption for Fig. 1).

of 35 mass units), attributed to loss of NH3 and then H2O (Scheme S1) [27,34]. Products corresponding to peaks e–f, appear to have parent ions with m/z 236.1 (Fig. S3e,f); this may indicate that these identified products are either not Trp + 32 derived, are – enone isomers of the Trp + 32 products, or readily lose H+ from the parent species. Comparative analysis of the full scan spectra obtained from the LC–MS/MS analyses of non-treated, control Mb, and HOSCN-treated Mb revealed the formation of 3 additional products with unique masses (m/z 320.1, m/z 336.1 and m/z 261.1) which were formed consistently in replicate experiments. No evidence was obtained to support the formation of products corresponding to the addition of SCN (+58, m/z 263.1), OCN (+42, m/z 247.1) or OSCN (+77, m/z 279.1) to Trp, which may have been potential products formed by HOSCN, (SCN)2 or other decomposition products. The fragmentation spectra for the products with m/z 320.1, m/z 336.1 and m/z 261.1 are shown in Figs. S4–S6, respectively, with the fragment ions and tentative assignments collected in Table S2. Characterization of Trp-derived products formed on HOSCN-treated albumin Treatment of BSA with HOSCN followed by TCA precipitation also exhibited a dose-dependent oxidation of the two peptides

Fig. 6. LC–MS analysis of Mb oxidation products with m/z 221.1. Mb (50 lM) was treated with HOSCN (25–500 lM) for 30 min at 22 °C, prior to purification by TCA precipitation and enzymatic digestion with Pronase (20% w/w) for 16 h at 37 °C. Samples were then analyzed by LC–MS/MS, with selective monitoring of species with m/z 221.1. Letters a–g represent species that consistently increase with HOSCN treatment. Chromatograms are representative of all n = 3 independent experiments.

containing Trp, AWSVAR (m/z = 688.4) and FWGK (m/z = 536.3) (Fig. 8). For both peptides, only the corresponding [M+16+H]+ fragment was detected in the chromatogram. MS/MS experiments confirmed the presence of oxygen on the Trp of both fragments. As for Mb, the decrease of the signal for the BSA peptide AWSVAR was closely related to the increase of the signal of the peptide AW(+16)SVAR (Fig. 8). No significant oxidation of the Trp residues was observed when the HOSCN-treated BSA was isolated by filtration (Fig. 8).

6

V. Bonifay et al. / Archives of Biochemistry and Biophysics 564 (2014) 1–11

Table 1 MS/MS fragmentation data for Trp and the oxidation products generated on Mb (50 lM) following exposure to HOSCN (500 lM) for 30 min at 22 °C (pH 7.4), with m/z 221.1. Letters correspond to peaks presented in Fig. 6. m/z

Letter

205.1 221.1

Retention time (min)

MS/MS fragment ions

Assignment

34.0

188.1 (loss NH3)

Trp Structures: Scheme 1 trans-3a-hydroxypyrroloindole-2-carboxylic acid (i)

a

9.0

b

9.8

c d e f g

14.1 16.7 22.8 36.6 41.1

201.9 175.0 203.0 175.0 149.9 203.0 175.0 203.0 175.0 203.0 175.0 203.0 175.0 203.0 175.0

(loss of H2O and (loss of H2O) (loss of H2O and (COOH/NH3) (loss of H2O) (loss of H2O and (loss of H2O) (loss of H2O and (loss of H2O) (loss of H2O and (loss of H2O) (loss of H2O and (loss of H2O) (loss of H2O and

Further characterization of the Trp-derived products in the structurally-related human serum albumin (HSA) agree well with the peptide mass mapping studies. Pronase digestion of HSA (50 lM) treated with HOSCN (25–500 lM), and purified by TCA precipitation resulted in a significant, dose-dependent, loss of Trp residues (Fig. S7). Treatment of HSA with greater than a 5-fold molar excess of HOSCN resulted in the generation of a product corresponding to the addition of +16 mass units to Trp (m/z 221.1), corresponding to product ‘‘e’’ in the Mb studies (Fig. 7). With HSA, two product peaks with m/z 261.1 were also observed, though no evidence was obtained for the addition of two oxygen atoms to the Trp (m/z 237.1 products) nor formation of the m/z 320.1 or m/z 336.1 species. Similarly, no additional oxidation products were found on performing the full scan spectra of HOSCN-treated and control Pronase digests of HSA. Characterization of Trp-derived oxidation products with other oxidants In order to establish whether the oxidation products generated on Mb (and BSA/HSA) are generated specifically by SCN-derived oxidants, identical experiments were carried out with HOCl, HOBr or decomposed HOSCN, and the free radical generator 2,20 -azobis(2-amidino-propane) dihydrochloride (AAPH). Mb (50 lM) was treated with HOCl, HOBr and decomposed HOSCN (250 lM) for 30 min at 22 °C (pH 7.4) or for 24 h with 10 mM AAPH at 22 °C (in the presence of O2), prior to precipitation with TCA, enzymatic digestion with Pronase and LC–MS/MS analysis. Treatment of Mb with HOCl, HOBr and AAPH-derived radicals also led to a reduction in protein-associated Trp residues (Fig. 9). Further LC–MS/MS studies were performed to assess the formation of the HOSCN products with m/z 261.1, 320.1, and 336.1 as it has been reported previously that HOCl and HOBr generate both monooxygenated and di-oxygenated Trp derivatives (reviewed [35]). These results are summarized in Table S3. Treatment of Mb with a 5-fold molar excess of HOCl, HOBr, decomposed HOSCN or 10 mM AAPH, resulted in the generation of the m/z 320.1 products (retention time 36.0 and 40.5 min, respectively). However, the m/z 261.1 (retention time 37.0 and 38.5 min, respectively), and m/z 336.1 products were only formed with HOCl or HOBr, with the m/z 261.1 products also seen with decomposed HOSCN. Discussion Hypothiocyanous acid has a pKa of 5.3. Thus, less than 1% of OSCN is protonated at physiological pH. Previous mechanistic

CO) Hydroxy-tryptophan or oxindolyalanine (iv) CO) cis-3a-hydroxypyrroloindole-2-carboxylic acid (i) CO) Hydroxy-tryptophan or oxindolyalanine (iv) CO) C-3 alcohol (ii) CO) Linked to m/z 261.1 product (iii) CO) Linked to m/z 261.1 product (iii) CO)

studies have demonstrated that HOSCN (not OSCN) is the reactive form [5]. As evidenced previously and again in this study, solutions of HOSCN that are acidified below pH 2 are capable of oxidizing Trp in proteins (Table 2). In addition, when the same TCA protocol is employed, we observe that free Trp as well as Trp residues in small peptides are oxidized, in contrast to the previous study performed in the absence of TCA [18]. If HOSCN were the reactive species, the Trp oxidation reaction would be expected to occur at pH 4, where OSCN is fully protonated. However, free Trp and Trp residues in peptides and proteins are not oxidized by HOSCN at pH 4 (results not shown). Therefore, our data are consistent with thiocyanogen [(SCN) 2] rather than HOSCN as the agent responsible for the oxidation of Trp below pH 2. Indeed, (SCN) 2 is known to react with aromatic groups via electrophilic substitution [36]. Similarly, (SCN)2 has been invoked previously as the reactant responsible for the modification of aromatic amino acid side chains in peroxidase/H2O2/SCN systems [37]. However, these studies, while often cited, predated our understanding of the chemistry of (SCN)2 in aqueous medium. (SCN)2 is produced through comproportionation reactions of HOSCN and SCN [5]. Above pH 2, the comproportionation reaction is slow and intermediate (SCN)2 reacts rapidly with HOSCN to give disproportionation products [38]. When the pH is above 2, based upon previous kinetic studies, we would expect the transient concentration of (SCN)2 to be low, and (SCN)2 would be too unstable (for example with respect to its reaction with HOSCN) to participate in Trp oxidation. When comproportionation occurs below pH 2, the product (SCN)2 is relatively stable [39]. Thus, for example, (SCN)2 that is extracted into pH < 2 solutions is sufficiently longlived to react with thiols [40]. Peptide mass-mapping studies provide evidence for the formation of both mono-oxygenated Trp and di-oxygenated Trp species upon treatment of Mb with HOSCN at low pH [18]. LC–MS/MS characterization of the oxidation products carried out in this study, following Pronase digestion of Mb to its constitutive amino acids, support the peptide mass mapping data, with evidence for the formation of multiple products with m/z 221.1 (Trp + 16) and m/z 237.1 (Trp + 32). Two of the m/z 221.1 product peaks were assigned to trans and cis isomers of 3a-hydroxypyrroloindole2-carboxylic acid (peaks a and c, respectively, in Fig. 6) and one to the C-3 alcohol species (peak e, in Fig. 6). It is hypothesized that isoforms of both hydroxy-tryptophan and/or oxindolyalanine are also formed, though these products could not be conclusively identified due to the similar nature of their fragmentation patterns (Fig. S2, peaks b and d). Although a recent study has demonstrated that two of the Trp + 16 isoforms (5-hydroxytryptohan and

V. Bonifay et al. / Archives of Biochemistry and Biophysics 564 (2014) 1–11

7

Fig. 7. Fragmentation spectra of HOSCN-mediated Mb oxidation products with m/z 221.1. Mb (50 lM) was treated with HOSCN (25–500 lM) for 30 min at 22 °C, prior to purification by TCA precipitation and enzymatic digestion with Pronase (20% w/w) for 16 h at 37 °C. Samples were then analyzed by LC–MS/MS, with selective monitoring of species with m/z 221.1 (collision energy, 25%). Letters a–g represent individual species that consistently increase with HOSCN treatment and correspond to the peaks labeled in Fig. 6. Spectra are representative of n = 3 independent experiments.

oxindolyalanine) can be distinguished based on their fragmentation characteristics in both ESI and MALDI tandem MS [41], it was not possible to distinguish these species under the conditions

used in the present study. The remaining two peaks observed with an apparent mass of m/z 221.1 (peaks f and g, in Fig. 6) may be derived from m/z 261.1 product species. Multiple products with

8

V. Bonifay et al. / Archives of Biochemistry and Biophysics 564 (2014) 1–11

Scheme 1. Proposed structures and fragmentation patterns for Trp oxidation products with m/z 221.1.

V. Bonifay et al. / Archives of Biochemistry and Biophysics 564 (2014) 1–11

% Area normalized

(a)

120 100 80 60 40 20 0

% Area normalized

(b)

0

20

40 60 80 100 Excess of HOSCN

120

0

20

40 60 80 100 Excess of HOSCN

120

120 100 80 60 40 20 0

Normalised Trp peak area

Fig. 8. LC–MS of BSA. (a) Percent of the peptide AWSVAR m/z = 688.4 compare to the control for Protocol 1 (circles with solid line) and Protocol 2 (triangles with dashed line). (b) Area of the [M+16+H]+ peak compared to the corresponding area of the peptide when using 125 fold of HOSCN. The areas are standardized to the area of the peak corresponding of the peptide HPEYAVSVLLR (361 ? 371).

8 6 4

*

2

*

*

O

SC

N

PH -H D

A A

H O C l H O Br

H

O

SC

N

PB

0

Oxidant Fig. 9. Trp oxidation of Mb by various oxidants. Treatment of Mb (50 lM) with HOSCN, HOCl, HOBr and decomposed HOSCN (250 lM) for 30 min at 22 °C or 10 mM AAPH for 24 h, followed by purification by TCA (5% w/v) precipitation and Pronase digestion (20% w/w) for 16 h at 37 °C. Bar ‘‘PB’’ refers to protein treated with phosphate buffer. Results are expressed as the ratio of Trp area peak (m/z 205.1) to the area of the internal standard Trp-indole-d5 peak (m/z 210.1). ⁄ Represents a significant decrease (P < 0.05) in hhMb [Trp area]/[Trp-indole-d5 area] between oxidant treated and control phosphate-buffer (100 mM, pH 7.4) treated hhMb by 1-way ANOVA with Dunnett’s post hoc testing. Values are means ± S.E.M. (n P 3).

m/z 237.1 were also observed, consistent with the addition of two oxygen atoms to Trp. However, due to the high number of species generated and their similar characteristics, it is difficult to unequivocally assign a structure to each peak. Based on studies with 1O2, it is likely that isomers of 2a,3a-hydroxypyrroloindole2-carboxylic acid and dioxindolyalanine are generated by HOSCN under acidic conditions [42].

9

Evidence was also obtained for the production of three previously unidentified species (m/z 320.1, 336.1, and 261.1) that are generated upon treatment of proteins with HOSCN at low pH. Although characterization studies were inconclusive, it is proposed that the m/z 320.1 and 336.1 species are related, and are probably Trp-associated, given that their fragmentation spectra contain many fragments common to the oxygenated Trp products (see Figs. S4 and S5 for fragmentation patterns). The m/z 261.1 species identified from oxidation of Mb and HSA may also be Trp-related. This tentative assignment is based on the similar fragmentation pattern and co-elution of the two m/z 261.1 peaks (peaks a and b in Fig. S6), with peaks corresponding to m/z 221.1 (peaks f and g in Fig. 6). The data suggest that the m/z 261.1 species readily dissociate to give an oxygenated-Trp product, and thus may represent a mono-oxygenated Trp product, with an additional +40 mass units. These products were also observed with the other hypohalous acids, and the m/z 320.1 and 261.1 products were also observed in experiments performed with decomposed HOSCN. The formation of the oxygenated Trp species is proposed to occur via the formation of one (or more) initial, but undetected, indole-SCN intermediates (Scheme 2). Addition of SCN+ to the double bond on the indole group of Trp could occur at either the C2 or C3 position, thereby facilitating the formation of the various oxygenated isoforms of Trp by multiple pathways, as detailed in Scheme 2. The addition of SCN+ to the C2 position may allow for the production of the cyclized 3a-hydroxypyrroloindole-2-carboxylic acid oxidation product (Scheme 2, product 2), as SCN is a relatively good leaving group, which would assist in the production of this species. Furthermore, the ability of SCN (or X) to add at either of these sites may result in the generation of the other oxidation products, such as in the case of the m/z 261.1 and 320.1 species. A reduced number of Trp oxidation species were formed on HSA compared to Mb following HOSCN exposure at low pH. This may be attributed to a number of factors including differing protein structures (i.e. altered accessibility), and the presence of a heme group in Mb that may facilitate Trp oxidation [18]. With HSA, a species analogous to the m/z 261.1 species generated on Mb was observed, though no evidence was obtained for the formation of the Mb-derived m/z 320.1 and 336.1 species. These data with HOSCN at low pH are consistent with previous studies with HSA exposed to HOCl, HOBr or the MPO system, where evidence has been presented for the incorporation of oxygen at Trp238 [43]. Taken together, these data do not support the use of oxidized Trp residues as a biomarker for HOSCN, as the products observed are not unique to this oxidant, and are only formed to a significant extent under the low pH conditions achieved during TCA precipitation and protein purification. Carbamylated Lys residues (homocitrulline) remain the most well-defined marker for SCN-derived oxidants in vivo, though there are potential confounding issues, particularly with uremic patients [12]. HOSCN can be used as a chemical precursor to oxidize free Trp and protein Trp residues, however, the active chemical agent is most likely (SCN)2. Acidic conditions can be achieved in vivo, for example, the pH of phagolysosomes and enamel-biofilm interfaces during cariogenesis falls to approximately 4, though in general physiologic fluids have high buffering capacities near pH 7. An exception might be the gastric system, where low pH conditions occur routinely and SCN acts as a substrate for gastric peroxidase [44]. However, Trp is known to spontaneously undergo autoxidation at very low pH [45], thus the contribution of HOSCN to Trp oxidation by HOSCN in biological systems is unclear.

10

V. Bonifay et al. / Archives of Biochemistry and Biophysics 564 (2014) 1–11

Table 2 Percent of loss of Trp in proteins with 50-fold excess HOSCN. The results shown are an average of those measured by HPLC-FD and HPAEC-iPAD. The third entry demonstrates that acid precipitation in the absence of HOSCN does not cause significant oxidation of Trp. % of Trp oxidized

Buffer + acid precipitation HOSCN + acid precipitation HOSCN + filtration + precipitation after solvation in buffer HOSCN + filtration

BSA

Mb

0 72 ± 2 8 ± 0.3 8 ± 0.4

0 80 ± 3 14 ± 0.7 12 ± 0.5

Scheme 2. Proposed mechanisms for the reaction of SCN-derived oxidants with Trp residues. R = CH2CH(NH2)COOH, +16 and +32 refers to species which are modified by the labeled value relative to parent Trp (m/z 205.1), with m/z 221.1 and 237.1, respectively.

V. Bonifay et al. / Archives of Biochemistry and Biophysics 564 (2014) 1–11

Conclusions Oxidation of Trp by HOSCN results in the formation of a number of products, but this reaction only occurs to a significant extent at pH values below pH 2, and possibly via the formation of (SCN)2 (a species that is unstable a neutral pH). The Trp-derived products formed in these reactions are not unique to HOSCN, with identical products detected with HOCl and HOBr. This suggests that Trpderived oxidation products may have limited utility as biomarkers for HOSCN under normal biological conditions. Acknowledgments This research was supported by the National Science Foundation (CHE-0911328), the National Health and Medical Research Council of Australia (570829), and Australian Research Council (FT120100682, CE0561607, DP0988311). Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.abb.2014.08.014. References [1] K.M. Pruitt, J.O. Tenovuo, (Eds.), Immunology Series, Vol. 27: The Lactoperoxidase System: Chemistry and Biological Significance, Marcel Dekker, Inc., New York, 1985. [2] M.T. Ashby, J. Dent. Res. 87 (2008) 900–914. [3] P.G. Furtmuller, U. Burner, C. Obinger, Biochemistry 37 (1998) 17923–17930. [4] A. Slungaard, J.R. Mahoney Jr., J. Biol. Chem. 266 (1991) 4903–4910. [5] M.T. Ashby, Adv. Inorg. Chem. 64 (2012) 263–303. [6] T.M. Aune, E.L. Thomas, Biochemistry 17 (1978) 1005–1010. [7] M.J. Davies, Encycl. Radicals Chem., Biol. Mater. 3 (2012) 1425–1457. [8] O. Skaff, D.I. Pattison, M.J. Davies, Biochem. J. 422 (2009) 111–117. [9] T.J. Barrett, C.L. Hawkins, Chem. Res. Toxicol. 25 (2012) 263–273. [10] T.P. Botti, H. Amin, L. Hiltscher, R.W. Wissler, Atherosclerosis 124 (1996) 191– 202. [11] C.E.O. Scanlon, B. Berger, G. Malcom, R.W. Wissler, Atherosclerosis 121 (1996) 23–33. [12] Z. Wang, S.J. Nicholls, E.R. Rodriguez, O. Kummu, S. Hoerkkoe, J. Barnard, W.F. Reynolds, E.J. Topol, J.A. DiDonato, S.L. Hazen, Nat. Med. (N. Y., NY, U. S.) 13 (2007) 1176–1184. [13] H.E. Roxborough, I.S. Young, Med. Hypotheses 45 (1995) 125–128.

11

[14] M.T. Ashby, A.C. Carlson, M.J. Scott, J. Am. Chem. Soc. 126 (2004) 15976– 15977. [15] P. Nagy, J.L. Beal, M.T. Ashby, Chem. Res. Toxicol. 19 (2006) 587–593. [16] R.D.R., John W. Henderson, Brain A. Bidlingmeyer, Cliff Woodward, Agilent Technologies, 2000. [17] J.D. Chandler, B.J. Day, Biochem. Pharmacol. 84 (2012) 1381–1387. [18] C.L. Hawkins, D.I. Pattison, N.R. Stanley, M.J. Davies, Biochem. J. 416 (2008) 441–452. [19] K.A. Hadfield, D.I. Pattison, B.E. Brown, L. Hou, K.-A. Rye, M.J. Davies, C.L. Hawkins, Biochem. J. 449 (2013) 531–542. [20] S.W. Taylor, E. Fahy, J. Murray, R.A. Capaldi, S.S. Ghosh, J. Biol. Chem. 278 (2003) 19587–19590. [21] M. Hamblin, D.B. Friedman, S. Hill, R.M. Caprioli, H.M. Smith, M.F. Hill, J. Mol. Cell. Cardiol. 42 (2007) 884–895. [22] D.Q. Peng, G. Brubaker, Z. Wu, L. Zheng, B. Willard, M. Kinter, S.L. Hazen, J.D. Smith, Arterioscler. Thromb. Vasc. Biol. 28 (2008) 2063–2070. [23] Y. Huang, J.A. DiDonato, B.S. Levison, D. Schmitt, L. Li, Y. Wu, J. Buffa, T. Kim, G.S. Gerstenecker, X. Gu, C.S. Kadiyala, Z. Wang, M.K. Culley, J.E. Hazen, A.J. Didonato, X. Fu, S.Z. Berisha, D. Peng, T.T. Nguyen, S. Liang, C.C. Chuang, L. Cho, E.F. Plow, P.L. Fox, V. Gogonea, W.H. Tang, J.S. Parks, E.A. Fisher, J.D. Smith, S.L. Hazen, Nat. Med. 20 (2014) 193–203. [24] D.P. Nelson, L.A. Kiesow, Anal. Biochem. 49 (1972) 474–478. [25] P. Nagy, S.S. Alguindigue, M.T. Ashby, Biochemistry 45 (2006) 12610–12616. [26] Y. Ding, H. Yu, S. Mou, J. Chromatogr. A 982 (2002) 237–244. [27] M. Gracanin, C.L. Hawkins, D.I. Pattison, M.J. Davies, Free Radic. Biol. Med. 47 (2009) 92–102. [28] P. Eyer, F. Worek, D. Kiderlen, G. Sinko, A. Stuglin, V. Simeon-Rudolf, E. Reiner, Anal. Biochem. 312 (2003) 224–227. [29] J.C. Morris, J. Phys. Chem. 70 (1966) 3798–3805. [30] C.L. Hawkins, M.J. Davies, Free Radic. Biol. Med. 39 (2005) 900–912. [31] D.G. Drescher, K.S. Lee, Anal. Biochem. 84 (1978) 559–569. [32] K.S. Lee, D.G. Drescher, Int. J. Biochem. 9 (1978) 457–467. [33] I.A. Papayannopoulos, Mass Spectrom. Rev. 14 (1995) 49–73. [34] G.E. Ronsein, M.C. de Oliveira, M.H. de Medeiros, P. Di Mascio, J. Am. Soc. Mass Spectrom. 20 (2009) 188–197. [35] M.J. Davies, C.L. Hawkins, D.I. Pattison, M.D. Rees, Antioxid. Redox Signal. 10 (2008) 1199–1234. [36] J.L. Wood, Org. React. (N. Y.) (1946) 240–266. [37] T.M. Aune, E.L. Thomas, M. Morrison, Biochemistry 16 (1977) 4611–4615. [38] J. Kalmar, K.L. Woldegiorgis, B. Biri, M.T. Ashby, J. Am. Chem. Soc. 133 (2011) 19911–19921. [39] P. Nagy, K. Lemma, M.T. Ashby, Inorg. Chem. 46 (2007) 285–292. [40] M.T. Ashby, H. Aneetha, J. Am. Chem. Soc. 126 (2004) 10216–10217. [41] T. Todorovski, M. Fedorova, R. Hoffmann, J. Mass Spectrom. 46 (2011) 1030– 1038. [42] G.E. Ronsein, M.C. Oliveira, S. Miyamoto, M.H. Medeiros, P. Di Mascio, Chem. Res. Toxicol. 21 (2008) 1271–1283. [43] P. Salavej, H. Spalteholz, J. Arnhold, Free Radic. Biol. Med. 40 (2006) 516–525. [44] D. Das, P.K. De, R.K. Banerjee, Biochem. J. 305 (1995) 59–64. [45] M. Stewart, C.H. Nicholls, Aust. J. Chem. 25 (1972) 2139–2144.