Biochimica et Biophysica Acta 1830 (2013) 3846–3857

Contents lists available at SciVerse ScienceDirect

Biochimica et Biophysica Acta

journal homepage: www.elsevier.com/locate/bbagen

Protein disulfide and glutathione are alternative substrates in the one Cys catalytic cycle of glutathione 7

Valentina Bosello-Travain a, Marcus Conrad b, Giorgio Cozza c, Alessandro Negro c, Silvia Quartesan a, Monica Rossetto a, Antonella Roveri a, Stefano Toppo a, Fulvio Ursini a, Mattia Zaccarin a, Matilde Maiorino a,⁎ a Department of Molecular Medicine, University of Padova, Viale G. Colombo, 3, I-35121-Padua, Italy b Institute of Developmental Genetics, Deutsches Zentrum für Neurodegenerative Erkrankungen and Helmholtz Zentrum München, Ingolstädter Landstrasse 1, D-85764 Munich-Neuherberg, Germany c Department of Biomedical Sciences, University of Padova, Viale G. Colombo, 3, I-35121-Padua, Italy article info abstract

Article history: Background: Mammalian GPx7 is a monomeric of the endoplasmic reticulum (ER),

Received 15 November 2012 containing a Cys redox center (CysGPx). Although containing a peroxidatic Cys (CP) it lacks the resolving Received in revised form 11 February 2013 Cys (CR), that confers fast reactivity with thioredoxin (Trx) or related proteins to most other CysGPxs. Accepted 19 February 2013 Methods: Reducing substrate specificity and mechanism were addressed by steady-state kinetic analysis of wild Available online 27 February 2013 type or mutated mouse GPx7. The were heterologously expressed as a synuclein fusion to overcome limited expression. Phospholipid hydroperoxide was the oxidizing substrate. –substrate and protein– Keywords: protein interaction were analyzed by molecular docking and surface plasmon resonance analysis. Kinetics 3 −1 −1 ′ Glutathione peroxidase Results: Oxidation of the CP is fast (k+1 >10 M s ), however the rate of reduction by GSH is slow (k +2 = −1 −1 Peroxide 12.6 M s ) even though molecular docking indicates a strong GSH–GPx7 interaction. Instead, the oxidized 3 −1 −1 Redox switch CP can be reduced at a fast rate by human protein disulfide isomerase (HsPDI) (k+1 >10 M s ), but not by

Redoxin Trx. By surface plasmon resonance analysis, a KD =5.2μM was calculated for PDI–GPx7 complex. Participation of an alternative non-canonical CR in the peroxidatic reaction was ruled out. Specific activity measurements in the presence of physiological reducing substrate concentration, suggest substrate competition in vivo. Conclusions: GPx7 is an unusual CysGPx catalyzing the peroxidatic cycle by a one Cys mechanism in which GSH and PDI are alternative substrates. General significance: In the ER, the emerging physiological role of GPx7 is oxidation of PDI, modulated by the amount of GSH. © 2013 Elsevier B.V. All rights reserved.

1. Introduction Selenoperoxidases (SecGPxs) are typical of vertebrates. The catalytic moiety selenium is incorporated as selenocysteine, the ‘peroxidatic’

Non-heme include two non-homolog families that have selenocysteine, UP- in the growing peptide chain [2].Themembersof in common the typical thioredoxin (Trx) fold and ping-pong catalytic this subfamily encompass both monomeric and tetrameric members, mechanism. Thus, glutathione peroxidases (GPxs), using either seleni- typically using glutathione (GSH) as the electron donor. Kinetics sug- um or sulfur as their redox center, and (Prdxs), which gested that the UP selenol, upon oxidation by a hydroperoxide (ROOH), only use sulfur in their redox centers [1], have major similarities. is oxidized to selenenic acid (Eq. (1)), in turn reacting with GSH and forming a mixed seleno-disulfide (Eq. (2)). The latter is reduced by a sec- ond GSH, releasing GSSG and regenerating the catalyst (Eq. (3)) [3,4]: Abbreviations: CysGPxs, subfamily of glutathione peroxidases containing a Cys redox center; 1CysGPx, a glutathione peroxidase containing the CP only; 2CysGPxs, glutathione ROOH þ HSe⋯ðÞSecGPx →ROH þ HOSe⋯ðÞSecGPx ð1Þ peroxidases containing both the CP and the CR;CP, peroxidatic Cys; CR, resolving Cys; GPx7, glutathione peroxidase 7; ER, endoplasmic reticulum; GSH, reduced glutathione; GSH þ HOSe⋯ðÞSecGPx →H O þ GS–Se⋯ðÞSecGPx ð2Þ GSSG, oxidized glutathione; GPxs, glutathione peroxidases; PDI, protein disulfide isomer- 2 ase; PCOOH, phosphatidylcholine hydroperoxide; Prdxs, peroxiredoxins; ROOH, hydro- peroxide; Sec, Selenocysteine; SecGPxs, subfamily of glutathione peroxidases containing GSH þ GS–Se⋯ðÞSecGPx →GSSG þ HSe⋯ðÞSecGPx : ð3Þ a Sec redox center; Syn, synuclein; Trx, thioredoxin; TrxR, thioredoxin reductase; UP, peroxidatic Selenocysteine In contrast, in invertebrates and the plant kingdom, the sub-family ⁎ Corresponding author at: Department of Molecular Medicine, Viale G. Colombo 3, I-35121 Padova, Italy. Tel.: +39 049 8276103; fax: +39 049 8073310. of sulfur-containing peroxidases (CysGPxs) encompasses almost ex- E-mail address: [email protected] (M. Maiorino). clusively monomeric proteins containing a peroxidatic Cys residue

0304-4165/$ – see front matter © 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.bbagen.2013.02.017 V. Bosello-Travain et al. / Biochimica et Biophysica Acta 1830 (2013) 3846–3857 3847

(Cp) in place of the UP. Here the CP is accompanied by an intra-chain Cys residue located within the so-called ‘Cys block’ (Fig. 1), within the

α4 helix of the thioredoxin fold. This second Cys participates in the re- action as the ‘resolving Cys’ (CR). Hence invertebrate and plant CysGPxs can be referred to as ‘2CysGPxs’. The CR endows reactivity with some redoxins, i.e. the Trx or Trx-related proteins, while muta- tion of the CR, abolishes the peroxidase activity [5]. The reaction mechanism of invertebrate and plant 2CysGPxs is sim- Briefly, in 2CysGPx and atypical Prdx, an intra-chain disulfide in- ilar to that of atypical Prdxs [1,6].TheCP, oxidized to the sulfenic acid derivative upon reaction with ROOH, reacts with the C ,toforman termediate is produced upon reaction with the hydroperoxide that R fi intra-chain disulfide (Eqs. (4) and (5)). This disulfide intermediate is re- is then ef ciently reduced by the CXXC motif of the redoxin. In the duced by a CXXC motif of the redoxin in two steps: the N-terminal Cys peroxidases relying on Sec catalysis, typically lacking the CR, the reac- tion with the hydroperoxide generates a selenenic acid intermediate, of the CXXC motif attaches the CR of the peroxidase forming a mixed di- sulfide (Eq. (6)), which is displaced by the C-terminal Cys of the CXXC which is reduced by GSH in two steps. This, seemingly, descends from motif of the redoxin (Eq. (7)) [5,7]: the peculiar reactivity of selenium. The recent discovery of rare Prdxs and GPxs, containing the CP but missing the CR residue, hence referred to as 1CysPrdx and 1CysGPxs, challenges this paradigm. þ ⋯ðÞ⋯ → þ ⋯ðÞ⋯ ð Þ ROOH HS CPCysGPxCR SH ROH HOS CPCysGPxCR SH 4 In mammalian cells, two monomeric 1CysGPxs (GPx7 and GPx8) have been described in endoplasmic reticulum (ER) [8,9]. Phylogeny suggests that they represent a branch in the evolution of the monomeric SecGPx4. Given that the common ancestor of GPxs is a Cys-containing enzyme, the evolution of GPx7 and GPx8 from SecGPx4 indicates a re- cent, intriguing reversion to usage of Cys in the catalytic center [10].

Fig. 1. Alignment of different GPx7 homologues with other GPxs. Human GPx1 and 4 have been taken as examples for SecGPxs, Drosophila melanogaster GPx and yeast GPx3 for

2CysGPxs. The peroxidatic residue of GPxs (either Cys or Sec) is indicated by an arrow. The ‘Cys block’ within the α4 helix of the dimer interface, where the CR is usually located in the 2CysGPxs, is indicated. Note that the CR is present in the DmGPx and the yeast GPx3 sequences only. The putative signal peptide, the conserved PCNQF motif, and the ER retention sequence of GPx7 are boxed. Accession numbers are the following: NP_056511.2, for human GPx7 (HsGPx7); NP_077160.1, for Mus musculus GPx7 (MmGPx7); NP_001088904.1, for Xenopus laevis GPx7 (X. laevis GPx7); AAF47761.1, for the GPx of Drosophila melanogaster (DmGPx); P40581.1, for yeast GPx3; NP_000572.2, for human GPx1 (HsGPx1). Sequence identity score between HsGPx7 and MmGPx7 is 90%. 3848 V. Bosello-Travain et al. / Biochimica et Biophysica Acta 1830 (2013) 3846–3857

According to the above paradigm both, GPx7 and GPx8 are was induced by adding 1 mM isopropyl-b-D-thiogalactopyranoside expected to not accept electrons from a redoxin. Yet, it has been re- (IPTG) and cells were harvested after 3 h by centrifugation at 5000 ×g cently suggested that they react faster with protein disulfide isomer- for 30 min. The pellet was stored at −80 °C for no longer than one ase (PDI) and other PDI family members, than GSH [9]. In that study week or immediately lysed. For lysis, bacterial pellets obtained from however, the assay with PDI was carried out in the presence of GSH, 1 l of culture were suspended by 30 ml of cold lysis buffer (0.1 M thus preventing the precise assessment of the actual capability of Tris–HCl, pH 7.4, 5% Nonidet, 0.15 M KCl, 0.1% Triton X-100, 3 mM the peroxidase to use PDI directly as a reducing substrate. Moreover, GSH containing protease inhibitors (–0.1 mg/ml PMSF, 0.7 mg/ml an unrealistically low second order rate constant, deduced from the pepsatin, 0.5 mg/ml leupeptin – for 15 min at 4 °C, under slow agita- specific activity, was reported. In this study, we specifically address tion. After centrifugation at 29,000 ×g for 30 min at 4 °C, the superna- the analysis of the kinetics of the catalytic cycle of GPx7, aiming to tant was immediately used as enzyme source for activity/kinetic get information on the rate constant of the oxidative and reductive analysis. Protein was quantified by the Bradford assay [12]. steps of the cycle that are relevant for gaining an insight into the Quantitation of (Syn)GPx7 or (Syn)GPx7C85A within the 29,000 ×g physiological function of the enzyme. supernatant from induced bacterial lysate was obtained by western blotting (see below) using known amounts of purified (Syn)GPx7 for 2. Material and methods the standard curve and polyclonal antibody to human GPx7 obtained from Abnova (Heidelberg, Germany). To this end, the bacterial extract 2.1. RT-PCR analysis expressing (Syn)GPx7 was purified by nickel affinity chromatography (Ni-NTA, Qiagen, Hilden, Germany), followed by a Superdex 75 column Detection of GPx7 in human tissues was performed by PCR using a (26 mm × 620 mm) as described [5]. This procedure yielded the per- human cDNA panel as template (Clontech laboratories, Inc., Palo Alto, oxidase in its inactive form. A calibration curve was obtained from the CA, USA). QuantumRNA classic II 18S (Ambion Inc., Austin, TX, USA) digitized read-out of western blot spots. was used for internal control. GPx7 primers were the following: fw: Human recombinant PDI (HsPDI) was obtained as indicated above, ACTGGTGTCGCTGGAGAAGT rev: GTCTGGGCCAGGTACTTGAA. Forty after transforming the E. coli JM 109 strain by the PQE30/HsPDI con- PCR cycles were applied at the annealing temperature of 58 °C. struct. Bacterial pellets obtained from 1 l of culture were suspended by 60 ml of cold B-Per extraction reagent (Pierce, Rockford, IL, USA) 2.2. Plasmid constructs containing protease inhibitors as above, and stirred for 10 min on ice to obtain cell lysis. After centrifugation (20,000 ×g, 30 min), the The expression vector containing chimeric mouse GPx7 (MmGPx7) supernatant was diluted 1:1 (v/v) with 50 mM Tris–HCl, 1 M NaCl, (accession number NM_024198) fused to human synuclein–pRSET/ 20% (v/v) glycerol, 0.2% (v/v) Triton X-100 (pH 7.4), purified by (Syn)GPx7-, was obtained by conventional cloning of the PCR product Ni-NTA chromatography and size exclusion chromatography as de- obtained using a cDNA from mouse liver tissue using the following scribed, except that the reductant was omitted [5]. The protein was primers fw: GCGGATCCCAATCCGAGCAGGACTTCTACG; rev: GTGGAA then concentrated and quantified as above [12]. HsPDI is recovered TTCCACAAGTCTTTCGTTTCCGAAG. After digestion, the PCR product in its disulfide form after this procedure. was cloned in the BamHI and EcoRI sites of the plasmid pSynA, prepared Human thioredoxin, having the wild type sequence except for a C72S as described below. The product of this vector is (Syn)GPx7. The pRSET Cys72 to Ser mutation to prevent dimerization (HsTrx ) [13], was plasmid encoding for human alpha-synuclein (pSynA) has been de- obtained by recombinant expression and purification as described [5]. scribed [11]. A modified nucleotide sequence without the inner restric- tion sites BamHI and EcoRI encoding the same protein sequence was 2.4. Western blotting obtained by site-directed mutagenesis. The stop codon was replaced by a nucleotide linker sequence encoding for a thrombin cleavage site For quantifying (Syn)GPx7 in the supernatant from the induced bac- (ASLVPRGS amino acid sequence) and by nucleotide sequence con- terial lysate, western blotting was performed as previously described taining the restriction sites BamHI and EcoRI in order to make the fusion [14] using commercially available polyclonal antibody to GPx7 (Abnova, vector pSynA encoding for GPx7-pSynA/(Syn)GPx7-. Heidelberg, Germany), and revealed using - (Syn)GPx7 was mutated at the residue Cys 85 of the MmGPx7 C85A conjugated secondary antibody (Santa Cruz Biotechnology, Santa Cruz, sequence to yield (Syn)GPx7 by PCR, using pRSET/(Syn)GPx7 as CA, USA) and a freshly prepared luminol solution (1.1 mM luminol, template and iProof High Fidelity Polymerase turbo (Bio-Rad Laborato- 1 mM 4-jodophenol, 0.12% BSA, 1.4 M H2O2, 0.025 mM Tris–HCl ries, Veenendaal, Holland) using the following primers (changed co- pH 9.25). Densitometric analysis of immunoblots was obtained using dons in italics): fw: TAATGTGCTTGCCTTCCCTGCCAACCAGTTTGGCCAA an Image Station 440 (Kodak, Rochester, NY, USA). CAGG; rev: CCTGTTGGCCAAACTGGTTGGCAGGGAAGGCAAGCACATTA. HsPDI was obtained by cloning in the multicloning site (BamHI/ KpnI) of the pQE30 vector (Qiagen, Hilden, Germany) the PCR product 2.5. Peroxidase activity assay and kinetic analysis obtained from human cDNA by the primers The coupled assay for glutathione peroxidases was used for mea- fw: ATAAGGATCCGACGCCCCCGAGGAGGAGGAC surements of peroxidase activity with different substrates and for kinet- rev: ATTAGGTACCTTACAGTTCATCTTTCACAGCT (NCBI Reference ic analysis. A single-beam Beckman DU70 spectrophotometer was used. Before the assay, the buffer of the supernatant from the induced bacte- Sequence: NP_000909.2), which yielded the plasmid PQE30/HsPDI. rial lysate expressing (Syn)GPx7/Syn/(Syn)GPx7C85A was exchanged Correctness of constructs was verified by sequencing. with the assay buffer (0.1 M Tris–HCl, pH 7.6 containing 5 mM EDTA,

1mMNaN3, 0.1% Triton X-100) by a NAP-5 column chromatography, 2.3. Heterologous expression, preparation of the bacterial extract and repeated twice (GE healthcare, Buckinghamshire, UK). quantification of recombinant enzymes For GSH peroxidase activity, 3 mM GSH was then added, together with 0.15–0.3 mM NADPH and 3 U/ml glutathione reductase and vol- For production of recombinant (Syn)GPx7 or (Syn)GPx7C85A,com- ume adjusted to 2.5 ml with the assay buffer. Reaction started by petent BL21 (DE3) pLysS Escherichia coli cells, which allows high- 20–25 μM, phosphatidylcholine hydroperoxide (PCOOH). Reaction efficiency protein expression of under the control of a T7 promot- was recorded until complete consumption of substrate. PCOOH dis- er, were transformed with the vector pSynA/(Syn)GPx7. Expression persion was prepared and quantitated as described [15]. V. Bosello-Travain et al. / Biochimica et Biophysica Acta 1830 (2013) 3846–3857 3849

Trx or PDI peroxidase activities were determined as above by re- used a direct computational approach to give an account for the interac- placing GSH with 5.0 μM HsTrxC72S or 10 μM HsPDI, respectively, tion between enzyme and GSH (see below) and the calculation of the ki- and glutathione reductase with a nonlimiting amount of Plasmodium netic coefficient Φo was not specifically addressed. falciparum thioredoxin reductase (PfTrxR) prepared as described [5], or a non-limiting amount of rat liver thioredoxin reductase (RrTrxR) 2.6. Identification of the adduct between the CP of GPx7 and GSH (Sigma-Aldrich, Saint Louis, Missouri, USA) [16]. A non-limiting amount is an appropriate amount of ancillary enzyme whose activity is at least The supernatant from the induced bacterial lysate, containing one order of magnitude higher than the activity under scrutiny. This approx. 60 μg of (Syn)GPx7, obtained as described above was reduced corresponded to 3 U/ml for PfTrxR and 4 U/ml for RrTrxR. The extinc- by 1 mM dithiotreitol (DTT) on ice. After 30 min, the buffer was ex- tion coefficient of 6220 M−1 cm−1 was used for calculations [17]. changed to 0.1 M Tris–HCl pH 7, containing 1 mM EDTA, 0.1% v/v Triton HsTrxC72S or HsPDI concentration was routinely calculated spectro- X-100, 1 mM GSH in two consecutive passages through Micro Bio-Spin photometrically, recording, as a function of time, the fast phase of chromatography columns (Bio Rad Laboratories, Inc). To the recovered NADPH oxidation following a definite amount of Trx or PDI addition in sample, approximately 2 μM PCOOH was added. After 10 min of incu- the presence of PfTrxR or RrTrxR, respectively. One mole of HsTrxC72S bation at RT, the reaction ended by adding 25 mM N-ethyl maleimide contains one CXXC center, thus 1 mol of NADPH reduces 1 mol of oxi- (NEM) to avoid thiol scrambling. The sample then underwent acidic dized HsTrxC72S. One molecule of HsPDI contains two reducible CXXC precipitation, tryptic digestion and MS/MS analysis as reported below. centers, thus 2 mol of NADPH reduces 1 mol of oxidized HsPDI. This quantitation procedure was previously validated by titrating redoxin 2.7. MS/MS analysis thiol groups by reaction with 5,5′-dithiobis-2-nitrobenzoic acid (DTNB) and monitoring at 412 nm [16]. After NEM addition, samples were incubated for 30 min and Kinetic analysis was carried out by analyzing the single progression subjected to multiple acidic precipitation with 0.01% (w/v) sodium curve of NADPH oxidation, as already reported for GPx4 or DmGPx deoxycholate (DOC) and 6% (v/v) trichloroacetic acid (TCA). Tryptic [5,18,19].Briefly, the digitized absorbance data vs.time,producedby digestion of ethanol washed sample pellets was performed overnight the spectrophotometer, were used to evaluate the substrate concentra- at 37 °C in the presence of 40 mM ammonium bicarbonate buffer tion and the reaction rate at each time interval (typically from 1 to pH 8.0, containing 1 mM EDTA, with a protease to protein ratio of 3min).Thereactionratewascorrectedforthenon-specificGSH/HsPDI 1:50. After stopping digestion by formic acid acidification, samples oxidation rate. Non-specific PCOOH reduction in the absence of cell ex- were dried by vacuum centrifugation and suspended in 0.1% (v/v) tract was negligible and thus not considered. The data obtained were formic acid aqueous solution for MS analysis. Peptides mix from fitted to the simple version of the Dalziel equation [20],which,for each digested sample were separated before ionization by reverse GPxs, can be written as follows: phase chromatography on a 1200 Rapid Resolution system equipped with a nanofluidic chip directly interfaced with a nanoESI source = ¼ Φ þ Φ =½þΦ =½: ð Þ E vo 0 1 ROOH 2 RSH 8 from the 6520 Accurate-Mass Q-TOF LC/MS mass spectrometer (Agilent Technologies, Santa Clara, CA, USA). Separation was achieved Where [ROOH] and [RSH] represent the molar concentration of hy- by a 3% to 70% acetonitrile gradient onto a C18 Zorbax capillary col- droperoxide and thiol respectively. The Φ1 and Φ2 parameters were umn at a flow of 300 nl/min. MS acquisition was performed from obtained by double reciprocal plot, in which the reciprocal velocities 350 to 2400 m/z at a scan rate of 2 spectra/s, while MS/MS analysis were multiplied by enzyme molarity [18–20], as the slope and the inter- was performed from 59 to 3000 m/z at the same scan rate in a data- cept, respectively. Φ1 represents the reciprocal value of the rate constant dependent way on the four most abundant precursor in each MS for the first enzyme oxidation reaction (Eqs. (1) and (4) and Scheme 1) scan. Collision Induced Dissociation (CID) was used for selected pre- and Φ2 represents the reciprocal value of the sum of the rate constants of cursor ion fragmentation given a function for collision energy of the two reducing half reactions (Eqs. (2–3) and (6–7) and Scheme 1). 3.7 V/(100 Da) with an offset of 2.5 V. Obtained MS and MS/MS data

The value of Φ0, which is equivalent to the reciprocal of the maximum were processed by the MassHunter Workstation — Qualitative Analy- turnover number, is 0 in the majority of GPxs so far studied [5], in agree- sis (B.02.00) suite (Agilent Technologies, Santa Clara, CA, USA). Suit- ment with non-saturation kinetics where substrate transformation is able custom database was developed for use with Mascot release 2.3 much faster than enzyme substrate interaction. In our kinetic analysis, (www.matrixscience.com) merging the entire bacterial proteome the value of Φ0 calculated for GSH resulted > 0, suggesting a saturation with sequence from GPx7 fusion construct. Tolerance of ±5 ppm kinetics in respect to GSH. To better describe this aspect, in this study we and ±0.05 Da was considered respectively for precursor and frag- ment ions database matching. Identified peptides containing Cys from GPx7 were compared using in house tuned LabKey Server v11.2. ROOH ROH RSH H 2 O 2.8. Molecular modeling GPx(red) GPx(ox) GPx-SR

k+1 k+2 Human GPx7 structure was retrieved from the PDB (PDB code:

k+3 2P31) and processed in order to remove water molecules. Hydrogen atoms were added using standard geometries to the protein structure by the MOE program [21]. To minimize contacts between hydrogens, the structures were subjected to Amber99 force field minimization until the rms of conjugate gradient was b0.1 kcal mol−1 Å−1 keeping k + k = +2 +3 +2 the heavy atoms fixed at their crystallographic positions [21]. GSH

Scheme 1. General reaction pathway of GPxs describing the enzyme-substitution structure was built using MOE-builder tool, part of the MOE suite mechanism. The rate constant k+1 is equivalent to reciprocal of the value of the [21] and was subjected to MMFF94x energy minimization until the − − slope in double reciprocal plot, where 1/ROOH is plotted vs. [E]/v, which yields the rms of conjugate gradient was 0.05 kcal mol 1 Å 1. Charges were Φ ′ term 1 in the Dalziel Eq. (8). The rate constant k +2 represents the sum of the rate calculated using ESP methodology. Three different programs have constants of the two reducing half reactions (i.e. k + k ) and is equivalent to the +2 +3 been used: MOE-Dock [21], GOLD [22] and Glide [23]. The estimated reciprocal of the intercept in the same plot, yielding Φ2 in the Dalziel Eq. (8). RSH and R′SH can be either two molecules of GSH or the N-terminal and C-terminal SH of association constant (pK), was calculated using MOE suite, expressed the CXXC motif of the redoxins. See also Eqs. (1–3) and (4–7). as the summation of a directional hydrogen-bonding term, a 3850 V. Bosello-Travain et al. / Biochimica et Biophysica Acta 1830 (2013) 3846–3857 directional hydrophobic interaction term, and an entropic term. The In the human tissues investigated, GPx7 specific primers detected center of the docking box or of the docking sphere was set at Cys 57 the transcript. Notably the transcript was observed also in the liver, with a radius of 12 Å. where, in the mouse, it was not previously detected [8] (Fig. 2).

3.2. Heterologous expression of GPx7 2.9. Surface plasmon resonance (SPR) analysis of the GPx/PDI interaction Kinetic analysis of the reaction of GPx7 requires an amount of en- (Syn)GPx7 was purified and digested by thrombin to obtain pure zyme that can only be obtained by heterologous expression. In our MmGPx7, lacking the Syn fusion. To this end, (Syn)GPx7 was purified hands, the expression of GPx7 in E. coli by the usual pQE30 expression by Ni-NTA [5] and digested by thrombin, and further Ni-NTA purified to system yielded an amount of enzyme too low for kinetic studies. To remove Syn. This procedure yielded homogeneous MmGPx7. After ex- overcome this limit, we expressed GPx7 as a fusion protein with changing protein buffer into 20 mM Tris–HCl, 100 mM NaCl, 1 mM synuclein (Syn), a methodology previously adopted successfully in MgCl , 10 units of thrombin (Sigma) per milligram of protein was 2 our laboratory to prepare otherwise poorly expressed proteins. added and incubated overnight at room temperature. Reaction was Human Syn was fused with the N terminus of mouse GPx7, devoid stopped by adding protease inhibitors. Purified GPx of Drosophila of the signal peptide by the pSynA expression vector, which produces melanogaster, DmGPxwt,oritsC mutant DmGPxC91K [5] were subjected R N terminal His-tagged proteins (pSynA/(Syn)GPx7) (Fig. 3A). This to SPR analysis to quantify the kinetic constants for PDI binding using a strategy led to the required high yield of the chimeric peroxidase Biacore T100 biosensor system (BIAcore, Upsala Sweden). To this end, pu- (Syn)GPx7, which resulted in it being the most abundant protein in rified HsPDI was bound to the sensor surface (CM5 sensor chip) utilizing the bacterial supernatant (approx. 65% of the total proteins) (Fig. 3B). the reaction between amino groups of the protein and the carboxyl Although bacterial extract exhibited a peroxidase activity with groups of the sensor chip. Following the activation of the latter with GSH suitable for kinetic studies, purification by Ni-NTA affinity chro- 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide/N-hydroxysuccinimide matography caused an irreversible loss of activity. The most likely hy- (EDC/NHS) (0.2/0.05 M), a 10 μl/min constant flux of HsPDI (30 μg/ml in pothesis for this loss is an oxidative inactivation of the , 10 mM acetate at pH 4.0) was injected in the flow cell, until the immobi- which was examined by MS/MS analysis. In the purified protein, the lization level of 600 RU was reached. Ethanolamine (1 M, pH 8.5) was molecular weight of the C increased in a variable, but always rele- used for the deactivation of excess reactive carboxyl groups. For blank P vant percentage, of 32 or 48 amu (Fig. 4), demonstrating that the subtraction, the reference flow cell was activated with EDC/NHS and lack of activity was due to an artifactual oxidation of the C to sulfinic then deactivated with ethanolamine. Before proceeding with protein– P and sulfonic acid derivatives. The oxidative lability of C , which was protein interaction experiment, the HsPDI modified sensor chip was P also observed in the presence of metal chelating agents and reduc- equilibrated with the running buffer HBS-EP + (10 mM Hepes pH 7.4, tants such as mercaptoethanol or GSH, discouraged us from trying 150 mM NaCl, 3 mM EDTA, 0.05% Biacore surfactant P20) at flow rate of other chromatographic approaches. 30 μl/min. In addition, 30 μl/min of 50 mM DTT was injected 10 min be- fore each daily experimental set. Kinetic analysis of the binding was 3.3. Kinetics of GSH peroxidase activity performed by successive injections of different concentrations of each pu- rified enzyme (ranging 0.10–40 μM), diluted in HBS-EP+. Regeneration Previous studies suggest that monomeric GPxs reduce both of the sensor chip after each binding events was obtained with 0.1% SDS. PCOOH and H2O2 at similar rates [18]. Thus both substrates are suit- able substrates for kinetic analysis of chimeric GPx7. Nonetheless, as 3. Results in this study a non-homogeneous enzyme was used, the specificity of the assay required validation. The complete absence of any perox- 3.1. GPx7 structural organization and expression in human tissues idase activity with PCOOH in the bacterial extract, transformed with a control plasmid, eliminated the possibility of competing activities

The primary structure of GPx7 from different species, aligned with (Fig. 5A). Competing activity was however, observed when H2O2 that of homologous GPxs, shows the conserved CP in the expected posi- tion and the absence of the canonical CR that is almost invariably found in other members of the CysGPx subfamily (Fig. 1). As in the latter, the tetramerization interface in GPx7 is absent, consistently with a mono- meric structure. Peculiar to GPx7 however, is the N-terminal stretch that is a putative signal peptide typical of the proteins of the secretory pathway, as computationally predicted (Signal Pv3 program (http:// www.cbs.dtu.dk/services/SignalP/)). A C-terminal (K)REDL (KKEEL in Xenopus laevis GPx7) motif is also present, which suggests ER retention. The latter, however, does not fully match the annotation in Prosite ([KRHQSA]-[DENQ]-E-L) for ER resident proteins [24,25]. Although ini- tially reported to be in a cytosolic location [8], GPx7 was subsequently described as an ER protein by immunohistochemical studies [9].It was also shown that deletion of the REDL motif moves the protein from the ER to the Golgi [25]. In this study we provide further evidence for ER localization of mature GPx7, by demonstrating that both the N-terminal signal pep- tide and the C-terminal KREDL motif of mammalian GPx7 are func- tional (Supplementary Fig. 1). In fact, the putative N-terminal signal peptide is cleaved off in the mature protein, indicating that GPx7 fol- Fig. 2. Human tissues expressing GPx7. PCR analysis was performed on cDNAs from lows the secretory pathway (Supplementary Fig. 1A). Furthermore, human tissues using specific GPx7 primers. 18S was detected as internal standard. Lanes are as follows: 1: heart, 2: brain, 3: liver, 4: placenta, 5: skeletal muscle. M.W., deletion of the KREDL motif, allows mature GPx7 to partially escape molecular weight markers. The expected length of GPx7 and 18S is 298 and 320 bp, re- the cell, demonstrating that the KREDL motif is involved at least par- spectively. Since 40 PCR cycles were applied, no judgment of tissue-specific expression tially in retaining GPx7 within the cell (Supplementary Fig. 1B). is possible. Represented is one of two replicates. V. Bosello-Travain et al. / Biochimica et Biophysica Acta 1830 (2013) 3846–3857 3851

A Table 1. The active site of the GPx7 is oxidized by PCOOH with a 3 −1 −1 rate constant (k+1 = 9.5 × 10 M s ) approximately three or- ders of magnitude faster than the non-catalyzed oxidation of a thiolate by a hydroperoxide [27]. The rate constant for the reductive −1 −1 steps, instead, is quite low (k′+2 = 12.6 M s ), in agreement with that calculated for the reductive step of artificial 1CysGPxs anal-

ogous to GPx7, namely the Cys mutant of SecGPx4 and the CR mutant of the 2CysGPx of Drosophila (DmGPxC91K) [5,19]. Notably, this rate B constant is in the range of the non-enzymatic thiol–disulfide ex- change reaction [28].

3.4. Molecular docking of GSH to GPx7

In light of the low rate of reductive part of the cycle, we questioned whether GSH might be considered as a “real” GPx7 substrate. This issue was addressed by a computational analysis of the interaction between GSH and GPx7 in comparison with other GPxs. The molecular docking pose for the complex GSH–GPx7 indicates that GSH is bound to GPx7 in a manner similar to GPx1 and GPx4, and that the interaction is even stronger (Fig. 6) [1]. In the case of GPx7, the electrostatic network be- tween GSH negative charges and Arg 180/Arg 52 (GPx1 human num- bering), Lys 135/Lys 148 (GPx4 human numbering), is replaced by the interactions with His 63 and Thr 162 and a relevant hydrophobic net- work (provided by Pro 161, and in particular by Phe 59). This arrange- ment provides an important contribution to the complex stabilization. The association constant (pK) of GSH indicates that the GSH–GPx7 com- plex is even more stable than the GSH-GPx1 and GSH-GPx4. The hydro- phobic term (hyd) contribution in the association constant appears to be critical for the strength of interaction. All together, kinetic and docking data indicate that GSH interacts with GPx7 even though the rate of the overall peroxidatic reaction is limited. This most likely occurs at the thiol–disulfide exchange re-

Fig. 3. The pSynA/GPx7 vector containing the fusion (Syn)GPx7 and heterologous ex- action and/or the release of GSSG (Scheme 1). pression thereof. A) Scheme of the pSynA/GPx7construct. It contains an inducible T7

RNA polymerase promoter, and encodes for an N terminal His6 tag fused to human 3.5. Kinetics of PDI peroxidase activity α-synuclein and mouse GPx7, which lacks the putative N-terminal signal peptide. The asterisk indicates the thrombin cleavage site. B) SDS-PAGE analysis of the Measuring the peroxidatic reaction of chimeric GPx7 by the reducing 29,000 ×g supernatant from induced bacterial lysate. Lanes are as follows: 1, 2 μlofa 1:50 dilution of the 29,000 ×g supernatant from the induced bacterial lysate express- substrate HsPDI, 4 U of the mammalian RrTrxR was used as non ing (Syn)GPx7; MW, molecular weight markers. The amount of (Syn)GPx7 was quan- rate-limiting ancillary enzyme in the assay. In fact, the P. falciparum tified by densitometric analysis and estimated as 65% of the total supernatant proteins. TrxR, used for tracking HsTrxC72S oxidation [5], does not reduce PDI Typical yield is around 40 mg of (Syn)GPx7 per liter of bacterial culture. [16]. As for the GSH peroxidase activity measurement, the use of non- homogeneous enzyme and PCOOH in this reaction, eliminated the pres- was the hydroperoxide substrate (not shown). Thus PCOOH fulfilled ence of any competing activity satisfying Racker's rule [26] (not shown). Racker's rule dictating that there must be no more than one enzyme Steady-state kinetic treatment of the data showed that the oxi- 3 −1 −1 activity on with the same substrate in the sample [26]. dized (Syn)GPx7 oxidizes HsPDI (k′+2 = 3.5 × 10 M s ) orders The unlikely hypothesis that recombinant GPx7 could react with of magnitude faster than it does GSH (Table 2). As expected, the cal- an E coli redoxin and that a redoxin-catalyzed GSH oxidation would culated k+1 value was very close to that measured when GSH was then take place could not be excluded in principle. the reductant (Tables 1 and 2). Furthermore, no activity was detected It was however, falsified by the observation that reactivity with con- in the presence of HsTrxC72S. taminating proteins in crude samples, was never observed even for Consistently with kinetic data, despite the different reactivity of those enzymes for which reactivity with a ‘redoxin’ was positively dem- (Syn)GPx7 with GSH or HsPDI, the specific activity is indeed quite simi- onstrated (e.g. DmGPx and/or peroxiredoxins). Moreover, the identifi- lar in the presence of either 3 mM GSH or 10 μM HsPDI (Tables 1 and 2). cation of the addition product between the CP of the chimeric GPx7 and GSH demonstrated a direct enzyme–substrate interaction (Fig. 5B). 3.6. The reaction mechanism of GPx7 does not involve any Cys residue

The steady-state kinetic analysis of (Syn)GPx7 was carried out, as besides the CP previously reported for GPx4 and DmGPx [5,18,19], using data on reac- tion rate and hydroperoxide concentration calculated from digitized Absence of the canonical CR in GPx7, despite the reduction of the read-out at defined time intervals along each trace of the spectrophoto- active site by a redoxin such as PDI, suggested the need to screen metric assay. Consistently with the canonical ping-pong mechanism of whether a different Cys residue could substitute for it. Besides CP, GPxs, the double reciprocal plots, obtained at different concentrations the only Cys present in (Syn)GPx7 is the Cys 85 of mature mouse of GSH, produced parallel regression lines of the same slope (not GPx7, which lies within the strictly conserved PCNQF motif (Fig. 1). shown), confirming that also the chimeric GPx7 fits the enzyme substi- The Cys85/Ala substitution in (Syn)GPx7Ala85 failed to affect activity tution kinetic mechanism (Scheme 1). and kinetics of the peroxidatic reaction, irrespective of whether the Calculated specific activity, kinetic coefficients and second order reducing substrate was GSH or HsPDI (Table 3). This evidence defi- apparent rate constants of the different steps of the catalytic cycle nitely rules out the formation of a non-canonical disulfide in the reac- of (Syn)GPx7 using GSH as reducing substrate are reported in tion mechanism of GPx7, and conclusively supports the notion that 3852

MS2 spectrum from 759.6843 [z = 3], Cys + 2O .BsloTaane l iciiae ipyiaAt 80(03 3846 (2013) 1830 Acta Biophysica et Biochimica / al. et Bosello-Travain V.

C + 2 O

MS2 spectrum from 765.0173 [z = 3], Cys + 3O – 3857

Fig. 4. MS/MS analysis of the fragment of containing the catalytic Cys of (Syn)GPx7 purified by nickel chromatography. MS/MS spectrum of a 759.6843 (left) and 765.0173 (right) amu tryptic peptide obtained from Ni-NTA purified (Syn) GPx7. The spectrum represents the MmGPx7 44-64 peptide containing the CP (Cys56) in the sulfenic or sulfinic acid form (left or right panel respectively). Dotted lines point out identified peaks with Mascot matched m/z value within search tolerance constraints. Upon such values, b and y ion series are shown with possible modifications. Double headed arrows link ion fragments from various series and upon them the amino acid sequence assignment is reported for each mass delta. V. Bosello-Travain et al. / Biochimica et Biophysica Acta 1830 (2013) 3846–3857 3853 A

B

Fig. 5. Glutathione peroxidase activity measurements in the 29,000 ×g supernatant from induced bacteria (A) and MS/MS identification of the reaction intermediate (B). In (A) the spectrophotometric cuvette contained the assay mixture of GSH peroxidase activity, as reported under Material and methods, and the enzyme source, which was: the 29,000 ×g supernatant from the induced bacterial lysate transformed with pSynA/GPx7, containing 0.6 μg (Syn)GPx7 (a); the 29,000 ×g supernatant from the induced bacterial lysate transformed with pSynA, expressing equal amounts of Syn (b). In (c) two nanograms of purified GPx4 from rat testis was used as control. Reaction started by adding PCOOH at the time indicated. Only the sample containing (Syn)GPx7 (a) exhibited activity. In (b) the absence of PCOOH reactivity was confirmed by addition of GPx4 that consumed the whole substrate. Numbers represent the reaction rate expressed in nmol/min. In (B) the reduced 29,000 ×g supernatant from (Syn)GPx7 induced bacteria was incubated in the presence of GSH and PCOOH and treated for MS analysis, as reported under Material and methods. Annotated MS/MS spectra for the CP within a tryptic peptide are reported, show- ing the adduct between the CP of GPx7 and GSH.

C91K wt (Syn)GPx7 catalyzes a catalytic cycle in which the oxidized CP, but not might also oxidize PDI. SecGPx4, DmGPx as well as DmGPx ,a the CP–CR disulfide, is the actual oxidized intermediate that is subse- typical 2CysGPx, were analyzed as controls. None of these enzymes quently reduced by the redoxin. accepted HsPDI as a reducing substrate (Table 4). Oxidation of PDI emerges therefore as a rather specific feature of GPx7. 3.7. Reducing substrate preferences of monomeric GPxs 3.8. Kinetic constants of the interaction between GPx7 and PDI The notion that the catalytic cycle of GPx7 does not include the formation of a CP–CR disulfide, prompted the question whether the The interaction between MmGPx7 and HsPDI was quantitatively other monomeric GPxs, whose catalysis does not rely on disulfide, evaluated by surface plasmon resonance (SPR) (Fig. 7). To escape 3854 V. Bosello-Travain et al. / Biochimica et Biophysica Acta 1830 (2013) 3846–3857

Table 1 Specific activity, kinetic coefficients and apparent rate constants of (Syn)GPx7 for the reduction of PCOOH by GSH.

Enzyme Reducing substrate Specific activity Φ1 k+1 Φ2 k′+2 (μmol/min/mg prot.) (μMs) (M−1 s−1) (μMs) (M−1 s−1)

(Syn)GPx7 GSH 0.065 ± 0.014 105 ± 51 9.5 × 103 7.9 ± 8 × 103 12.6

Specific activity was measured in the 29,000 ×g supernatant from the bacterial lysate expressing (Syn)GPx7. Specific activity was measured in the presence of 25 μM PCOOH and 3 mM GSH, in the presence of NADPH and a non-limiting amount of GSH reductase. Enzyme concentration was calculated as described under Material and methods. Kinetic coef- ficients are reported as mean ± standard deviation of three independent experiments.

4. Discussion

GPx7 belongs to the subfamily of glutathione peroxidases that use a Cys residue for catalysis. It was first identified as a cytoplasmic pro- tein [8] and then found within the ER. In this study we provide further evidence for the localization previously supported by immunohisto- chemistry [9,25]. In a previous study, GPx7 was claimed to be not a glutathione per- oxidase but a PDI peroxidase [9]. However in that study the require- ments for a correct definition of enzymatic activity were not satisfied because the assay was conducted in the presence of GSH. Moreover, for the reaction in the presence of GSH, an unrealistically low rate constant was measured (0.58 M−1 s−1). This was indeed calculated without any kinetic steady-state treatment to dissect the different steps of the catalytic cycle [9]. On the other hand, the fact that the canonic criteria for a monomeric glutathione peroxidase to Fig. 6. Molecular docking pose for the complex GSH–GPx7. Molecular docking of GSH react with a Trx-related protein is not fulfilled by the GPx primary bound to the active site of the human GPx7 (PDB code: 2P31); on the right side, hydro- structure, further questioned the real substrate specificity of GPx7. philic (red) and hydrophobic (green) residues contribution for GSH–HsGPx1, GSH– HsGPx4, and GSH–HsGPx7 complexes, with associated constants values (pK). The pK Besides having a monomeric nature, the subfamily of CysGPxs re- values of GSH with GPx1, GPx4 and GPx7 indicate that the GSH–GPx7 complex is quires a canonical CR to react with a protein of the Trx family. In con- more stable, where the hydrophobic term (Hyd) brings major contribution to binding trast, this canonical CR is missing in GPx7 [5]. strength. In this study the validated procedure for the kinetic analysis of GPxs was used, which required a suitable source of enzymatic activi- ty. Unfortunately, GPx7 revealed to be extremely prone to oxidative inactivation. The peroxidase activity was indeed largely lost during possible artifacts, for these measurements the protein used was GPx7 Ni-NTA chromatography, even in the presence of reductants such as lacking the N-terminal Syn fusion. We assumed that the oxidation of mercaptoethanol or GSH. MS/MS analysis confirmed the artifactual the catalytic residue would not affect protein–protein interaction, over-oxidation of the CP. It is this marked susceptibility to oxidation which involves a large surface of the protein. This approach was in- of the active site during Ni-NTA chromatography that probably ex- trinsically validated by the positive result. Indeed, the analytical plains why low or even absent GSH peroxidase activity was previous- data could be successfully fitted with a 1:1 binding model and the av- ly reported [8,9]. erage values of the kinetic constants of binding could be precisely Searching for an expression system allowing minimal sample pro- −1 −1 −1 measured: ka = 346 M s ,kd = 0.0018 s ,KD = 5.2 μM cessing, we found that mouse GPx7 could be expressed in bacteria at (being KD =kd/ka). The very weak, non-specific interaction detected high yield as a N-terminal fusion with human synuclein, (Syn)GPx7. wt between HsPDI and other peroxidases (DmGPx and the CR mutant The soluble fraction of the bacterial lysate was used as source of activ- DmGPxC91K) did not permit any meaningful evaluation of interaction ity for kinetic studies, validated by the absence of competing reac- constants, in agreement with kinetic evidence. tions. Actually Racker's rule requiring a single enzymatic activity for

Table 2 Specific activity, kinetic coefficients and apparent rate constants of (Syn)GPx7 for the reduction of PCOOH by HsPDI.

Enzyme Reducing substrate Specific activity Φ1 k+1 Φ2 k′+2 (μmol/min/mg prot.) (μMs) (M−1 s−1) (μMs) (M−1 s−1)

(Syn)GPx7 HsPDI 0.048 ± 0.016 204 ± 60 4.9 × 103 285 ± 50 3.5 × 103

Specific activity and kinetic analysis as reported under Table 1, except that PDI (10 μM) replaced GSH, and rat Trx reductase replaced GSH reductase.

Table 3 Specific activity, kinetic coefficients and apparent rate constants of (Syn)GPx7C85A for the reduction of PCOOH by GSH or HsPDI.

Enzyme Reducing substrate Specific activity Φ1 k+1 Φ2 k′+2 (μmol/min/mg prot.) (μMs) (M−1 s−1) (μMs) (M−1 s−1)

(Syn)GPx7C85A GSH 0.057 ± 0.018 153 ± 40 6.5 × 103 80 ± 6 × 103 12.4 (Syn)GPx7C85A HsPDI 0.059 ± 0.010 172 ± 30 5.8 × 103 166 ± 70 6.0 × 103

Specific activity and kinetic analysis as reported under Table 1 and 2 except that the 29,000 ×g supernatant contained the (Syn)GPx7 mutant (Syn)GPx7C85A. V. Bosello-Travain et al. / Biochimica et Biophysica Acta 1830 (2013) 3846–3857 3855

Table 4 As suggested, a proton shuffling in the active site seems to facilitate Specific activity of individual GPxs with different reducing substrates. both deprotonation of the chalcogen and protonation of the anionic − Enzyme Reducing substrate Specific activity (μmol/min/mg protein) leaving group (RO ), on which the catalytic efficiency depends. Nota- bly, this oxidation mechanism of the active site is shared by all GPxs, (Syn)GPx7 ⁎ GSH 0.065 irrespective of the presence of selenium or sulfur as redox center [4]. HsTrxC72S Undetectable GPx7 strongly binds GSH and accepts it as reducing substrate, and ⁎ HsPDI 0.048 thus the enzyme must be rated as a true ‘GSH peroxidase’. The calcu- RrGPx4 −1 −1 lated k′+2 value (12.6 M s ) is similar to that of the artificial mu- GSH 17.2 HsTrxC72S Undetectable tant UP/CP of GPx4, where the enzymatic activity is limited, as in HsPDI Undetectable GPx7, by the reductive part of the catalytic cycle. This step of the cat- DmGPxwt alytic cycle is very fast when the redox center is a selenol or when Trx GSH 0.08 reduces the CP–CR disulfide [4]. Despite the slow turnover, GSH binds HsTrxC72S 5.12 to GPx7 similarly to GPx1 and GPx4 [1,29] although on different res- HsPDI Undetectable fi DmGPxC91K idues. A strong hydrophobic network signi cantly contributes to the GSH 0.26 complex stabilization. HsTrxC72S Undetectable Searching for an alternative substrate to GSH, we confirmed by ki- HsPDI Undetectable netic analysis the previous indication that PDI is oxidized by the ac- Specific peroxidase activity of the indicated enzymes was measured as described under tive site of GPx7. Notably, the rate constant for the oxidation of Material and methods with 25 μM PCOOH as the oxidizing substrate. GSH, HsTrxC72S or 3 −1 −1 HsPDI by (Syn)GPx7 (k′+2 = 3.5 × 10 M s ) is more than two HsPDI when present were 3 mM, 5 μMor10μM, respectively. The source of (Syn) orders of magnitude faster than that for the step encompassing oxida- GPx7, the supernatant from the bacterial lysate expressing the enzyme, RrGPx4, was fi a purified preparation of GPx4 from rat testis, DmGPxwt and DmGPxC91K represent tion of GSH and release of GSSG. At rst glance, this was somehow the wild type and the CR mutated forms of the DmGPx, respectively [5]. surprising as GPx7 does not contain the CR partner of the CP in con- ⁎ Data from Table 1. trast to all the other redoxin peroxidases of the CysGPx subfamily thus far analyzed [5,7,30]. The option that a non-canonical Cys resi-

due could substitute for the missing CR was positively ruled out by kinetic studies was satisfied using PCOOH as the oxidizing substrate mutagenesis of the sole non-peroxidatic Cys (Cys 85) of the mouse when either GSH or HsPDI was the reducing substrate. Using H2O2, GPx7. Thus GPx7 breaks the paradigm of the structure–function rela- however, did not satisfy that requirement. tionship of CysGPx. Apparently, the rule that dictates that the In the active site, conserved among GPxs, the oxidation rate of N-terminal Cys of the CXXC motif of the redoxin can only attach the 3 −1 −1 (Syn)GPx7 peroxidatic Cys by PCOOH (k+1 >10 M s ) is with- CR participating to the disulfide in the peroxidase (Eq. (6)) is not ab- in the range of that observed for the other CysGPxs [4]; i.e., orders of solute. In the reaction with GPx7 the N-terminal Cys of the CGHC magnitude faster than the non-enzymatic oxidation of a thiolate [27]. motif of PDI seemingly reacts with the oxidized CP of the peroxidase

Fig. 7. Representative SPR analysis of binding of different GPxs to reduced HsPDI. Typical sensograms of the binding of 5 μM each of purified MmGPx7, DmGPxwt, and DmGPxC91K to immobilized and reduced HsPDI. Curve a = MmGPx7; Curve b = DmGPxC91K; Curve c = DmGPxwt; Curve d = HBS-EP+, pH 7.4. All samples were exchanged with the HBS-EP+, pH 7.4 buffer by NAP chromatography before analysis. (RU = response unit, where 1 RU represents the binding of 1 pg of protein per square mm.) The insert shows the SDS-PAGE analysis of the digestion product of the chimeric GPx7 used for SPR experiments (MmGPx7, trace a). Lanes are as follows: 1: Ni-NTA purified (Syn)GPx7; 2 and 3: the purified prod- uct obtained after digestion by thrombin, which represents MmGPx7 (0.8 and 0.4 μg respectively); 4: Molecular weight size markers. The M.W. of (Syn)GPx7 and MmGPx7 is 36 and 19 kDa, respectively. 3856 V. Bosello-Travain et al. / Biochimica et Biophysica Acta 1830 (2013) 3846–3857 to form a mixed disulfide intermediate that rearranges, forming a di- Zennaro, Department of Molecular Medicine, University of Padova for sulfide on PDI, according to the reactions: SPR analysis. This study was supported in parts by grants to M.M. (Progetto di Ateneo CPDA087343/08) and to F.U. (Strategic Project ROOH þ ðÞCysGPx7 ⋯SH→ROH þ ðÞCysGPx7 ⋯SOH ð9Þ STPD082FN3-002).

ðÞCysGPx7 ⋯SOH þ HS⋯ðÞCGHC ⋯SH→ðÞCysGPx7 ⋯S–S⋯ðÞCGHC ⋯SH þ ð Þ H2O 10 Appendix A. Supplementary data

Supplementary data to this article can be found online at http:// dx.doi.org/10.1016/j.bbagen.2013.02.017.

Furthermore, the absence of a reactive CR on the peroxidase is expected to favor the oxidation of PDI. As the C-terminal Cys of the CGHC motifs of PDI is a poor nucleophile [31], following the formation References fi of the mixed disul de between GPx7 and PDI, the formation of the di- [1] L. Flohé, S. Toppo, G. Cozza, F. Ursini, A comparison of thiol peroxidase mecha- sulfide on the peroxidase would compete with the formation of the nisms, Antioxid. Redox Signal. 15 (2011) 763–780. disulfide on PDI. [2] I. Chambers, J. Frampton, P. Goldfarb, N. Affara, W. McBain, P.R. Harrison, The struc- Recently, a paper showing that in cells GPx7 forms a disulfide be- ture of the mouse glutathione peroxidase : the selenocysteine in the active site is encoded by the “termination” codon, TGA, EMBO J. 5 (1986) 1221–1227. tween the CP and Cys 85 was published [32]. Admittedly, this was a [3] P. Mauri, L. Benazzi, L. Flohé, M. Maiorino, P.G. Pietta, S. Pilawa, A. Roveri, F. Ursini, surprise, since we failed to detect such a product in our chimeric Versatility of selenium catalysis in PHGPx unraveled by LC/ESI-MS/MS, Biol. – GPx7, which was specifically screened by MS/MS analysis for this spe- Chem. 384 (2003) 575 588. [4] S. Toppo, L. Flohé, F. Ursini, S. Vanin, M. Maiorino, Catalytic mechanisms and spec- cies. Although our finding obviously does not exclude that the disul- ificities of glutathione peroxidases: variations of a basic scheme, Biochim. fide is formed in vivo, there were differences in the technical Biophys. Acta 1790 (2009) 1486–1500. [5] M. Maiorino, F. Ursini, V. Bosello, S. Toppo, S.C.E. Tosatto, P. Mauri, K. Becker, A. conditions for the MS analysis between our work and the paper of fi fi fi Roveri, C. Bulato, L. Benazzi, A. De Palma, L. Flohé, The thioredoxin speci city of Wei et al. that supports the disul de identi cation. Protein thiol Drosophila GPx: a paradigm for a -like mechanism of many gluta- groups, indeed were not blocked during trypsin digestion under alka- thione peroxidases, J. Mol. Biol. 365 (2007) 1033–1046. line conditions. Thus the common artifact, due to thiol scrambling [6] Z.A. Wood, E. Schröder, J. Robin Harris, L.B. Poole, Structure, mechanism and reg- ulation of peroxiredoxins, Trends Biochem. Sci. 28 (2003) 32–40. during the digestive procedure cannot be ruled-out. [7] T. Schlecker, M.A. Comini, J. Melchers, T. Ruppert, R.L. Krauth-Siegel, Catalytic Besides specific redox chemistry at the active site, oxidation of mechanism of the glutathione peroxidase-type tryparedoxin peroxidase of HsPDI by MmGPx7 also relies on specific protein–protein interaction. Trypanosoma brucei, Biochem. J. 405 (2007) 445–454. fi fi fi [8] A. Utomo, X. Jiang, S. Furuta, J. Yun, D.S. Levin, Y.-C.J. Wang, K.V. Desai, J.E. Green, SPR af nity constant measurement con rms the speci city of the in- P.-L. Chen, W.-H. Lee, Identification of a novel putative non-selenocysteine teraction between HsPDI and MmGPx7 in agreement with activity containing phospholipid hydroperoxide glutathione peroxidase (NPGPx) essen- measurements showing that PDI does not react with monomeric per- tial for alleviating oxidative stress generated from polyunsaturated fatty acids – oxidases other than GPx7, nor does GPx7 react with Trx. in breast cancer cells, J. Biol. Chem. 279 (2004) 43522 43529. [9] V.D. Nguyen, M.J. Saaranen, A.-R. Karala, A.-K. Lappi, L. Wang, I.B. Raykhel, H.I. Unfortunately, due to the presence of competing reactions, our ex- Alanen, K.E.H. Salo, C.-C. Wang, L.W. Ruddock, Two endoplasmic reticulum PDI perimental set-up did not permit the study of reactivity between peroxidases increase the efficiency of the use of peroxide during disulfide bond – GPx7 and H O . Regardless, in view of both the simplest structure of formation, J. Mol. Biol. 406 (2011) 503 515. 2 2 [10] S. Toppo, S. Vanin, V. Bosello, S.C.E. Tosatto, Evolutionary and structural insights H2O2 vs. PCOOH and that the monomeric peroxidases generally into the multifaceted glutathione peroxidase (Gpx) superfamily, Antioxid. – react almost identically with PCOOH or H2O2 [18], we may safely pre- Redox Signal. 10 (2008) 1501 1513. sume that the GPx7 active site is oxidized by H O as well as PCOOH. [11] A. Negro, A.M. Brunati, A. Donella-Deana, M.L. Massimino, L.A. Pinna, Multiple 2 2 phosphorylation of alpha-synuclein by protein tyrosine kinase Syk prevents Which of the two substrates that is relevant in vivo cannot be deter- eosin-induced aggregation, FASEB J. 16 (2002) 210–212. mined at this stage of knowledge. What we do know is that mono- [12] M.M. Bradford, A rapid and sensitive method for the quantitation of microgram meric structure is the crucial constraint for fast reactivity with quantities of protein utilizing the principle of protein-dye binding, Anal. Biochem. 72 (1976) 248–254. PCOOH [29] and that this might be relevant to GPx7 physiological [13] X.X. Ren, M.M. Björnstedt, B.B. Shen, M.L.M. Ericson, A.A. Holmgren, Mutagenesis function as is indeed becoming evident for GPx4 [33]. of structural half-cystine residues in human thioredoxin and effects on the regu- lation of activity by selenodiglutathione, Biochemistry 32 (1993) 9701–9708. [14] A. Roveri, M. Maiorino, C. Nisii, F. Ursini, Purification and characterization of 5. Conclusions phospholipid hydroperoxide glutathione peroxidase from rat testis mitochondri- al membranes, Biochim. Biophys. Acta 1208 (1994) 211–221. GPx7 emerges as a unique member of the CysGPx subfamily of [15] M. Maiorino, C. Gregolin, F. Ursini, Phospholipid hydroperoxide glutathione per- – fi oxidase, Methods Enzymol. 186 (1990) 448 457. glutathione peroxidases. It ef ciently oxidizes PDI by a mechanism [16] J. Lundström, A. Holmgren, Protein disulfide-isomerase is a substrate for thioredoxin involving a peroxidatic Cys residue only. In this context, the use of reductase and has thioredoxin-like activity, J. Biol. Chem. 265 (1990) 9114–9120. GSH as a substrate is very intriguing. The observed low rate, despite [17] G. Weber, C. Long, Biochemists' Handbook, Ltd., C. Long, E. and FN Spon, London, the strong interaction with the active site, argues against a function- 1961. [18] F. Ursini, M. Maiorino, C. Gregolin, The selenoenzyme phospholipid hydroperox- ally relevant reduction of hydroperoxides using GSH. Considering ide glutathione peroxidase, Biochim. Biophys. Acta 839 (1985) 62–70. the expected concentration of GSH and PDI in vivo, compatible with [19] M. Maiorino, K.D. Aumann, R. Brigelius-Flohé, D. Doria, J. van den Heuvel, J. those used in this study [34,35], kinetic analysis reveals a competition McCarthy, A. Roveri, F. Ursini, L. Flohé, Probing the presumed of selenium-containing peroxidases by mutational analysis of phospholipid hydro- between the alternative substrates GSH and PDI. Upon a fast oxida- peroxide glutathione peroxidase (PHGPx), Biol. Chem. Hoppe Seyler 376 (1995) tion by ROOH, the oxidized active site of GPx7 can oxidize PDI only 651–660. – when GSH concentration is permissively low. This consideration [20] K. Dalziel, Initial steady state velocities in the evaluation of enzyme coenzyme- substrate reaction mechanisms, Acta Chem. Scand. 11 (1958) 1706–1723. frames the discussion of the physiological role of GPx7 in the oxida- [21] MOE-Dock, Molecular Operating Environment (MOE 2009.10), The Chemical tion of PDI within the ER as a reaction compatible with control or Computing Group, Montreal, Canada, 2009, http://www.chemcomp.com. modulation by the actual concentration of GSH [35]. [22] GOLD suite (version 5.1), The Cambridge Crystallographic Data Centre (CCDC), Cambridge, UK, 2011, www.ccdc.cam.ac.uk/products/life_sciences/gold/. [23] R.A. Friesner, J.L. Banks, R.B. Murphy, T.A. Halgren, J.J. Klicic, D.T. Mainz, M.P. Acknowledgements Repasky, E.H. Knoll, M. Shelley, J.K. Perry, D.E. Shaw, P. Francis, P.S. Shenkin, Glide: a new approach for rapid, accurate docking and scoring. 1. Method and as- sessment of docking accuracy, J. Med. Chem. 47 (2004) 1739–1749. We thank Professor Stefano Moro, Molecular Modeling Section of [24] N. Hulo, A. Bairoch, V. Bulliard, L. Cerutti, E. De Castro, P.S. Langendijk-Genevaux, M. the University of Padova, for invaluable technical support and Dr Lucio Pagni, C.J.A. Sigrist, The PROSITE database, Nucleic Acids Res. 34 (2006) D227–D230. V. Bosello-Travain et al. / Biochimica et Biophysica Acta 1830 (2013) 3846–3857 3857

[25] I. Raykhel, H. Alanen, K. Salo, J. Jurvansuu, V.D. Nguyen, M. Latva-Ranta, L. [31] A.K. Lappi, M.F. Lensink, H.I. Alanen, K.E.H. Salo, M. Lobell, A.H. Juffer, L.W. Ruddock, A molecular specificity code for the three mammalian KDEL receptors, Ruddock, A conserved arginine plays a role in the catalytic cycle of the protein di- J. Cell Biol. 179 (2007) 1193–1204. sulphide , J. Mol. Biol. 335 (2004) 283–295. [26] P. Hinkle, Efraim Racker, 1913–1991, J. Bioenerg. Biomembr. 24 (1992) 517–519. [32] P.-C. Wei, Y.-H. Hsieh, M.-I. Su, X.J. Jiang, P.-H. Hsu, W.-T. Lo, J.-Y. Weng, Y.-M. [27] C.C. Winterbourn, D. Metodiewa, Reactivity of biologically important thiol com- Jeng, J.-M. Wang, P.-L. Chen, Y.-C. Chang, K.-F. Lee, M.-D. Tsai, J.-Y. Shew, W.-H. pounds with superoxide and hydrogen peroxide, Free Radic. Biol. Med. 27 Lee, Loss of the oxidative stress sensor NPGPx compromises GRP78 chaperone ac- (1999) 322–328. tivity and induces systemic disease, Mol. Cell (2012) 1–13. [28] M.M. Gallogly, D.W. Starke, J.J. Mieyal, Mechanistic and kinetic details of catalysis [33] A. Seiler, M. Schneider, H. Förster, S. Roth, E.K. Wirth, C. Culmsee, N. Plesnila, E. of thiol–disulfide exchange by glutaredoxins and potential mechanisms of regu- Kremmer, O. Rådmark, W. Wurst, G.W. Bornkamm, U. Schweizer, M. Conrad, Gluta- lation, Antioxid. Redox Signal. 11 (2009) 1059–1081. thione peroxidase 4 senses and translates oxidative stress into 12/15-lipoxygenase [29] F. Ursini, M. Maiorino, R. Brigelius-Flohé, K.D. Aumann, A. Roveri, D. Schomburg, L. dependent- and AIF-mediated cell death, Cell Metab. 8 (2008) 237–248. Flohé, Diversity of glutathione peroxidases, Methods Enzymol. 252 (1995) 38–53. [34] F.Q. Schafer, G.R. Buettner, Redox environment of the cell as viewed through the [30] H. Sztajer, B. Gamain, K.D. Aumann, C. Slomianny, K. Becker, R. Brigelius- redox state of the glutathione disulfide/glutathione couple, Free Radic. Biol. Med. Flohé, L. Flohé, The putative glutathione peroxidase gene of Plasmodium 30 (2001) 1191–1212. falciparum codes for a thioredoxin peroxidase, J. Biol. Chem. 276 (2001) [35] A.-K. Lappi, L.W. Ruddock, Reexamination of the role of interplay between gluta- 7397–7403. thione and protein disulfide isomerase, J. Mol. Biol. 409 (2011) 238–249.