Author’s Accepted Manuscript Selenocysteine oxidation in glutathione peroxidase catalysis: an MS-supported quantum mechanics study Laura Orian, Pierluigi Mauri, Antonella Roveri, Stefano Toppo, Louise Benazzi, Valentina BoselloTravain, Antonella De Palma, Matilde Maiorino, Giovanni Miotto, Mattia Zaccarin, Antonino Polimeno, Leopold Flohé, Fulvio Ursini

PII: DOI: Reference:

www.elsevier.com

S0891-5849(15)00279-8 http://dx.doi.org/10.1016/j.freeradbiomed.2015.06.011 FRB12471

To appear in: Free Radical Biology and Medicine Received date: 1 April 2015 Revised date: 18 May 2015 Accepted date: 9 June 2015 Cite this article as: Laura Orian, Pierluigi Mauri, Antonella Roveri, Stefano Toppo, Louise Benazzi, Valentina Bosello-Travain, Antonella De Palma, Matilde Maiorino, Giovanni Miotto, Mattia Zaccarin, Antonino Polimeno, Leopold Flohé and Fulvio Ursini, Selenocysteine oxidation in glutathione peroxidase catalysis: an MS-supported quantum mechanics study, Free Radical Biology and Medicine, http://dx.doi.org/10.1016/j.freeradbiomed.2015.06.011 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting galley proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Author’s Accepted Manuscript Selenocysteine oxidation in glutathione peroxidase catalysis: an MS-supported quantum mechanics study Laura Orian, Pierluigi Mauri, Antonella Roveri, Stefano Toppo, Louise Benazzi, Valentina BoselloTravain, Antonella De Palma, Matilde Maiorino, Giovanni Miotto, Mattia Zaccarin, Antonino Polimeno, Leopold Flohé, Fulvio Ursini

PII: DOI: Reference:

www.elsevier.com

S0891-5849(15)00279-8 http://dx.doi.org/10.1016/j.freeradbiomed.2015.06.011 FRB12471

To appear in: Free Radical Biology and Medicine Received date: 1 April 2015 Revised date: 18 May 2015 Accepted date: 9 June 2015 Cite this article as: Laura Orian, Pierluigi Mauri, Antonella Roveri, Stefano Toppo, Louise Benazzi, Valentina Bosello-Travain, Antonella De Palma, Matilde Maiorino, Giovanni Miotto, Mattia Zaccarin, Antonino Polimeno, Leopold Flohé and Fulvio Ursini, Selenocysteine oxidation in glutathione peroxidase catalysis: an MS-supported quantum mechanics study, Free Radical Biology and Medicine, http://dx.doi.org/10.1016/j.freeradbiomed.2015.06.011 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting galley proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Selenocysteine oxidation in glutathione peroxidase catalysis: an MS-supported quantum mechanics study Laura Orian*1, Pierluigi Mauri*2, Antonella Roveri*3, Stefano Toppo*3, Louise Benazzi2, Valentina Bosello-Travain3, Antonella De Palma2, Matilde Maiorino3, Giovanni Miotto3, Mattia Zaccarin3, Antonino Polimeno1, Leopold Flohé§3 and Fulvio Ursini3 1

Department of Chemistry, University of Padova, Italy Institute for Biomedical Technologies, National Research Council, Milano, Italy. 3 Department of Molecular Medicine, University of Padova, Italy § Corresponding Author: Leopold Flohé, Distinguished Visiting Professor at the Department of Molecular Medicine, University of Padova, Italy. Viale G. Colombo 3, I-35121, Padova, Italy e-mail: [email protected] 2

*These authors equally contributed to this study

Abstract: Glutathione peroxidases (GPxs) are enzymes working with either selenium or sulfur catalysis. They adopted diverse functions ranging from detoxification of H2O2 to redox signaling and differentiation. The relative stability of the selenoenzymes, however, remained enigmatic in view of the postulated involvement of a highly unstable selenenic acid form during catalysis. Nevertheless, density functional theory calculations obtained with a representative active site model verify the mechanistic concept of GPx catalysis and underscore its efficiency. However, they also allow that the selenenic acid, in the absence of the reducing substrate, reacts with a nitrogen in the active site. MS/MS analysis of oxidized rat GPx4 complies with the predicted structure, an 8membered ring, in which selenium is bound as selenenylamide to the protein backbone. The intermediate can be re-integrated into the canonical GPx cycle by glutathione, whereas, under denaturing conditions, its selenium moiety undergoes β-cleavage with formation of a dehydroalanine residue. The selenenylamide bypass prevents destruction of the redox center due to overoxidation of the selenium or its elimination and likely allows fine-tuning of GPx activity or alternate substrate reactions for regulatory purposes. Keywords: Glutathione peroxidases; selenium catalysis; selenenylamide formation; DFT calculation; MS/MS analysis of catalytic intermediates.

Abbreviations: CID, collision-induced dissociation; CysGPx(s), glutathione peroxidase(s) containing a peroxidatic cysteine; 1-CysGPx(s), glutathione peroxidase(s) lacking a resolving 1

cysteine; 2-CysGPx(s), glutathione peroxidase(s) containing a peroxidatic cysteine and a resolving cysteine; CP, catalytic Cys, CR, resolving Cys; CS, charge-separated species; DFT, density functional theory; Dha, dehydroalanine; ESI, electrospray ionization; GPx(s), glutathione peroxidase(s); GPx1, ‘classical’ glutathione peroxidase (E.C. 1.11.1. 9); GPx4, phospholipid hydroperoxide glutathione peroxidase (E.C. 1.11.1.12); GPx4U46C, mutated form of rat GPx4 containing the peroxidatic Sec replaced with Cys; GSH, glutathione; HPLC, high performance liquid chromatography; LC-ESI MS/MS, liquid chromatography-electrospray ionization-tandem mass spectrometry; LC-MS, liquid chromatography–mass spectrometry; LC-MS/MS, liquid chromatography-tandem mass spectrometry; MS, mass spectrometry; MS1, primary mass spectrum; MS/MS, tandem mass spectrometry; NF-κB, nuclear factor kappa-light-chain-enhancer of activated B cells; PC-OOH, phosphatidyl choline hydoperoxide; Q-TOF, quadrupole time of flight; RMSD, root mean square deviation; RrGPx4, rat GPx4 (UniProt ID P36970-2); Sec, selenocysteine; SecGPx, glutathione peroxidases containing a peroxidatic selenocysteine; TS, transition state; U, selenocysteine in one letter code.

Introduction

The first GPx was discovered in 1957 as an enzyme protecting hemoglobin from oxidative denaturation [1]. In the meantime, this enzyme (now GPx1) and homologs either working with selenium (SecGPx) or sulfur catalysis (CysGPx) [2] have been shown to fulfill metabolic roles far beyond the name-giving activity of the ‘classical’ glutathione peroxidase (GPx1), i. e. reduction of H2O2 by glutathione (GSH) (compiled in [3]). Mammalian GPx1 also dampens insulin signaling [4] and NF-κB activation [5], mitochondrial GPx4 forms an essential sperm constituent [6-8], nuclear GPx4 contributes to chromatin compaction [9] and cytosolic GPx4 regulates NF-κB activation [10], 12/15-lipoxygenase-driven apoptosis [11] and ferroptosis [12, 13], and increases hepatitis C virus infectivity [14]. GPx7, and likely GPx8, contribute to protein folding [15]. GPx8 also dampens insulin and fibroblast growth factor signaling [16]. Others prevent carcinogenesis but support tumor growth [17], and yeast homologs activate adaptive responses [18]. GPx catalysis is commonly interpreted as a sequence of independent bimolecular reactions, whereby the ground state enzyme (E), with its active site selenocysteine (Sec) residue, is oxidized by a hydroperoxide to an intermediate F and then stepwise reduced by GSH via a semi-reduced intermediate G (Fig. 1a). G has been shown to be a Se-glutathionylated enzyme in pig GPx4 [19], while the chemical nature of F has been debated since the discovery of selenium as the catalytic moiety of GPx1 [20-22]. The most common hypothesis postulates an oxidation of Sec to a selenenic acid [17]. Selenenic acids, however, are extremely unstable. A chemical synthesis was achieved only once, its Se-OH being hidden in a clathrate [23]. It, therefore, cannot surprise that the postulated selenenic acid intermediate has so far resisted any chemical verification. Instead, 2

seleninic acid formation in crystallographic [24] or x-ray photoelectron spectroscopic studies [25], β-cleavage of Sec [26-28] or its unspecific crosslinking via selena-disulfide bridges [19] were detected. The intriguing question, therefore, is how these enzymes can survive continuous catalytic cycling despite the notorious instability of the postulated selenenic acid intermediate. A second unresolved problem is the extreme velocity of the first catalytic step of both, the Sec- and the Cys GPxs [29]. An often-read explanation is the dissociation of the catalytic Sec or Cys, respectively, enforced by neighboring residues in the active site. However, a low pKa of the Sec or Cys, respectively, is certainly a prerequisite for their reaction with a hydroperoxide, but by itself does not explain the catalytic efficiency of the enzymes. The bimolecular rate constants of fully dissociated low molecular thiols with H2O2 hardly exceed 30 M-1s-1 [30] and a quantum mechanics calculation for the homologous reaction of selenocysteine yielded a k of 35.4 M-1s-1 [31]. In sharp contrast, the corresponding rate constants k+1 reach 106 M-1s-1 in CysGPxs and 5 x 107 M-1s-1 in SecGPxs (compiled in [29]). The primary goal of this investigation was to unravel these remaining enigmas of GPx catalysis; the outcome, inter alia, is a plausible mechanism and the discovery of an unusual selenenylamide formation, which prevents β-cleavage of the active site Sec and provides a basis for fine-tuning of GPx activity and alternate substrate reactions.

Materials and Methods

Enzymes Bovine GPx1 (from bovine red blood cells) was from Sigma. Two batches were used: the batch G6137 where the two N-terminal residues were missing and the N-terminus was blocked, apparently by a carbamylation, and the batch 49753 containing different N-terminal truncations (Δ 13, Δ 11, Δ 12 and Δ 10). GPx4 was purified from rat testis cytosol as previously described [32]. Recombinant rat GPx4 U46C was expressed in E. coli and purified. The expression vector containing rat GPx4 (RrGPx4, UniProt ID P36970-2) was prepared by conventional cloning of the PCR product obtained from cDNA of rat liver with the following primers: fw = AATACATATGTGTGCATCCCGCGATGAT; rev = ATTAAAGCTTCTAGAGATAGCACGGCAGGTCCT. After digestion, the PCR product was cloned in the NdeI/HindIII multicloning site of the pT7-7 expression vector (Addgene, Cambridge MA, USA). The correctness of the construct was verified by sequencing. The mutant sequence RrGPx4 Sec46→Cys (GPx4U46C) was generated using the wild type expression plasmid as template and the QuikChange site-directed mutagenesis kit 3

(Stratagene, Cedar Creek, TX, USA). The primers used for amplification were: fw = CAACGTGGCCTCGCAATGTGGCAAAACCGACGTAAAC; rev = GTTTACGTCGGTTTTGCCACATTGCGAGGCCACGTTG. The correctness of the constructs was verified by sequencing. For production of recombinant RrGPx4U46C, competent BL21 (DE3) pLysS E. coli cells, allowing efficient expression of genes under the control of a T7 promoter, were transformed with the vector pT7-7

GPx4U46C.

The

expression

was

induced

by

adding

1

mM

isopropyl-β-D-

thiogalactopyranoside and after 3 h cells were harvested by centrifugation at 5,000 x g for 30 min. Cell pellets were either immediately lysed or kept at -80 °C. The pellet from 0.5 L of culture was lysed under slow agitation at 4 °C in 50 ml of lysis buffer [0.1 M Tris–HCl, pH 7.4, 0.05 M KCl, 5 % (v/v) Nonidet, 0.1 % (v/v) Triton X-100, 100 mM 2-mercaptoethanol] and protease inhibitors [complete protease inhibitor cocktail tablets (1 tablet/50 ml), Roche Diagnostics GmbH, Mannheim, Germany] for 20 minutes. The homogenate was centrifuged at 100,000 x g for 60 min at 4 °C and the supernatant immediately used for enzyme purification. The enzyme was purified to homogeneity by the same procedure used for purification of native enzyme [32]. Proteins were quantified by the Bradford assay using BSA as standard. Homogeneity of enzyme preparations were evaluated by SDS PAGE on T = 14 % gel.

GPx activity GPx4 activity was measured by the conventional test [33], using either H2O2 or phosphatidyl choline hydoperoxide (PC-OOH) and GSH as substrates. Under the standard conditions (30 μM PC-OOH and 3 mM GSH), the specific activities of different preparations of GPx4 used in this study were 135+/-11 and 0.45 +/- 0.006 μmoles/ min x mg protein for native enzyme and the U46C mutein, respectively. GPx1 activity was measured as described in [34].

LC-MS of reduced and oxidized intact proteins Since the reduced form of GPx1 and GPx4, also in the absence of the hydroperoxide substrate, spontaneously shifts to the oxidized form in few minutes, enzyme reduction, removal of the reducing substrate and oxidation by hydroperoxide were carried out using an optimized on-line procedure. This procedure permitted a fast transfer of the sample from the pre-column, where the redox reaction takes place, to the analytical column. Moreover, the chromatographic separation of the species thus achieved obviates the difficulty to accurately discriminate between two species differing by just 2 a.m.u. A 40 nL C8 nano-precolumn coupled to a C8 160 nL analytic nanocolumn was used, linked to the Q-TOF ESI - source (6520 Accurate-Mass Q-TOF LC-MS/MS 4

equipped with high performance liquid chromatography (HPLC) - Chip Cube MS Interface; Agilent Technologies, Santa Clara, CA, United States). A switching valve between the pre-column and the analytic column allowed an accurate timing of sample treatment and the fast and complete removal of the reagents before HPLC analysis. The amount of proteins processed ranged from 150 to 300 fmoles in 0.1 M Tris-HCl, 1 mM NaEDTA, 5 mM β-mercaptoethanol, pH 8.3. Reduction was carried out by flushing reduction buffer (K2HPO4/KH2PO4 50 mM , pH7, 20 mM DTT, 20 % glycerol and 5 % acetonitrile) for 10 minutes at 1 Pl/min into the pre-column, equilibrated with at least 50 volumes of the same buffer. The efficiency of the reduction was validated by the measured mass of the eluted protein. Oxidation was accomplished by flushing the immobilized sample at 1 Pl/min for 3 min with oxidizing buffer (K2HPO4/KH2PO4 50 mM, pH 7, 10 PM H2O2, 20 % glycerol and 5 % acetonitrile). Samples were then flushed with 4 Pl of an aqueous buffer containing 0.3 % TFA, 20 % glycerol and 25 % acetonitrile, switched to the 160 nL analysis column and resolved at 0.3 Pl/min with a gradient. Buffer A was 0.1 % formic acid in water and Buffer B acetonitrile : methanol 90 : 10, 0.1 % formic acid. The percentage of B increased from 28 % to 44 % in 10 min, then from 44 % to 70 % B in 5 min. Re-equilibration was performed by 28 % B for 10 min. Eluted proteins were ionized at 1.7 kV, fragmentor set at 350 V, source gas temperature was 325 C° and gas flow rate 4.8 L min-1. Acquisition parameters were set at MS scan rate of 1 spectra sec-1 in the range between 130 and 1.700 m/z in high resolution mode (R = 20.000). Data were acquired in profile mode and analyzed with MassHunter Workstation Software Qualitative Analysis rel. B06 (Agilent Technologies, Santa Clara, CA, USA). Deconvolution of MS signals from whole protein was performed with “pMod”, an improved version of maximum entropy algorithm, in a m/z range encompassing at least 7 differently charged clusters. Baseline subtraction factor was set to its minimum value, relative height peak filters to 1 % of highest peak and significance filter to a value greater or equal to 20,00. Target mass range used for deconvolution was 10,000 to 40,000 Daltons. All other parameters were left at their default values.

LC-MS/MS analysis of tryptic peptides For these experiments, to the enzymes, stored at -20 C° in 0.1 M Tris-HCl pH 8.3, 1 mM Na-EDTA, 5 mM β-mercaptoethanol, 5 mM fresh β-mercaptoethanol was added to ensure exhaustive reduction. The efficiency of the reduction was tested by HPLC-MS analysis of the intact protein. Notably, also this procedure, as the on-line procedure, fails to fully reduce one disulfide between β-mercaptoethanol and Cys-35 in GPx1 and GPx4. Moreover, a proportion of the enzyme (10-20 %) was already oxidized to the -2 a.m.u. form when analyzed. Half of this sample, at a final 5

volume of 1.5 ml, was exposed to 10 μM H2O2 for 5 minutes at room temperature (oxidized enzyme) and the remaining half immediately used as control (reduced enzyme). Following removal of the oxidant by a micro BioSpin 6 column, proteins were precipitated with 6 % (w/v) trichloroacetic acid ,washed with cold ethanol and digested with trypsin. Trypsin was added to 3 μg of protein at an enzyme/substrate ratio of 1:50 (w/w) in 100 mM ammonium bicarbonate, pH 7.9. After overnight incubation at 37 °C, the pH was adjusted to 2 with trifluoroacetic acid to stop the reaction. One μg of the tryptic digest mixture was desalted on PepClean™ C-18 spin columns (Pierce, Rockford, Il, USA) and injected to LC-ESI MS/MS. In some experiments aimed to demonstrate the conversion of the -2 a.m.u. form to dehydroalanine (Dha), tryptic petides were obtained as in the previous section but having also 5 mM mercaptoethanol present during digestion. When a short proteolysis (10 min) was adopted, the enzyme/substrate ratio was 1:4 (w/w). Two different MS instruments and pertinent data processing procedures were used. Orbitrap technology: μLC-MS/MS analysis of tryptic fragments was performed with a Surveyor MS high pressure liquid chromatography pump (Thermo Fisher Scientific, San Josè, CA, USA), coupled to an LTQ-OrbitrapXL-ETD mass spectrometer (Thermo Fisher Scientific) by a NSI interface (Thermo Fisher Scientific). The column was a Biobasic C18 (Thermo Fisher) eluted with an acetonitrile gradient from 0.1 % formic acid in water (buffer A) to 0.1 % formic acid in acetonitrile (buffer B). The flow rate was 100 μL/minute split in order to achieve a final flux of 1.5 μL/minute. The spray capillary voltage was set at 1.5 kV, while the ion transfer capillary temperature was held at 220°C. Full MS spectra were recorded in positive ion mode over a 700– 1800 m/z range, with a resolving power of 100,000 FWHM (at m/z = 400) with 2 microscans per second. Analysis of MS/MS spectra was performed using the 3.3.1 Bioworks version, based on SEQUEST algorithm (Thermo Finnigan Corp., San Jose, CA, US). The experimental MS/MS spectra were correlated to tryptic peptide sequences by comparison with the theoretical mass spectra. For SEQUEST analysis, a high stringency was guaranteed as follows: the chosen minimum values of Xcorr were greater than 1.5, 2.0 and 2.5 for single-, double-, triple- charged ions, respectively; normalized correlation ('Cn) was better than 0.1. The peptide mass search tolerance was set to 1.00 Dalton, precursor ion tolerance was set to 1.400 a.m.u. and the threshold to 1,000. Only the first-best matching peptide was taken into consideration and only if the same peptide was found in multiple MS/MS spectra. Moreover, searches were done with ‘no enzyme’ mode to identify Cys and Sec modifications. In this case, tolerance was set to 2.00 a.m.u. for peptides and 1.00 a.m.u. for MS/MS ions. Finally, to assign a final score to proteins, the SEQUEST output data were filtered setting peptide probability to 1x10-3 and consensus score higher than 10.

6

Q-TOF technology: Nano-LC/MS analysis of 6 – 12 ng of each digest was performed with a 1200 Rapid Resolution system (Agilent Technologies, Santa Clara, CA, USA), directly coupled to the nano-ESI source of a 6520 Accurate-Mass Q-TOF LC/MS system (Agilent Technologies, Santa Clara, CA, USA) by means of an HPLC-Chip apparatus. Loaded samples were enriched on a 360 nL enrichment column and resolved on a 75 μm x 150 mm separation column packed with Polaris C18-A 3 μm material by means of a gradient ranging from 8 % to 50 % of buffer B (as above) over 15 min, then from 50 % to 65 % B in 3 min. All spectra were recorded in positive ion mode in the range between 140 and 1700 m/z for MS and 59 to 1700 m/z for MS/MS. In both MS modes the resolving power was 20,000 FWHM at a scan rate of 3 spectra s-1 (for both MS and MS/MS). Spray capillary voltage was set at 1.73 kV, while the ion transfer capillary temperature was held at 325 °C. MS/MS precursors were selected at an isolation width of 4 m/z and fragmented by collision-induced dissociation (CID) by application of a formula weighting collision energy based on precursor charge. When required, specific targeted precursor masses were selected prior to analysis and acquired at an MS/MS scan rate of 1 spectrum s-1 and fragmented at fixed collision energies from 35 to 50 eV. MS/MS data for identification were extracted by means of the MassHunter Workstation Software Qualitative Analysis (Agilent Technologies, Santa Clara, CA, USA) and converted to Mascot generic format. Data were analysed with Mascot server (ver. 2.3, Matrix Science Ltd.) being classified by a probability-based implementation of the Mowse algorithm. Experimental mass spectra were correlated with theoretical mass spectra. For confidence of peptide identification, a precursor mass tolerance of no more than 10 ppm and an MS/MS product mass tolerance of no more than 0.05 a.m.u. were applied. For semi-quantitative evaluation, ionic current from peptides of interest was then extracted (‘extracted ion current: EIC’) with the MassHunter Workstation Software Qualitative Analysis (Agilent Technologies, Santa Clara, CA, USA) along the retention time dimension. The area under each peak was manually integrated.

Effect of starting the reaction cycle from the oxidized enzyme Enzymes were reduced as above and the reducing agent was removed by a buffer exchange repeated twice using NAP-5 and NAP-10 Sephadex G-25 columns (GE Healthcare UK Limited, Little Chalfont, Buckinghamshire, UK) equilibrated with 0.1 M Tris-HCl pH 7.8, containing 1 mM Na-EDTA and 0.1 % (v/v) Triton X-100. Half of the sample was immediately used (reduced enzyme; control) and the remaining half treated for 5 minutes at room temperature with 25 μM PCOOH (oxidized enzyme) in case of GPx4, or with 10 μM H2O2 in case of GPx1. To the samples, 0.2 mM NADPH and 0.5 IU/ml glutathione reductase (from Sigma) were added and the volume adjusted to 2.5 ml with the same buffer. In the experiments where the oxidized enzyme was used, 7

the reaction started with the addition of 0.25 mM to 1 mM glutathione.

Effect of dimedone For these experiments, GPx1 was reduced with 5 mM mercaptoethanol, then 5 mM dimedone was added, immediately followed by 5 μM H2O2. Dimedone (10 mM) was dissolved in 0.5 M Tris-HCl pH 8.8, and the pH adjusted to 7.0 with HCl. The amount of protein injected was 150 fmoles. Following an incubation of 180 min the sample was analyzed as described above for intact protein size and tryptic fragments.

Density functional theory (DFT) calculations The computational mechanistic study was carried out employing state-of-the-art DFT methodologies [35] as implemented in Gaussian programs suite [36, 37]. The geometries of the intermediates and transition states of the catalytic cycles were fully optimized, i.e. without imposing any constraints. A cluster of 7 amino acids was chosen (see text) as representative of the GPx enzyme. Preliminary geometry optimizations of the cluster of amino acids (and a water molecule) were carried out testing B3LYP [38] functional with 6-31G(d) basis set (level of theory: B3LYP/631G(d)) and PBE1PBE [39] functional with 6-31G(d) basis set (level of theory: PBE1PBE/631G(d)). The former level of theory was chosen on the basis of the best congruence with the crystallographic structure (see Results). In addition the same level of theory has been used in a previous mechanistic study [40] and has been validated in subsequent studies in which a larger portion of the enzyme is included at lower level of theory [41]. This enforces the reliability of the computational protocol here employed. In the present study we have decided to include one water molecule mainly for structural purposes. The location of this water molecule was chosen in line with previous DFT calculations of the GPx active site [40, 41]. In fact, it has been shown that the presence of two water molecules in the enzymatic core improves the RMSD of the calculated geometry [42]. Therefore, also those waters that form during the catalysis were kept. A larger basis set was also tested in geometry optimization of the initial cluster [level of theory: B3LYP/6311G(d,p)], but this did not lead to significant improvement of the RMSD (see Results). Frequency calculations were carried out at the same level of theory, i.e. B3LYP/6-31G(d), to confirm the obtained minimum energy structures (no imaginary frequencies) and transition state structures (one imaginary frequency associated to the correct reaction coordinate). QST3 algorithm [43], implemented in Gaussian software was used to search the transition state structures using a guess obtained from a series of constrained geometry optimizations varying the relevant reaction coordinate. 8

One H2O2 (initially located in correspondence to the substrate accession site revealed by MD calculations [44]) and two glutathione molecules, here modeled as ethane-thiol (CH3CH2SH), are involved in the catalytic reactions. Since their exact binding sites prior to the reactions is unknown, the calculations of reliable entropic contributions is not possible. Thus, entropy effects are not included in the energetics. This approximation has been considered adequate also in precedent studies [40]. Solvent effects have been taken into account using the Polarizable Continuum Model [45]; a dielectric constant of 4.0 has been used, as suggested in previous studies for a protein environment [40]; solvent correction calculated at B3LYP/6-31G(d) level was added to single point energies. If not stated differently, energy values in the text are calculated at B3LYP/6-311+G(d,p) and include ZPC + solvent correction (calculated at B3LYP/6-31G(d)). Refinements to optimize the energetics other than full optimization of the intermediates and transition states and single point calculations with standard thermodynamic and solvent corrections were not attempted.

Results

DFT data of the canonical catalytic cycle and the mechanism of Sec oxidation Quantum mechanics-based calculations corroborated the proposed canonical scheme of SecGPx catalysis (Fig. 1a), with F being the selenenic acid form. Applying the DFT, we obtained an energetic profile along the entire catalytic cycle comprising the postulated intermediates and pertinent transition states (Fig. 1b and Movies S1-4). It is similar for CysGPx, with F being a sulfenic acid. Only the activation energies are higher, in line with the lower efficiency of sulfur versus selenium catalysis [29]. For both types of enzymes, the activation energies for the reductive part are larger, which complies with k´+2 being lower than k+1. However, the activation energies of the reductive part of the cycle could be more favorable, since, for technical limitations, we had to choose ethane-thiol as substitute substrate. Binding of physiological substrates such as GSH, thioredoxins and other thiol proteins [6, 15, 18, 19, 46-48] involves remote residues whose inclusion would have exceeded the limits DFT calculations. The energy profile of the entire catalytic cycle, as here obtained, is similar to previous DFT studies on GPx, performed with a smaller and less representative set of residues, which was based on human GPx3 [40, 41]. The active site cluster, here selected to represent the ground state E, comprises all conserved amino acids that, according to site-directed mutagenesis, affect GPx activity. In rat cytosolic GPx4, these are Sec46, Gln81, Trp136 [46, 49, 50] and Asn137 [51] called the ‘catalytic tetrad’ [51]. For optimum compliance with established x-ray structures, we added the conserved Phe138 and Gly47 9

plus a second Gly to mimic a peptide bond linking Gly47 to the non-conserved down-stream residue. With this selection of residues, geometry optimization yielded an active site model that perfectly matched with crystallographic data (Fig. 2). The use of constraints such as fixing the αcarbons in x-ray-verified positions thus proved to be unnecessary and the model was considered representative for all subtypes of glutathione peroxidases in their ground state E. For calculating the Sec (or Cys) oxidation by H2O2 and the downstream reactions, we complemented the active site with one water molecule, since water proved to be involved in various elementary steps of the catalysis of the GPx mimic ebselen [52, 53] and had also been discussed to possibly contribute to GPx catalysis [40]. This single water molecule proved indeed to be pivotal for GPx catalysis. It allows a charge separation that, irrespective of absence or presence of substrate, leads to abstraction of the proton from the catalytic selenol or thiol, respectively. Only in the presence of water, the DFT calculations yielded the selenolate or thiolate form of the GPx model, which is amply documented to prevail at physiological pH in natural GPxs [49, 50] and analogous peroxiredoxins [54, 55]. In the latter enzyme family, the active site thiolate is essentially generated by positive charges of conserved arginine residues and a Cys-coordinated Thr (Ser) [55, 56]. Strong basic residues being missing in the active site of GPxs, it is here the water-mediated charge separation that provides the nucleophilicity to Sec (Cys) required to react with a hydroperoxide. Deprotonation of the active site chalcogen is, however, not the only contribution of the charge separation to catalysis. Although the dislocated proton in our model is seen at the ring nitrogen of Trp136 (Movies S1 and S2), it could in essence move to any of the acceptor groups of the reaction center, in each case occupying an extremely labile location. Shuttling back from such positions, the mobile proton can contribute to the cleavage of the peroxide bond of a properly bound hydroperoxide by creating H2O as an ideal leaving group. Fig. 3 illustrates the most favorable interplay between H2O2, water and the tetrad residues for the oxidative part of the catalytic cycle. The E to F transition is initiated by formation of a weak complex with H2O2 ([E•H2O2] in Fig. 1b and 3). The pose of H2O2 in this complex complies with the one obtained by molecular dynamics by Fogolari et al [44]. By polar interactions, H2O2 is fixed between the carbonyl oxygen of the Asn137-Phe138 bond, the selenol and the water, which in turn is integrated in a relay of mobile protons spanning from the Asn137 carboxamide nitrogen over the Gln81 carboxamide and the water to the imino nitrogen of Trp136. In the transition state (TS1), the selenol proton moves via H2O2 and water to the Trp nitrogen, thus creating a charge-separated species (CS; Fig. 1b). The resulting complex [CS•H2O2] is so unstable that it decays without any energy barrier. In a concerted mechanism, the peroxide bond is cleaved by two events: the nucleophilic attack of the selenolate on the more buried peroxide oxygen and shuttling back of the 10

proton from Trp136 via water to the surface-exposed oxygen. The products are the selenenic acid form F and an additional water or an alcohol, if the substrate is an alkyl-hydroperoxide (for details see Movies S1-2).

Identification of oxidized GPx forms by DFT-supported MS analysis Although DFT-calculations corroborated the concept of F being a selenenic acid, we failed to verify this structure by MS. When we incubated rat SecGPx4 with low concentrations of hydroperoxide in the absence of GSH, we could not detect any enzyme form or fragment suggestive of Sec being oxidized to a selenenic or seleninic acid. Also the attempt to trap the selenenic acid residue by dimedone, which is widely used to derivatize cysteine sulfenic acid residues [57] and supposed to equally react with selenenic acids [58], failed (Suppl. Fig. S1). Already in 2003, we had reported that porcine GPx4 seemed to shrink by 2 a.m.u. upon oxidation instead of being enlarged by an oxygen atom [19]. This preliminary evidence is here confirmed for rat GPx4 and human GPx1 and, thus, appears to be a general phenomenon in SecGPxs. In carefully controlled MS analyses, oxidized human GPx1 and rat GPx4 adopt masses that are 2 a.m.u. smaller than those of the reduced enzymes (Fig. 4 and 5), suggesting a fast loss of H2O after formation of the selenenic acid form. Of notice, apart from these mass decrements and small shifts in HPLC retention times, no additional change upon oxidation of the enzymes could be detected by HPLC-MS analysis. These observations kept puzzling us, since structural considerations excluded a selenadisulfide formation in these enzymes that could explain the mass decrements. We were also unable to verify any reaction of Sec with other functional side chains in the reaction center by MS analysis. The solution of the enigma was suggested by the DFT-structure of F, where the selenium adopts a position that enables a reaction with the nitrogen of the peptide bond downstream of Gly47. The predicted product is a novel oxidized intermediate F´, in which the selenium is integrated as selenenylamide in a relaxed 8-membered ring (Fig. 6 and Movie S5). LC-MS/MS analysis complied with this structure. Selenium-containing tryptic peptides derived from both, GPx1 (residues 42-57) and GPx4 (residues 34-48), showed the same mass decrements of 2 a.m.u. as the intact proteins. By MS/MS sequencing we could narrow down the position of the missing 2 a.m.u. to the fragment UGTTVR in GPx1 and UGK in GPx4 (Figs. 7 and 8). A final structural characterization of these tryptic peptides proved to be impossible, since the Sec-containing ones were obtained in minute amounts only (for reasons see below). However, considering the identity of the active site structure of GPx1 and GPx4, we can reasonably assume that a bond is formed between selenium and one of the downstream peptide nitrogens, as they are shared by both fragments. This consideration leaves two options: a five-membered ring formed 11

between the Se of Sec and the immediate downstream peptide nitrogen of Gly47 or an eightmembered ring involving the N of the next downstream peptide bond (Gly47-Lys48 in GPx4 or Gly-Thr in GPx1). A five-membered ring seemed also to be possible according to DFT calculations. However, its energy (enthalpy in solution) was 27.5 kcal mol-1 higher than that of F´ and a transition state indicating the formation of this structure from the selenenic acid form could not be identified. Instead, a transition state for the selenenic acid form to the eight-membered ring was clearly found and the resulting structure presents as energetically favored and relaxed. Collectively, these considerations justify the conclusion that the -2 a.m.u. species (F´) most likely contains the eight-atom ring structure predicted by DFT calculation (Fig. 6). An analogous F´ formation was not predicted for the CysGPx4 mutein. A sulfenylamide analog of the selenenylamide F´ appeared energetically possible. However, the DFT calculations failed to find a transition state for its formation from the sulfenic acid form. This surprised, since an analogous sulfenylamide bond had been detected by crystallographic analysis of oxidized protein tyrosine phosphatase 1B [59, 60]. We therefore tried to identify the still suspected sulfenylamide form of the rat CysGPx4 mutein by LC-MS following the same procedure used in the MS-analysis of the natural SecGPxs. In agreement with the DFT-predictions, we could not detect any trace of a molecular species with a molecular mass lowered by 2 a.m.u. upon oxidation. Instead, we found an increase of the molecular mass by 32 a.m.u. in the oxidized CysGPx4, suggesting an enzymatic species having the catalytic Cys over-oxidized to a sulfinic acid (Fig. 9a). This outcome was, in fact, predicted by DFT calculation (Movie S7). MS sequencing of the tryptic peptide T34-48 derived from the oxidized CysGPx4 in essence confirmed the assumption showing the catalytic Cys further over-oxidized to a sulfonic acid and two non-catalytic cysteines disulfide-bonded (Fig. 9b). Functional characterization of the selenenylamide form F´ The activation energies for the F to F´ transition and the reduction of F´ by thiols are relatively high (50.7 and 42.0 kcal/mol, respectively), when compared to the complex [EyyH2O2] to F and the complex [FyRSH] to G transition of the canonical GPx cycle (21.9 and 18.1 kcal/mol, respectively). Therefore, these cyclization and thiolysis reactions should hardly be competitive. A modification of GPx activity measurement verified this conclusion. For standard GPx determination, pre-incubation with GSH for about 5 minutes is recommended [34] to activate ‘sluggish enzyme forms’, the nature of which had remained elusive [61, 62]. They are formed whenever a SecGPx is exposed to a hydroperoxide in the absence of a thiol substrate or even when only stored under aerobic conditions [63]. Here we first incubated SecGPx4 with low concentrations of phosphatidyl choline hydroperoxide (25 μM), which according to LC-MS/MS almost exclusively yields F´, and started the reaction with GSH. At 0.25 mM GSH, 12

full activity was not resumed but after ~70 seconds. The visibly slower rate, however, almost disappeared at physiological GSH concentration (Fig. 10). This ‘lag phase’ evidently reflects the time required for the generation of G from F´, which means for the re-integration of F´ into the canonical catalytic cycle via thiolysis of the Se-N bond. Similar results were obtained with GPx1 (not shown). Collectively, we conclude that F´ is indeed the slow-reacting intermediate, which is generated when F does not meet a suitable reductant in time, but fast enough to prevent detection or trapping of the primary intermediate F.

Dehydroalanine formation by ș-cleavage of selenocysteine in GPx Loss of selenium from GPx1 accompanied by complete inactivation has been reported to occur under oxidizing conditions and implicated in redox signaling [26]. Mechanistically, βcleavage of a still unidentified oxidized modification of the active site Sec is assumed, the resulting product being a dehydroalanine residue. Dha was indeed detected in peptides derived from human GPx1 [26] and rat GPx3 and selenoprotein P [28] by mass spectrometry. We, therefore, systematically screened the MS spectra of intact GPx1and derived fragments for Dha content. Our sample GPx1 contained traces of material with a molecular mass compatible with the loss of H2Se from native GPx1. The amount in the commercial sample was estimated to range around 1 % of the total material and this percentage did not change, when the enzyme was subjected to the routine procedure to generate the oxidized species (Fig. 11a). Expectedly, the by far predominant molecular species were those having the Sec involved in the selenenylamide bond. Precisely the opposite data was obtained, when the tryptic peptide T42-57 comprising the active site Sec was investigated: Now about 99 % of the material proved to have the selenium eliminated (Fig. 11b) and the selenenylamide form, which predominated in the intact protein, amounted to 1% only. This βelimination of the selenium can be largely prevented, if the tryptic digestion is performed in the presence of β-mercatoethanol (Fig. 11c). Now the majority of the fragment is obtained with the selenium thiylated by mercaptoethanol, while the selenenylamide form completely disappears. The correctness of the assignment of masses to molecular species is further corroborated by the isotopic distribution of the fragment masses displaying the isotopic pattern typical of selenium in the selenenylamide and Se-thiylated peptides (Fig. 11e and f), but not in the Dha form (Fig. 11d). In retrospect, these data explain the difficulties we faced while elucidating the molecular structure of oxidized GPx forms. More importantly, the results reveal that the selenenylamide bond is stable within the peculiar reaction center of the glutathione peroxidases, but becomes extremely unstable, as soon as the three-dimensional architecture of the proteins is disrupted.

13

Discussion

Technical problems In our attempt to understand the reaction cycle of glutathione peroxidases in more detail, we had to face difficulties that result from the unusual kinetics of these enzymes, limited structural knowledge and, as far as the SecGPxs are concerned, from the reactivity of selenium in whatever functional form. The GPx catalysis in general is characterized by ping-pong kinetics with infinite Vmax and Km, thus corresponding to the mechanism IV in the system of enzymatic mechanisms compiled by Keith Dalziel [64]. In mechanistic terms, the enzyme and the catalytic intermediates form binary complexes with each of the substrates followed by co-substrate-independent reactions within the complexes (mechanism IVi) or the mechanism just represents a sequence of bimolecular reactions not involving any enzyme/substrate complex (mechanism IVii). Lacking maximum velocities argue in favor of mechanism IVii, but also comply with mechanism IVi, if the formation of complexes is slower than the reaction within the complexes. Considering substrate specificities, the latter case applies to GPxs. In consequence, any attempt to obtain crystallographic evidence for complexes between the enzyme or the catalytic intermediates with substrates must fail. Catalytic intermediates are instable by definition and this is particularly true in case of GPxs. Accordingly, only the DFT calculations for the ground state enzyme (Fig. 2) could be validated by x-ray data. In 2-CysGPxs, which are not the subject of the present study, massive structural changes, similar to those known from 2-Cys peroxiredoxins [65], accompany the E to F transition [48, 50]. However, discrete structural changes may also occur in the reaction cycle of 1-CysGPxs and SecGPxs. Binding of their medium to high size substrates involves residues that are far outside the modeled reaction center [47, 48, 50, 66]. The possible impact on the active site structure remained unconsidered when calculating the reductive part of the catalytic cycle with a non-natural small substitute substrate. Quantitatively, therefore, the calculated activation energies for the oxidative part of catalysis can be rated as more reliable than those of the reductive part. The instability of Sec derivatives rendered the verification of the DFT-calculated nature of F or F´, respectively, a nightmare. We did not expect to detect the selenenic acid postulated for the canonical GPx cycle by LC-MS/MS directly, because we were aware of its notorious instability. However, we were quite disappointed when failing to identify this component by trapping agents amply used to derivatize sulfenic acids. When chasing the selenenylamide-containing component, the poor yields of a selenium-containing tryptic peptide bothered us. Only when we started to search 14

for Dha formation in the GPxs, we discovered the real problem. Whenever the selenocysteine residue is derivatised, it becomes prone to β-cleavage under the conditions of tryptic digestion, the so far only exception being Se-thiylation. As evident from Fig. 11, about 99 % of the selenium present in F´ was lost during tryptic digestion and shortening the digestion to 10 minutes did not change the outcome significantly. Other digestion procedures, as used by Ma et al. [28], did not yield better results. Evidently, the selenium moiety is reasonably stable only within the architecture of the native active site. In so far, we can regard it as a lucky strike that we finally, by means of the characteristic selenium isotopic distribution, found the traces of peptides containing the selenenylamide bond and could partially sequence them. On the other hand, the tendency of derivatized selenocysteine to undergo β-cleavage also implies the possibility that selenium modifications in selenoproteins similar, although distinct, from the one here identified can be easily overlooked in MS analysis. Another unpleasant property of selenocysteine is its ability to undergo exchange reaction with disulfides more readily than any thiol. Unlike thiol/disulfide exchange, the homologous reactions with selenium cannot be prevented by low pH and could only be decreased, but not prevented by thiol alkylation with, e.g., iodo-acetamide, since traces of hidden thiols, liberated by denaturation for proteolysis, re-initiated the reshuffling. In fact, we detected numerous peptides containing Se-S bonds and even diselenides. Most of them could clearly be classified as artifacts by means of Sec to Cys distances in the native enzymes, others might have resulted from polymerization reactions of selenium with surface-exposed cysteines, although we largely avoided intermolecular reactions by use of detergents and low protein concentrations. The only way to limit these disturbances to a reasonable extent was to shorten the usual overnight trypsin digestion of the samples. At an incubation time of 10 minutes, the artifacts disappeared, yet we had to struggle with incomplete proteolysis.

The GPx efficiency problem Our DFT calculations of the GPx reaction cycle seemingly just confirm what had generally been accepted: oxidation of the deprotonated Sec (Cys) to a selenenic (sulfenic) acid and stepwise reduction of the latter by thiols. However, as already mentioned in the introduction, the deprotonation of the active site selenol (thiol) by itself does not explain why the catalytic Sec (or Cys) of the GPxs reacts faster with hydroperoxide than any dissociated low molecular mass selenol or thiol by 4 to 6 orders of magnitude. The DFT calculations not only explain the low pKa of the catalytic residue, but also the unusual reactivity and the unusual kinetics. It is the H2O-mediated charge separation that enforces the selenol (thiol) deprotonation, thus creating a sufficiently 15

aggressive nucleophile. At the same time, the proton delocalized from the chalcogen remains in the catalytic tetrad, preferentially bound to the conserved Trp. Thereby a highly energized situation is created, which can relax, once a hydroperoxide is suitably complexed between the deprotonated chalcogen and the Trp. Then, the charge separation is reversed, yet not restoring the starting situation. In a concerted reaction, the delocalized proton shuttles back via water towards the negatively charged Se or S, respectively, but hits the more surface-exposed oxygen of the peroxide bond, while the nucleophilic selenium (sulfur) simultaneously attacks the second oxygen from the opposite site. In consequence, the peroxide bond is cleaved, and H2O as ideal leaving group, instead of OH-, can escape from the reaction center and the selenenic acid is formed. It is this concerted reaction, which makes the difference between the catalytic H2O2 decomposition and any nonenzymatic process and explains the extreme velocity of the enzymatic catalysis. Earlier concepts to explain the catalytic efficiency of GPx often assumed a direct activation of its active site selenol (thiol) by side chains of the tetrad residues via hydrogen bonding [24, 49, 51]. Such interaction is refuted by the DFT calculations, in line with the chalcogens’ tendency to form hydrogen bridges, which substantially declines from O to S and Se. Hydrogen bonding only exists between the tetrad components. They contribute to the overall stability of the reaction center, to binding of H2O2 and facilitate the proton shuttling, but do not directly involve the selenium or sulfur. In essence, the DFT-based mechanism presents as a complex water-mediated acid-base catalysis, which, under the structural constrains of the active site architecture, results in fast oxidation of the chalcogen. Otherwise, the mechanism here proposed complies with established knowledge. The calculated charge separation is in line with alkylation kinetics [49, 50, 63] revealing that the selenol (thiol) in the GPx reaction center can be deprotonated irrespective of the presence of a hydroperoxide substrate. As discussed above, the lack of any saturation kinetics, as observed with GPxs [29, 63, 67, 68], can be interpreted as absence of any enzyme/substrate complex or as indicating a very fast reaction within such complex. The DFT calculations indeed reveal a specific, although weak, complexation of H2O2 by the enzyme and, thus, support mechanism IVi (see above). The instability of [CS•H2O2], however, guarantees that this complex can never accumulate, whereby mechanism IVii (no complex at all) is mimicked. With this clarification, we can also offer a physical definition of the rate constant k+1 for the oxidation of the enzyme by H2O2: It is the net forward rate constant for the complex formation. The oxidation of the enzyme within [CS•H2O2] occurs without any energy barrier. It is a not rate-limiting instantaneous event.

16

The GPx stability problem With the discovery of the selenenylamide formation in SecGPxs we offer an explanation how these enzymes overcome over-oxidation even in the absence of any reducing substrate. The selenocysteine, when oxidized to a selenenic acid in the first catalytic step, reacts sufficiently fast with the enzyme’s physiological thiol substrate. If the latter becomes limited, the selenenic acid, without a safety mechanism, would readily react with any accessible thiol, amino, amido or other nucleophilic group with the risk of being decomposed by selenium loss via β-cleavage and major structural disturbance. The typical SecGPxs evidently prevent such irreversible inactivation by the formation of a selenenylamide bond with a peptide nitrogen of the protein backbone. In this form (F´) the enzyme is stable and readily re-animated to full activity by its physiological reducing substrate. Thus, in this enzyme sub-family a mechanism evolved to overcome a transient hydroperoxide challenge in the absence of reducing substrate, as may occur under conditions of substantial oxidative stress. This self-protecting mechanism is analogous to the formation of inter- or intramolecular disulfide bonds in typical 2-Cys-peroxiredoxins, atypical 2-Cys-peroxiredoxins and 2-CysGPxs, in which the catalytic Cys (CP), after oxidation to sulfenic acid, reacts with a resolving Cys (CR) to build a stable disulfide. A sulfenylamide formation, although reported for protein phosphatases, has never been implicated in any of the thiol peroxidases and, in fact, could not be detected in the Cys mutein of GPx4. Unlike natural thiol peroxidases, this CysGPx4, which does not have a CR, reacted with over-oxidation of its catalytic Cys to a sulfinic/sulfonic acid and massive aggregation, when exposed to conditions leading to F´ formation in the natural SecGPx4. Similar observations were made with 2-Cys-peroxiredoxins of E. coli [69] and M. tuberculosis [70], when their CR was compromised by site directed mutagenesis. In short, oxidized SecGPxs are stabilized in their selenenylamide form, 2-Cys-peroxiredoxins and likely 2-CysGPxs as disulfides. Similarly, the natural vertebrate GPx7 copes with oxidative challenge forming an intramolecular disulfide involving a non resolving Cys [71]. How other 1-CysGPx homologs such as GPx5 and GPx8 do, still remains enigmatic.

Reversible versus irreversible inactivation of SecGPx and possible implications Recently, Cho et al. [26] have reported on a fundamental difference between the response of GPx1 and 2-Cys-peroxiredoxins, here PrxII, to H2O2 challenge: GPx1 is reported to be irreversibly inactivated in red blood cells due to selenium elimination, whereas PrxII is over-oxidized to a sulfinic form, from which it is re-activated by sulfiredoxin. Although we do not doubt that selenium 17

elimination from selenoproteins is a permanent risk and may happen due to persistent oxidative stress, the inferred physiological relevance of irreversible GPx inactivation by H2O2 deserve some critical comments. The authors suggest that the decline of GPx activity with aging of erythrocytes results from β-elimination of Se from an over-oxidized Sec residue, likely a seleninic acid. They support their conclusion with the detection of Dha in tryptic peptides derived from isolated oxidized GPx1 and labeling of the Dha of denatured oxidized GPx1 in cell lysates by a biotin-conjugated cysteamine-based probe. However, the formation of a seleninic acid form is unlikely according to our DFT calculations and could experimentally not be observed at all. Instead, almost exclusively the selenenylamide form F´, which we do not consider an oxidatively destructed enzyme, is detected upon oxidation of SecGPx. Under the conditions of proteolytic digestion, however, F´ and likely any kind of denatured oxidized SecGPx loses Se with formation of Dha and, thus, leads to a dramatic overestimation of the latter in the native enzyme. Moreover, the oxidation condition applied by Cho et al. [26], one hour incubation at 1mM H2O2, are far beyond any physiological challenge. According to the still valid review by Chance et al. [72], the average cellular steady state of H2O2 is estimated to range between 10-9 to 10-7 M. The H2O2 concentration may be speculated to be one or two orders of magnitude higher in the intimate vicinity of an activated NADPH oxidase. Even at these speculative levels and in the absence of reducing substrate we could not detect any de novo Dha formation in the native enzyme, but only the artifactual one after tryptic cleavage (Figs. 7, 8 and 11). We therefore conclude that under near physiological conditions, there is no fundamental difference between SecGPxs and 2-Cys-peroxiredoxins in their response to H2O2: the former are transformed to the less active selenenylamide form F´ and the latter to the sulfinic form. Both modifications are reversible: F´ is re-activated within seconds to minutes by GSH (Fig.10), overoxidized 2-Cys-peroxiredoxin is reactivated by sulfiredoxin, probably at a lower rate. This analogy justifies the conclusion that transient activity changes of both enzyme types may have analogous consequences. They may, e. g., be reversibly inactivated to save H2O2 required for signaling according to the “floodgate theory” proposed by Wood, Poole and Karplus [73]. Moreover, we may speculate about other roles of F´ in redox regulation. F´ has a substrate interaction site distinct from the main-stream intermediate F. According to DFT calculation and our functional experiments (Fig. 10), the reaction of F´ with GSH to yield G is less favored than the F to G transition of the canonical GPx cycle. Possibly, F´ is optimized for the reaction with alternate substrates and, thus, serves as transducer in redox regulation, as has been proposed for the oxidized yeast 2-CysGPx3 by the group of Michel Toledano [18] and for oxidized 2-Cys-peroxiredoxins by Morgan and Veal [74]. 18

Synopsis

We calculated the energetic profile of the entire catalytic cycle of GPx by the density functional theory, starting from the widely accepted assumption that, in the ground state enzyme E, a deprotonated Sec or Cys is oxidized by hydroperoxide to a selenenic or sulfenic acid (F), respectively, which is stepwise reduced by thiols. This canonical GPx cycle was corroborated by DFT calculations. The first step of the catalysis was addressed with the intention to understand its extreme velocity. A DFT-calculated model of the active site, comprising all residues that have been shown to affect activity and being representative of > 99 % of published GPx sequences, revealed that a charge separation within the active site is obtained with the aid of a single water molecule. It results from a delocalization of the selenol (thiol) proton to the imino nitrogen of the conserved Trp in the catalytic tetrad. The charge separation explains the low pKa of the catalytic Sec or Cys, which is a prerequisite for reactivity with any hydroperoxide. Moreover, the complex between the chargeseparated species and H2O2 is energetically so unstable that it decomposes without any energy barrier. In this complex decay, the delocalized proton associates with OH- to generate H2O as ideal leaving group, while Se- (S-) combines with OH+ to build the selenenic (sulfenic) acid. The mechanism complies with steady state kinetics and, for the first time, offers a plausible explanation for the extreme rates for the reaction of GPxs with hydroperoxides. We failed to identify the postulated selenenic (sulfenic) acid intermediate by LC-MS/MS. Instead, we identified a selenenylamide bond involving a peptide backbone nitrogen (alternate first intermediate F´) in two SecGPx species, rat GPx4 and bovine GPx1, and found the catalytic Cys oxidized to sufinic/sulfonic acid in the Cys mutein of GPx4, when the enzymes were oxidized in the absence of GSH. The experimental outcome complied with DFT predictions. The selenenylamide formation is reversed by GSH. F´ is thereby transformed into the second intermediate G (E-Se-SG) of the canonical cycle. This re-integration of F´ is less favored than the canonical F to G transition. The selenenylamide bypass is considered to be a safety mechanism to specifically protect SecGPxs from over-oxidation under shortage of reducing substrate. Possible other functions of F´ are discussed in the light of topical theories of redox regulation.

Acknowledgments This study was supported in part by the grant HFSP to F.U. (RGP0013/2014) to FU 19

References [1] [2]

[3] [4]

[5]

[6]

[7]

[8]

[9]

[10]

[11]

[12]

[13]

Mills, G. C. Hemoglobin catabolism. I. Glutathione peroxidase, an erythrocyte enzyme which protects hemoglobin from oxidative breakdown. J Biol Chem 229:189-197; 1957. Toppo, S.; Vanin, S.; Bosello, V.; Tosatto, S. C. Evolutionary and structural insights into the multifaceted glutathione peroxidase (Gpx) superfamily. Antioxid Redox Signal 10:15011514; 2008. Flohé, L. The impact of thiol peroxidases on redox regulation. Free radical research:DOI: 10.3109/10715762.10712015.11046858; 2015. McClung, J. P.; Roneker, C. A.; Mu, W.; Lisk, D. J.; Langlais, P.; Liu, F.; Lei, X. G. Development of insulin resistance and obesity in mice overexpressing cellular glutathione peroxidase. Proceedings of the National Academy of Sciences of the United States of America 101:8852-8857; 2004. Kretz-Remy, C.; Mehlen, P.; Mirault, M. E.; Arrigo, A. P. Inhibition of I kappa B-alpha phosphorylation and degradation and subsequent NF-kappa B activation by glutathione peroxidase overexpression. The Journal of cell biology 133:1083-1093; 1996. Ursini, F.; Heim, S.; Kiess, M.; Maiorino, M.; Roveri, A.; Wissing, J.; Flohé, L. Dual function of the selenoprotein PHGPx during sperm maturation. Science 285:1393-1396; 1999. Maiorino, M.; Bosello, V.; Ursini, F.; Foresta, C.; Garolla, A.; Scapin, M.; Sztajer, H.; Flohé, L. Genetic variations of gpx-4 and male infertility in humans. Biology of reproduction 68:1134-1141; 2003. Foresta, C.; Flohé, L.; Garolla, A.; Roveri, A.; Ursini, F.; Maiorino, M. Male fertility is linked to the selenoprotein phospholipid hydroperoxide glutathione peroxidase. Biology of reproduction 67:967-971; 2002. Conrad, M.; Moreno, S. G.; Sinowatz, F.; Ursini, F.; Kolle, S.; Roveri, A.; Brielmeier, M.; Wurst, W.; Maiorino, M.; Bornkamm, G. W. The nuclear form of phospholipid hydroperoxide glutathione peroxidase is a protein thiol peroxidase contributing to sperm chromatin stability. Molecular and cellular biology 25:7637-7644; 2005. Brigelius-Flohé, R.; Friedrichs, B.; Maurer, S.; Schultz, M.; Streicher, R. Interleukin-1induced nuclear factor kappa B activation is inhibited by overexpression of phospholipid hydroperoxide glutathione peroxidase in a human endothelial cell line. Biochem J 328 ( Pt 1):199-203; 1997. Seiler, A.; Schneider, M.; Forster, H.; Roth, S.; Wirth, E. K.; Culmsee, C.; Plesnila, N.; Kremmer, E.; Radmark, O.; Wurst, W.; Bornkamm, G. W.; Schweizer, U.; Conrad, M. Glutathione peroxidase 4 senses and translates oxidative stress into 12/15-lipoxygenase dependent- and AIF-mediated cell death. Cell Metab 8:237-248; 2008. Friedmann Angeli, J. P.; Schneider, M.; Proneth, B.; Tyurina, Y. Y.; Tyurin, V. A.; Hammond, V. J.; Herbach, N.; Aichler, M.; Walch, A.; Eggenhofer, E.; Basavarajappa, D.; Radmark, O.; Kobayashi, S.; Seibt, T.; Beck, H.; Neff, F.; Esposito, I.; Wanke, R.; Forster, H.; Yefremova, O.; Heinrichmeyer, M.; Bornkamm, G. W.; Geissler, E. K.; Thomas, S. B.; Stockwell, B. R.; O'Donnell, V. B.; Kagan, V. E.; Schick, J. A.; Conrad, M. Inactivation of the ferroptosis regulator Gpx4 triggers acute renal failure in mice. Nature cell biology 16:1180-1191; 2014. Yang, W. S.; SriRamaratnam, R.; Welsch, M. E.; Shimada, K.; Skouta, R.; Viswanathan, V. S.; Cheah, J. H.; Clemons, P. A.; Shamji, A. F.; Clish, C. B.; Brown, L. M.; Girotti, A. W.; Cornish, V. W.; Schreiber, S. L.; Stockwell, B. R. Regulation of ferroptotic cancer cell death by GPX4. Cell 156:317-331; 2014.

20

[14]

[15]

[16]

[17] [18] [19]

[20] [21] [22] [23]

[24] [25]

[26]

[27]

[28]

[29]

[30] [31]

Brault, C.; Levy, P.; Duponchel, S.; Michelet, M.; Salle, A.; Pecheur, E. I.; Plissonnier, M. L.; Parent, R.; Vericel, E.; Ivanov, A. V.; Demir, M.; Steffen, H. M.; Odenthal, M.; Zoulim, F.; Bartosch, B. Glutathione peroxidase 4 is reversibly induced by HCV to control lipid peroxidation and to increase virion infectivity. Gut; 2014. Bosello-Travain, V.; Conrad, M.; Cozza, G.; Negro, A.; Quartesan, S.; Rossetto, M.; Roveri, A.; Toppo, S.; Ursini, F.; Zaccarin, M.; Maiorino, M. Protein disulfide isomerase and glutathione are alternative substrates in the one Cys catalytic cycle of glutathione peroxidase 7. Biochim Biophys Acta 1830:3846-3857; 2013. Bosello-Travain, V.; Forman, H. J.; Roveri, A.; Toppo, S.; Ursini, F.; Venerando, R.; Warnecke, C.; Zaccarin, M.; Maiorino, M. Glutathione peroxidase 8 is transcriptionally regulated by HIFalpha and modulates growth factor signaling in HeLa cells. Free Radic Biol Med 81:58-68; 2015. Brigelius-Flohé, R.; Maiorino, M. Glutathione peroxidases. Biochim Biophys Acta 1830:3289-3303; 2013. Delaunay, A.; Pflieger, D.; Barrault, M. B.; Vinh, J.; Toledano, M. B. A thiol peroxidase is an H2O2 receptor and redox-transducer in gene activation. Cell 111:471-481; 2002. Mauri, P.; Benazzi, L.; Flohé, L.; Maiorino, M.; Pietta, P. G.; Pilawa, S.; Roveri, A.; Ursini, F. Versatility of selenium catalysis in PHGPx unraveled by LC/ESI-MS/MS. Biol. Chem. Hoppe-Seyler 384:575-588; 2003. Flohé, L. The labour pains of biochemical selenology: the history of selenoprotein biosynthesis. Biochim Biophys Acta 1790:1389-1403; 2009. Flohé, L.; Günzler, W. A.; Schock, H. H. Glutathione peroxidase: a selenoenzyme. FEBS Lett. 32:132-134; 1973. Rotruck, J. T.; Hoekstra, W. G.; Pope, A. L.; Ganther, H. E.; Swanson, A.; Hafemann, D. Relationship of selenium to GSH peroxidase. Fed. Proc. 31:691; 1972. Goto, K.; Saiki, T.; Okazaki, R. synthesis, structure, and reactions of stable selenenic acids bearing novel bridged calix(6)arene fraimworks. Phosphoroussulphur and silicon 136:475478; 1998. Epp, O.; Ladenstein, R.; Wendel, A. The refined structure of the selenoenzyme glutathione peroxidase at 0.2-nm resolution. European journal of biochemistry / FEBS 133:51-69; 1983. Wendel, A.; Pilz, W.; Ladenstein, R.; Sawatzki, G.; Weser, U. Substrate-induced redox change of selenium in glutathione peroxidase studied by x-ray photoelectron spectroscopy. Biochim Biophys Acta 377:211-215; 1975. Cho, C. S.; Lee, S.; Lee, G. T.; Woo, H. A.; Choi, E. J.; Rhee, S. G. Irreversible inactivation of glutathione peroxidase 1 and reversible inactivation of peroxiredoxin II by H2O2 in red blood cells. Antioxid Redox Signal 12:1235-1246; 2010. Kraus, R. J.; Prohaska, J. R.; Ganther, H. E. Oxidized forms of ovine erythrocyte glutathione peroxidase. Cyanide inhibition of a 4-glutathione:4-selenoenzyme. Biochim Biophys Acta 615:19-26; 1980. Ma, S.; Caprioli, R. M.; Hill, K. E.; Burk, R. F. Loss of selenium from selenoproteins: conversion of selenocysteine to dehydroalanine in vitro. Journal of the American Society for Mass Spectrometry 14:593-600; 2003. Toppo, S.; Flohé, L.; Ursini, F.; Vanin, S.; Maiorino, M. Catalytic mechanisms and specificities of glutathione peroxidases: variations of a basic scheme. Biochim Biophys Acta 1790:1486-1500; 2009. Winterbourn, C. C.; Metodiewa, D. Reactivity of biologically important thiol compounds with superoxide and hydrogen peroxide. Free Radic Biol Med 27:322-328; 1999. Cardey, B.; Enescu, M. A computational study of thiolate and selenolate oxidation by hydrogen peroxide. Chemphyschem : a European journal of chemical physics and physical chemistry 6:1175-1180; 2005.

21

[32]

[33]

[34] [35] [36]

[37]

[38]

[39] [40]

[41]

[42]

Roveri, A.; Maiorino, M.; Nisii, C.; Ursini, F. Purification and characterization of phospholipid hydroperoxide glutathione peroxidase from rat testis mitochondrial membranes. Biochim Biophys Acta 1208:211-221; 1994. Roveri, A.; Maiorino, M.; Ursini, F. Enzymatic and immunological measurements of soluble and membrane-bound phospholipid-hydroperoxide glutathione peroxidase. Methods in enzymology 233:202-212; 1994. Flohé, L.; Günzler, W. A. Assays of glutathione peroxidase. Methods in enzymology 105:114-121; 1984. Parr, R. G.; Yang, W. Density-Functional Theory of Atoms and Molecules. Oxford University Press; 1989. Frisch, M. J. .; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M. A.; Cheeseman, J. R.; Montgomery, J., J.A.; Vreven, T.; Kudin, K. N.; Burant, J. C.; Millam, J. M.; Iyengar, S. S.; Tomasi, J.; Barone, V.; Mennucci, B.; Cossi, M.; Scalmani, G.; Rega, N.; Petersson, G. A.; Nakatsuji, H.; Hada, M.; Ehara, M.; Toyota, K.; Fukuda, R.; Hasegawa, J.; Ishida, M.; Nakajima, T.; Honda, Y.; Kitao, O.; Nakai, H.; Klene, M.; Li, X.; Knox, J. E.; Hratchian, H. P.; Cross, J. B.; Bakken, V.; Adamo, C.; Jaramillo, J.; Gomperts, R.; Stratmann, R. E.; Yazyev, O.; Austin, A. J.; Cammi, R.; Pomelli, C.; Ochterski, J. W.; Ayala, P. Y.; Morokuma, K.; Voth, G. A.; Salvador, P.; Dannenberg, J. J.; Zakrzewski, V. G.; Dapprich, S.; Daniels, A. D.; Strain, M. C.; Farkas, O.; Malick, D. K.; Rabuck, A. D.; Raghavachari, K.; Foresman, J. B.; Ortiz, J. V.; Cui, Q.; Baboul, A. G.; Clifford, S.; Cioslowski, J.; Stefanov, B. B.; Liu, G.; Liashenko, A.; Piskorz, P.; Komaromi, I.; Martin, R. L.; Fox, D. J.; Keith, T.; Al-Laham, M. A.; Peng, C. Y.; Nanayakkara, A.; Challacombe, M.; Gill, P. M. W.; Johnson, B.; Chen, W.; Wong, M. W.; Gonzalez, C.; Pople, J. A. Gaussian 03, Revision D.02. Wallingford, CT: Gaussian, Inc.; 2004. Frisch, M. J. .; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M. A.; Cheeseman, J. R.; Scalmani, G.; Barone, V.; Mennucci, B.; Petersson, G. A.; Nakatsuji, H.; Caricato, M.; Li, X.; Hratchian, H. P.; Izmaylov, A. F.; Bloino, J.; Zheng, G.; Sonnenberg, J. L.; Hada, M.; Ehara, M.; Toyota, K.; Fukuda, R.; Hasegawa, J.; Ishida, M.; Nakajima, T.; Honda, Y.; Kitao, O.; Nakai, H.; Vreven, T.; Montgomery, J. A.; Peralta, J. E.; Ogliaro, F.; Bearpark, M.; Heyd, J. J.; Brothers, E.; Kudin, K. N.; Staroverov, V. N.; Kobayashi, R.; Normand, J.; Raghavachari, K.; Rendell, A.; Burant, J. C.; Iyengar, S. S.; Tomasi, J.; Cossi, M.; Rega, N.; Millam, N. J.; Klene, M.; Knox, J. E.; Cross, J. B.; Bakken, V.; Adamo, C.; Jaramillo, J.; Gomperts, R.; Stratmann, R. E.; Yazyev, O.; Austin, A. J.; Cammi, R.; Pomelli, C.; Ochterski, J. W.; Martin, R. L.; Morokuma, K.; Zakrzewski, V. G.; Voth, G. A.; Salvador, P.; Dannenberg, J. J.; Dapprich, S.; Daniels, A. D.; Farkas, Ö.; Foresman, J. B.; Ortiz, J. V.; Cioslowski, J.; Fox, D. J. Gaussian 09, Revision B 01. Wallingford, CT: Gaussian, Inc.; 2009. Stephens, P. J.; Devlin, F. J.; Chabalowski, C. F.; Frisch, M. J. Ab initio calculation of vibrational absorption and circular dichroism spectra using density functional force fields. J. Phys. Chem 98:11623-11627; 1994. Adamo, C.; Barone, V. J. Chem. Phys. Toward reliable density functional methods without adjustable parameters: The PBE0 model. 110:6158-6169; 1999. Prabhakar, R.; Vreven, T.; Morokuma, K.; Musaev, D. G. Elucidation of the mechanism of selenoprotein glutathione peroxidase (GPx)-catalyzed hydrogen peroxide reduction by two glutathione molecules: a density functional study. Biochemistry 44:11864-11871; 2005. Prabhakar, R.; Vreven, T.; Frisch, M. J.; Morokuma, K.; Musaev, D. G. Is the protein surrounding the active site critical for hydrogen peroxide reduction by selenoprotein glutathione peroxidase? An ONIOM study. J Phys Chem B 110:13608-13613; 2006. Prabhakar, R.; Musaev, D. G.; Khavrutskii, I. V.; Morokuma, K. Does the active site of mammalian glutathione peroxidase (GPx) contain water molecules? An ONIOM study. J. Phys. Chem B 108:12643-12645; 2004. 22

[43] [44]

[45] [46]

[47]

[48]

[49]

[50] [51]

[52] [53] [54]

[55]

[56]

[57]

[58] [59]

Ayala, P. Y.; Schlegel, H. B. A combined method for determining reaction paths, minima, and transition state geometries. J Chem Phys 107:375-384; 1997. Fogolari, F.; Corazza, A.; Toppo, S.; Tosatto, S. C.; Viglino, P.; Ursini, F.; Esposito, G. Studying interactions by molecular dynamics simulations at high concentration. J Biomed Biotechnol 2012:303190; 2012. Tomasi, J.; Mennucci, B.; Cammi, R. Quantum mechanical continuum solvation models. Chem. Rev. 105(8):2999-3094 (2005. Schlecker, T.; Comini, M. A.; Melchers, J.; Ruppert, T.; Krauth-Siegel, R. L. Catalytic mechanism of the glutathione peroxidase-type tryparedoxin peroxidase of Trypanosoma brucei. Biochem J 405:445-454; 2007. Sztajer, H.; Gamain, B.; Aumann, K. D.; Slomianny, C.; Becker, K.; Brigelius-Flohé, R.; Flohé, L. The putative glutathione peroxidase gene of Plasmodium falciparum codes for a thioredoxin peroxidase. J. Biol. Chem. 276:7397-7403; 2001. Maiorino, M.; Ursini, F.; Bosello, V.; Toppo, S.; Tosatto, S. C.; Mauri, P.; Becker, K.; Roveri, A.; Bulato, C.; Benazzi, L.; De Palma, A.; Flohé, L. The thioredoxin specificity of Drosophila GPx: a paradigm for a peroxiredoxin-like mechanism of many glutathione peroxidases. J Mol Biol 365:1033-1046; 2007. Maiorino, M.; Aumann, K. D.; Brigelius-Flohé, R.; Doria, D.; van den Heuvel, J.; McCarthy, J.; Roveri, A.; Ursini, F.; Flohé, L. Probing the presumed catalytic triad of selenium-containing peroxidases by mutational analysis of phospholipid hydroperoxide glutathione peroxidase (PHGPx). Biol. Chem. Hoppe Seyler 376:651-660; 1995. Ma, L. H.; Takanishi, C. L.; Wood, M. J. Molecular mechanism of oxidative stress perception by the Orp1 protein. J Biol Chem 282:31429-31436; 2007. Tosatto, S. C.; Bosello, V.; Fogolari, F.; Mauri, P.; Roveri, A.; Toppo, S.; Flohe, L.; Ursini, F.; Maiorino, M. The catalytic site of glutathione peroxidases. Antioxid Redox Signal 10:1515-1526; 2008. Antony, S.; Bayse, C. A. Modeling the mechanism of the glutathione peroxidase mimic ebselen. Inorg Chem 50:12075-12084; 2011. Orian, L.; Toppo, S. Organochalcogen peroxidase mimetics as potential drugs: a long story of a promise still unfulfilled. Free Radic Biol Med 66:65-74; 2014. Poole, L. B. The Catalytic Mechanism of Peroxiredoxins. In: Flohé, L.; Harris, J. R., eds. Peroxiredoxin Systems: Structure and Function. New York: Springer; 2007: 61-81. Zeida, A.; Reyes, A. M.; Lebrero, M. C.; Radi, R.; Trujillo, M.; Estrin, D. A. The extraordinary catalytic ability of peroxiredoxins: a combined experimental and QM/MM study on the fast thiol oxidation step. Chemical communications 50:10070-10073; 2014. Flohé, L.; Budde, H.; Bruns, K.; Castro, H.; Clos, J.; Hofmann, B.; Kansal-Kalavar, S.; Krumme, D.; Menge, U.; Plank-Schumacher, K.; Sztajer, H.; Wissing, J.; Wylegalla, C.; Hecht, H. J. Tryparedoxin peroxidase of Leishmania donovani: molecular cloning, heterologous expression, specificity, and catalytic mechanism. Archives of biochemistry and biophysics 397:324-335; 2002. Klomsiri, C.; Nelson, K. J.; Bechtold, E.; Soito, L.; Johnson, L. C.; Lowther, W. T.; Ryu, S. E.; King, S. B.; Furdui, C. M.; Poole, L. B. Use of dimedone-based chemical probes for sulfenic acid detection evaluation of conditions affecting probe incorporation into redoxsensitive proteins. Methods in enzymology 473:77-94; 2010. Liu, J.; Li, F.; Rozovsky, S. The intrinsically disordered membrane protein selenoprotein S is a reductase in vitro. Biochemistry 52:3051-3061; 2013. Salmeen, A.; Andersen, J. N.; Myers, M. P.; Meng, T. C.; Hinks, J. A.; Tonks, N. K.; Barford, D. Redox regulation of protein tyrosine phosphatase 1B involves a sulphenyl-amide intermediate. Nature 423:769-773; 2003.

23

[60] [61] [62]

[63] [64] [65] [66]

[67]

[68] [69]

[70]

[71]

[72] [73] [74]

van Montfort, R. L.; Congreve, M.; Tisi, D.; Carr, R.; Jhoti, H. Oxidation state of the activesite cysteine in protein tyrosine phosphatase 1B. Nature 423:773-777; 2003. Little, C.; Olinescu, R.; Reid, K. G.; O'Brien, P. J. Properties and regulation of glutathione peroxidase. J Biol Chem 245:3632-3636; 1970. Flohé, L. The selenoprotein glutathione peroxidase. In: D. Dolphin, R. P., and O. Avramovic, ed. Glutathione: Chemical, Biochemical, and Medical Aspects - Part A.: John Wiley & Sons, Inc.; 1989: 643-731. Flohé, L.; Günzler, W. A. Glutathione peroxidase. In: Flohé, L.; Benöhr, H. C.; Sies, H.; Waller, H. D.; A., W., eds. Glutathione. Stuttgart: Georg Thieme Verlag; 1974: 132-145. Dalziel, K. Initial steady state velocities in the evaluation of enzyme-coenzyme-substrate reaction mechanisms. Acta Chem. Scand. 11:1706-1723; 1957. Flohé, L.; Toppo, S.; Cozza, G.; Ursini, F. A comparison of thiol peroxidase mechanisms. Antioxid Redox Signal 15:763-780; 2011. Aumann, K. D.; Bedorf, N.; Brigelius-Flohé, R.; Schomburg, D.; Flohé, L. Glutathione peroxidase revisited--simulation of the catalytic cycle by computer-assisted molecular modelling. Biomed. Environ. Sci. 10:136-155; 1997. Takebe, G.; Yarimizu, J.; Saito, Y.; Hayashi, T.; Nakamura, H.; Yodoi, J.; Nagasawa, S.; Takahashi, K. A comparative study on the hydroperoxide and thiol specificity of the glutathione peroxidase family and selenoprotein P. J Biol Chem 277:41254-41258; 2002. Ursini, F.; Maiorino, M.; Gregolin, C. The selenoenzyme phospholipid hydroperoxide glutathione peroxidase. Biochim Biophys Acta 839:62-70; 1985. Baker, L. M.; Poole, L. B. Catalytic mechanism of thiol peroxidase from Escherichia coli. Sulfenic acid formation and overoxidation of essential CYS61. J Biol Chem 278:9203-9211; 2003. Trujillo, M.; Mauri, P.; Benazzi, L.; Comini, M.; De Palma, A.; Flohé, L.; Radi, R.; Stehr, M.; Singh, M.; Ursini, F.; Jaeger, T. The mycobacterial thioredoxin peroxidase can act as a one-cysteine peroxiredoxin. J Biol Chem 281:20555-20566; 2006. Maiorino, M.; Bosello-Travain, V.; Cozza, G.; Miotto, G.; Roveri, A.; Toppo, S.; Zaccarin, M.; Ursini, F. Understanding mammalian glutathione peroxidase 7 in the light of its homologs. Free Radic Biol Med; 2015. Chance, B.; Sies, H.; Boveris, A. Hydroperoxide metabolism in mammalian organs. Physiological reviews 59:527-605; 1979. Wood, Z. A.; Poole, L. B.; Karplus, P. A. Peroxiredoxin evolution and the regulation of hydrogen peroxide signaling. Science 300:650-653; 2003. Morgan, B. A.; Veal, E. A. Functions of typical 2-Cys peroxiredoxins in yeast. Sub-cellular biochemistry 44:253-265; 2007.

Legends to the figures

Fig. 1. GPx catalysis. Figure 1a illustrates the basic scheme of GPx catalysis. The ground state enzyme E is oxidised by a hydroperoxide yielding F, which is stepwise reduced by two thiols via a semi-reduced intermediate G. E represents the reduced enzyme with its peroxidatic (seleno)cysteine in the selenol or thiol form, respectively. In the oxidised intermediate F, these residues have tentatively been regarded as selenenic or sulfenic acids. In G the (seleno)cysteine is thiylated by a thiol substrate. k+1 and k´+2 are the experimentally accessible net forward rate constants for the 24

oxidative and reductive part of the catalytic cycle. Figure 1b shows the energetic profile of the cycle for a SecGPx (●) and the corresponding Cys mutein (▲), as DFT-calculated with a single water molecule present in E. TS1, 2, 3: transition states; CS, charge separated species; RSH: ethane-thiol; RSSR: corresponding disulfide; [E•H2O2], [CS•H2O2], [F•RSH], [G•RSH], [E•RSSR]: complexes of E or intermediates with substrates and product.

Fig. 2. Active site of glutathione peroxidases. The active site representative of ground state GPx, as calculated by DFT, is shown in comparison with the x-ray structure of human GPx4. Root Mean Square Deviation (RMSD) of atomic positions values of the x-ray structure (Pdb 2OBI; white) versus minimized GPx core structures were 1.373, 1.155 and 1.144, when calculated by PBE1PBE/6-31G(d) (yellow), B3LYP/6-31G(d) (pink) and B3LYP/6-31G(d,p) (cyan), respectively. Brown ball, selenium atom. Residue numeration refers to cytosolic rat GPx4.

Fig. 3. Essentials of the DFT-calculated oxidation of SecGPx by H2O2. The calculation was performed with E, as shown in Fig. 2, but with a single water molecule and H2O2 bound to the reaction center. The polar contact network of E is shown as yellow dashed lines. The tetrad residues Sec46, Asn137, Trp136, and Gln81, as well as Gly47 contribute to catalysis. The transition state, leading to the charge-separated form CS, involves three concurrent steps: i) proton transfer from Sec to hydrogen peroxide, ii) proton transfer from hydrogen peroxide to water molecule, iii) proton transfer from water to nitrogen of the tryptophan ring. CS is bearing a negative charge on selenium and a positive one on the ring nitrogen of tryptophan. CS evolves with no energy barrier to the selenenic form F and a water molecules via a concerted mechanism. Formation of the chargeseparated form does not depend on binding of H2O2 but just on the presence of water (see Movie S1). Binding of H2O2 to CS then generates the identical unstable CS•H2O2 complex which decays as described.

Fig. 4. Oxidation GPx1 produces a mass decrease by 2 a.m.u. The left panels (a1 and b1) show the Extracted Ion Current (EIC) over chromatographic retention time (‘Acquisition time’) of a selected multi-charged peak of the primary mass spectrum (MS1; * in a2 and b2) from reduced or oxidized GPx1 respectively. The insets in a1 and b1 show the result of the deconvolution of MS1 spectra. The insets in a2 and b2 show the zoom on the most abundant multi-charged signals (marked as *). The molecular mass of the reduced GPx1 corresponds to the full sequence minus the N-terminal Met-Cys residues plus a mass increment of 43 a.m.u. strongly suggestive of a carbamylation at the N-terminus, which indeed proved to be blocked. An adduct containing one molecule of β25

mercaptoethanol (+75.9983) resistant to reduction was also detected. GPx1 was reduced 5 mM mercaptoethanol, then oxidized with 10 μM H2O2 and separated on a C8 reverse phase nanocolumn. The amount of protein injected was 200 fmoles. Resolution 20.000, accuracy 5ppm.

Fig. 5. Oxidation GPx4 produces a mass decrease by 2 a.m.u. Processing of mass spectra shown in the individual panels correspond to Fig. 4. The molecular mass of the reduced form of GPx4 corresponds to the value expected for the total sequence minus the N-terminal residues Met and Cys. Also here an adduct with one molecule of β-mercaptoethanol (+75.9983) resistant to reduction was detected. GPx4 was reduced with 5 mM mercaptoethanol, then oxidized with 5 μM H2O2 and otherwise processed as GPx1 (Fig. 4). The 922.0098 signal in panels b1 and b2 is the calibration standard.

Fig. 6. DFT-predicted active site structure of F´. Oxidized Sec46 forms a selenenylamide bond with the nitrogen of the peptide bond linking the conserved Gly47 to the following amino acid, which here is calculated as Gly, but is Lys in rat GPx4 and Thr in bovine GPx1.

Fig. 7. Tryptic fragment of GPx1 containing the -2 a.m.u. species. The right part shows the zoom on the 640 – 900 m/z region of the MS/MS spectrum obtained by CID fragmentation of the doubly charged ion 867.44 corresponding to the GPx1 tryptic peptide VLLIENVASLUGTTVR minus 2 a.m.u. Fragmentation ion y6 (682.24) identifies the smallest peptide stretch accounting for -2 a.m.u. comprising the U residue. Table on the left summarizes the complete b and y ion series obtained by MS/MS for the peptide. Matched theoretical values are marked in blue (y) and red (b). Tryptic peptides from the oxidized enzyme were obtained by overnight digestion and analyzed by Q-TOF technology.

Fig. 8. Tryptic fragment of GPx4 containing the -2 a.m.u. species. The right part shows the zoom on the 230 – 360 m/z region of the MS/MS spectra obtained by CID fragmentation of the doubly charged ion 802.32 corresponding to the GPx4 tryptic peptide GCVCIVTNVASQUGK plus one βmercaptoethanol (+ 75.9983 a.m.u) and minus 2 a.m.u. Fragmentation ion y3 (353.16) identifies the smallest peptide stretch accounting for -2 a.m.u. comprising the U residue. Fragmentation ion b3 (336.24) locates the β-mercaptoethanol linked to Cys35, as further confirmed by fragmentation ion b2 (237.12). The table on the left summarizes the complete b and y ion series obtained by MS/MS for each peptide. Matched theoretical values are marked in blue (y) and red (b). Tryptic peptides from oxidized enzymes were obtained by overnight digestion and analyzed by Orbitrap technology. 26

Fig. 9. Oxidized form of the CysGPx4 mutein. Signal deconvolution the reduced CysGPx4 mutein form (a1) matches the expected value (19208.3326 a.m.u.) plus one molecule of β- mercaptoethanol (+75.9983) located on Cys75. MS1 deconvolution of CysGPx4 mutein exposed for 2 min to 5 μM H2O2 (a2) yields a mass shift of + 32 a.m.u. MS/MS sequencing (b) of the tryptic peptide T34-48 (GCVCIVTNVASQCGR) of oxidized CysGPx4 reveals over-oxidation of the catalytic Cys46. MS/MS fragmentation of the corresponding 764.33 [M+2H+]2+ ion confirms the sequence by b(red) and y-fragments (blue) and shows the presence of a disulfide between the non-catalytic cysteines 35 and 37 and the catalytic Cys46 oxidized to sulfonic acid derivative. See Methods for further details.

Fig. 10. Re-integration of F´ into the canonical GPx cycle by GSH. Traces a, b, c, and d show activity measurements (coupled test system; see Methods) of pre-oxidized rat SecGPx4, when started by addition of GSH at concentrations indicated. The first 10 ± 2 sec (*) after this addition could not be evaluated due to stirring artifacts. At reaction start, the enzyme is present as F´ according to the data presented in Fig. 4. A third order polynomial analysis (R2 ≥ 0.99) reveals that the traces can clearly be divided into three distinct phases: a period of gradual resumption of full activity from the beginning of curve analysis up to the green dashed line (“lag phase”); a period of constant turnover between the green and red dashed line (“constant rate”; essentially dependent on GSH concentration, which is kept constant by regeneration); a final decline of the rate due to hydroperoxide consumption. In traces a-c, a substantial lag phase is calculated, which decreases with increasing GSH concentration, becoming undetectable at 1mM (trace d). Necessarily, this calculated lag phase does not take in account the first ~10 seconds after the reaction start. The time to reach constant turnover in trace a-d reflects that required to transform F´ into G of the canonical GPx cycle.

Fig. 11. Conversion of selenenylamide to dehydroalanine upon tryptic digestion. Panel a shows the MS1 deconvolution of oxidized GPx1. The deconvoluted mass at 22386.55 (zoomed in the inset) matches the expected value for the Sec to dehydroalanine conversion (‘SecÆDha’). The relative amount of the -2 a.m.u. species and SecÆDha species, estimated as area under the peak are reported in the bar plot: #, GPx1 bearing Sec Æ Dha; ##, GPx1(-2 a.m.u.). The panels b and c show the extracted ion counts of the tryptic peptide T42-57 (VLLIENVASLUGTTVR) derived from oxidized GPx1, as obtained by digestion in the absence (b) or in the presence (c) of 5 mM βmecaptoethanol (ME). The insets show the relative amounts of the peptides, estimated as area under 27

the peak (ND = not detectable). d, e, f are the MS1 spectra of the three tryptic peptides derived from the U containing peptide in oxidized GPx1: VLLIENVASL(Dha)GTTVR (d), VLLIENVASLU(2)

GTTVR (e) and VLLIENVASLU(ME)GTTVR (f). The isotopic distribution shows the presence of

one selenium atom in e and f. GPx1 was reduced by 5 β-mM mercaptoethanol, then oxidized by 5 μM H2O2, hence analyzed on a C8 reverse phase nano-column linked to MS. Resolution 20.000, accuracy 5-30 ppm for whole protein, 2-5 ppm for peptides.

Highlights xDFT calculations corroborate the catalytic cycle of GSH peroxidases. xIn oxidized GSH peroxidases the Se moiety is stabilized as Se-N bond. xThiolysis of the Se-N bond re-integrates the enzyme into the canonical cycle. xUpon denaturation Se-N-involved Sec loses Se to form dehydroalanine. xThe Se-N bypass of GPx catalysis implies new perspectives of redox regulation.

28

Legends to the Movies (Supplementary) Movie S1 Original DFT-results for formation of a charge-separated form CS of SecGPx from ground state enzyme E with the aid of H2O. a) Active site of E, as calculated without water plus a single water molecule added; b) transition state with main atom movements and distances shown; c) CS with the selenol being deprotonated and the proton shifted via a long-range proton shuttling to the imino-nitrogen of Trp136. An analogous charge separation is observed with the CysGPx mutein, which is in line with the observation that Sec and Cys in the active site of any kind of GPx behave like being dissociated. Movie S2 DFT-results for the water-supported formation of the CSxH2O2 complex from the ExH2O2 complex. a) [ExH2O2]; b) transition state (TS1); c) [CSxH2O2]. As in Movie S1, a longrange proton shuttling is required to yield the unstable complex, which decomposes without energy barrier (see Fig. 1b and 3). It can be formed via the route shown in Movie1 by binding H2O2 to CS or, as shown here, within [ExH2O2]. Thereby the hydroperoxide would contribute to the charge separation and its own destruction. Movie S3 DFT-data for the transition of F to G. a) F plus ethane-thiol used as mimic of the physiological reducing substrate GSH; b) transition state TS2; c) the reaction product G. In TS2 the thiol proton of the substrate is moving via a water molecule to the OH of the oxidized Sec, thus creating a further H2O as ideal leaving group. Simultaneously, the Se-S bond of G is formed. This key event is primarily supported through proton attractions by water and the carboxamide groups of Asn 137 and Gln 81. Movie S4 DFT-data of the regeneration of E from G. a) G plus the second thiol substrate; b) transition state TS3; c) E plus product RSSR. In TS3, the thiol proton of the substrate is mobilized by Trp136 and Gln81 and attacks the selenium of the Se-S bond in G, while the thiolate of the substrate binds to the sulfur. The productive side chain arrangement in TS3 is stabilized by multiple ionic interactions. Movie S5 DFT-data for the formation of the selenenylamide form F´ from the selenenic acid form F of SecGPx. a) Starting situation F. b) Transition state TS4. The proton of the peptide bond downstream of Gly47 has moved to the OH of the selenenic acid to form water and a selenenylamide bond has been formed. The process is facilitated by water-mediated proton shuttling and polar interactions involving Gln81 and Asn137. c) The alternate oxidized SecGPx form F´. Movie S6 DFT-data for the thiolysis of the selenenylamide form F´. a) F´ with ethane-thiol as mimic of the physiological substrate GSH. b) Transition state TS5. Via an interposed water molecule the carbonyl of Gly47 attracts the thiol proton of the substrate. This interaction twists the plane of the peptide bond and stretches the Se-N bond. Finally, the substrate in its thiolate form attacks the Se of the breaking Se-N bond. Again, the process is facilitated by polar interactions involving Gln81, Trp136 and Asn137 as well as backbone carbonyls and water. c) The product of the reaction is the intermediate G, which re-integrates F´ into the canonical GPx cycle. Movie S7 DFT-data for the formation of sulfinic from sulfenic acid form. a) sulfenic acid. b) Transition state TS. Sulfur attacks oxygen of H2O2 splitting the molecule. Oxygen is finally covalently bonded to sulfur and the proton released binds to the leaving water molecule. c) The sulfinic acid product of the reaction.

Figure 1

Figure 1

Figure 2

Figure 2

Figure 3

Figure 3

Figure 4

Figure 4

Figure 5

Figure 5

Figure 6

Figure 6

Figure 7

Figure 7

Figure 8

Figure 8

Figure 9

Figure 9

Figure 10

Figure 10 0.00

*

-0.08

-0.06

b: 0.50mM GSH Lag phase 32 sec

-0.10

c: 0.75mM GSH Lag phase 9 sec. d: 1.00mM GSH Lag phase 0 sec

-0.12

Δ Extinction

-0.04

-0.02

a: 0.25mM GSH Lag phase 62 sec

0

50

100

150

Time (sec)

200

250

Figure 11

Figure 11

Graphical Abstract (for review)

DFT-based cartoon demonstrating the reversible formation of a covalent bond between selenium, the element of the moon (red ball), with a downstream peptide nitrogen (blue ball) in the active site of oxidized glutathione peroxidases.