Tumor Cell Phenotype Is Sustained by Selective MAPK Oxidation in Mitochondria

Mitochondria are major cellular sources of hydrogen peroxide (H2O2), the production of which is modulated by oxygen availability and the mitochondrial energy state. An increase of steady-state cell H2O2 concentration is able to control the transition from proliferating to quiescent phenotypes and to signal the end of proliferation; in tumor cells thereby, low H2O2 due to defective mitochondrial metabolism can contribute to sustain proliferation. Mitogen-activated protein kinases (MAPKs) orchestrate signal transduction and recent data indicate that are present in mitochondria and regulated by the redox state. On these bases, we investigated the mechanistic connection of tumor mitochondrial dysfunction, H2O2 yield, and activation of MAPKs in LP07 murine tumor cells with confocal microscopy, in vivo imaging and directed mutagenesis. Two redox conditions were examined: low 1 µM H2O2 increased cell proliferation in ERK1/2-dependent manner whereas high 50 µM H2O2 arrested cell cycle by p38 and JNK1/2 activation. Regarding the experimental conditions as a three-compartment model (mitochondria, cytosol, and nuclei), the different responses depended on MAPKs preferential traffic to mitochondria, where a selective activation of either ERK1/2 or p38-JNK1/2 by co-localized upstream kinases (MAPKKs) facilitated their further passage to nuclei. As assessed by mass spectra, MAPKs activation and efficient binding to cognate MAPKKs resulted from oxidation of conserved ERK1/2 or p38-JNK1/2 cysteine domains to sulfinic and sulfonic acids at a definite H2O2 level. Like this, high H2O2 or directed mutation of redox-sensitive ERK2 Cys214 impeded binding to MEK1/2, caused ERK2 retention in mitochondria and restricted shuttle to nuclei. It is surmised that selective cysteine oxidations adjust the electrostatic forces that participate in a particular MAPK-MAPKK interaction. Considering that tumor mitochondria are dysfunctional, their inability to increase H2O2 yield should disrupt synchronized MAPK oxidations and the regulation of cell cycle leading cells to remain in a proliferating phenotype.


Introduction
The cell's redox status controls the progression of the cell cycle, including misregulation in cancer [1,2]. Oxidants, such as H 2 O 2 , play an important role in the activation of signaling molecules, which control the complex machinery involved in cell proliferation, differentiation, apoptosis, and senescence. An attractive notion is that the continuous increase in oxidant concentration may trigger disparate cell responses: slight variations in H 2 O 2 concentration (0.7-20 mM H 2 O 2 ) help determine normal cell fate, i.e., proliferation [3,4], arrest, senescence or apoptosis [5]. Moreover, an increase in H 2 O 2 steady-state concentration ([H 2 O 2 ] ss ) has been observed in vivo in the transition from proliferative hepatoblasts to quiescent and differentiated hepatocytes [6].
Mitochondria are major cellular sources of H 2 O 2 , the production of which is modulated by the mitochondrial energy state and generation of nitric oxide [7]. High mitochondrial H 2 O 2 yield is associated with late rat brain and liver development and signals the end of proliferation [6,8]. From this perspective, development can be understood as a transition from anaerobic metabolism to a five-fold increase in metabolism in mature cells; arrest and differentiation are associated to high mitochondrial activity and membrane potential [9]. Mitochondria are dysfunctional in cancer: the activity of mitochondrial complexes is decreased, the mitochondrial generation of H 2 O 2 is substantially decreased [10], the mitochondrial-K + channel axis is suppressed [11], the oxidant-dependent inhibition p38 MAPK is impaired, and p53 suppresses mitochondrion-driven apoptosis [12]. Hence, it may be surmised that tumor cells -alike embryonic tissues-live at a very low [H 2 O 2 ] ss [6,10,13].
To understand the mechanisms of redox modulation by MAPKs, studies have been focused on the oxidative inhibition of phosphatases (MKP) [25] and on the interaction of MAPKKs with antioxidant proteins, such as thioredoxin [26]. However, the direct effects of oxidants on MAPKs and the mechanisms of the transition from proliferation to arrest remain obscure.
Recent data indicate that MAPKs are present in mitochondria, as well as other kinases like PKC and Akt [27,28]. However, the connection among tumor mitochondrial dysfunction, H 2 O 2 yield, and activation of MAPKs still awaits elucidation. In the present work, we provide evidence of MAPKs subcellular redistribution upon activation, including their transit through mitochondria. We demonstrate that the redox state modulates the mitochondrial interaction of MAPKs to MAPKKs by oxidation of conserved cysteine domains of MAPKs to sulfinic and sulfonic acid; this biochemical mechanism determines MAPKs differential activation and traffic to nuclei and ultimately, sustains the phenotype of LP07 tumor cells.  [6,10]. In these cells, mitochondria have a low H 2 O 2 production rate but they still respond to oxidative stress as the normal ones do [6,10]. In the present study, we thus examined the redox transition as represented by low (1 mM) and high (50 mM) H 2 O 2 concentrations. This transition offers the opportunity to test a) the circuit of redox signaling based upon mitochondria and, b) a mechanistic of low H 2 O 2 yield for persistent cell proliferation. We show in Figure 1A  (JNK1/2 inhibitor). As inferred by the effect of U0126, the temporal increase of cyclin D1 observed with 1 mM H 2 O 2 depended as well on the activation of ERK1/2, while downregulation of cyclin D1 at 50 mM of H 2 O 2 was reverted by the p38 and JNK1/2 inhibitors ( Fig. 1C and D). It is concluded that cell cycle variations are related to the differential redox activation of extracellular or stress activated MAPK.

Mitochondria are a meeting point of MAPKs and their upstream activators
Activation of MAPKs in response to a variety of stimuli, including oxidative stress [19,29] has been defined in a two-compartment system: phosphorylation (activation) in cytosol followed by translocation to nuclei [16,30]. Preliminary data on MAPK mitochondrial turnover and its modulation by redox status [24,31] prompted us to examine a three-compartment model for MAPK redistribution upon activation in LP07 cells. To confirm the presence of MAPKs in LP07, cells were labeled with MitoTracker Deep Red and immune stained with anti ERK1/2, p38, and JNK1/2 primary antibodies and secondary antibodies conjugated with Cy3. Images were obtained by confocal microscopy and further analyzed by intensity correlation analysis (ICA) [32] (Fig. 2A); the presence of a diagonal in the 2D fluorescence intensity histogram demonstrated that MAPKs are constitutively expressed in mitochondria. Furthermore, mitochondrial subfractionation showed that MAPKs colocalized with their cognates MEK1/2, MKK3 and MKK4 in the mitochondrial outer membrane and in the intermembrane space (Fig. 2B), as corroborated by the different fraction markers. ERK1/2 molecules were also detected within the organelles by electron microscopy (Fig. 2C).

Traffic of MAPKs and MAPKKs elicited by redox stimulation involves the mitochondrial compartment
Comparison of the activation kinetics and nuclear accumulation of ERK1/2, JNK1/2, and p38 upon oxidative perturbation yielded a remarkable difference in MAPK responses. After stimulation with 1 mM H 2 O 2 , p-ERK1/2 and total ERK1/2 increased in mitochondria, cytosol and nuclei (Fig. 3). An hour after stimulation, p-ERK1/2 returned to the basal level in mitochondria but remained elevated in nuclei. Conversely, 50 mM H 2 O 2 entailed a considerable lower rate of ERK1/2 translocation and reduced its activation by ten-fold, and the kinase was retained in mitochondria in detriment of nuclear accumulation (Fig. 3). The sum of the kinetics integrals for each H 2 O 2 treatment remained constant (Fig. 3, circled numbers) which emphasizes the notion of redox interdependence between the mitochondrial and nuclear compartments.
A similar approach to assess the redox regulation of the kinetics JNK1/2 and p38 ( Fig. 4) revealed that these were barely affected at low (1 mM) H 2 O 2 levels. Instead, upon treatment with 50 mM H 2 O 2 , total and phosphorylated JNK1/2 and p38 increased in mitochondria and cytosol and then translocated and accumulated in the nuclei (Fig. 4).
In order to assess the in vivo translocation of MAPKs into mitochondria, LP07 cells were transfected with either GFP-hERK2 or GFP-hJNK1, stained them with MitoTracker Deep Red, and continuously followed for MAPK redistribution upon H 2 O 2 stimulation by video confocal microscopy. Low H 2 O 2 caused GFP-hERK2 entrance to mitochondria and subsequent translocation to nuclei (Fig. 5A). A similar behaviour but at high H 2 O 2 stimulation was found for GFP-hJNK1 (Fig. 5B). In the right panel of Figure 5, fluorescence values were plotted every minute during 40 min. In both redox conditions, a previous passage to mitochondria anticipated further traffic to nucleus; however, in the experimental conditions, GFP-hJNK1 turnover resulted faster than that of GFP-hERK2. The patterns described by confocal microscopy were similar to those observed by western blot in Figs. 3 and 4.
The subcellular redistribution of MAPKKs and their redox regulation are shown in Fig. S1. 1 mM H 2 O 2 stimulus elicited an initial MEK1/2 outward movement from mitochondria to cytosol (Fig. S1A). As ERK1/2, p-MEK1/2 was retained in mitochondria when stimulated with 50 mM H 2 O 2 . MEK1/2 translocation to nuclei could not be detected in agreement with previous reports [29,33].
The pattern of redistribution and activation of MKK4 (Fig. S1B) and MKK3 (Fig. S1C) following stimulation with high H 2 O 2 mimicked that of MEK1/2 at 1 mM H 2 O 2 , i.e., the MAPKKs moved out from the mitochondria to the cytosol. Fig. S1D is a scheme representative of putative cycle of upstream MAPKKs in the studied conditions to carry up MAPKs to nucleus. The degree of MAPK nuclear retention is related to cell cycle progression [16,33]; ERK1/2 retention correlates to cell proliferation, while retention of JNK1/2 and p38 correlates with cell cycle arrest (see Figs. 1 and 2).
Oxidation favors differential MAPKs phosphorylation in vitro, but does not change the intrinsic catalytic activity Direct effects of H 2 O 2 on MAPK catalytic activity were assessed with human recombinant ERK2-GST or immunoprecipitated JNK1/2 and p38 immobilized on agarose. Kinases exposed to H 2 O 2 were subsequently incubated with substrates or upstream kinases and 32 P-cATP. Phosphorylation efficiency of ERK2 by MEK1/2 was enhanced at low H 2 O 2 concentrations, whereas decreased at high H 2 O 2 concentration (Fig. 6A). Oppositely, phosphorylation of JNK1/2 and p38 by respective MKK4 and MKK3, was enhanced at high H 2 O 2 level (Fig. 6A). However, H 2 O 2 treatments did not affect the intrinsic catalytic activity of ERK2, JNK1/2, or p38 as proved by absence of effects on the phosphorylation of myelin basic protein (substrate for ERK1/2) or ATF-2 (substrate for p38 and JNK1/2) (Fig. 6A). We therefore surmise that exposure of ERK2, JNK1/2, and p38 to determined H 2 O 2 concentrations increases phosphorylation efficiency by introducing a post-translational modification that would enhance the interaction with their respective MAPKKs.
To see whether phosphorylation efficiency varies by redox effects on binding, we treated recombinant MAPKs with 0.1-10 mM H 2 O 2 , and subsequently incubated them with cytosolic or mitochondrial fractions containing the respective MAPKKs. Similarly to phosphorylating activities, ERK2-GST binding to MEK1/2 resulted enhanced at low H 2 O 2 level (0.1 mM H 2 O 2 ) ( Fig. 6B) whereas p38-GST and JNK2-GST binding to cognate MAPKKs was facilitated at 10 mM H 2 O 2 (Fig. 6B). Interestingly, oxidation of ERK2-GST enhanced its dimerization and activation as shown by the interaction with endogenous ERK1/2 and the increase in phosphorylation (Fig. 6B). However, these interactions and ERK2 activation were enterely disrupted at H 2 O 2 concentrations above 1 mM.

H 2 O 2 modulates ERK-MEK interaction and shuttle to nuclei
To examine whether redox effects on binding in vivo resemble the ones observed in vitro, LP07 cells were stimulated with 1 and 50 mM H 2 O 2 , p-MEK1/2 and ERK1/2 were precipitated from mitochondria, and cytosol and complex formation was followed by western blot in a pull-down assay. As observed in the in vitro assay, in vivo p-MEK1/2-ERK1/2 interaction was substantially increased at low H 2 O 2 and decreased at high H 2 O 2 concentration (Fig. 7A).
To investigate whether modulation of MEK-ERK interaction in mitochondria affects shuttle to nuclei, cells were transfected with ERK2 and its mutants H230R or Y261N, both with restricted docking to MEK1/2 [33]. At low H 2 O 2, transfected ERK2 wild type followed a typical sequence of translocation to mitochondria and nucleus as shown in Figures 2 and 5A. Oppositely, ERK2 mutants with poor binding to MEK1/2 were retained in the organelles in detriment of their translocation to nuclei (Fig. 7B). These findings suggest that the traffic of MAPKs to the nuclei and thus, cell behaviour, depend on H 2 O 2 -induced changes in their loop of activation, as resulted from redox variations in the domains docking the upstream MAPKKs in mitochondria [34,35].
ERK2, p38, and JNK2 cysteine thiols are specifically oxidized by H 2 O 2 Considering the susceptibility to oxidation of cysteine thiol moieties in proteins, we explored the relevance of these amino acids in the regulation of MAPKs pathways. ERK2-GST immobilized on agarose was exposed to the thiol blocker 4-vynilpyridine (4-VP) and then incubated with a mitochondrial fraction. 4-vinylpyridine treatment resulted in a markedly reduced ERK-MEK interaction (Fig. 8A).
Oxidized cysteines responsible for the differential binding of MAPKs were identified by LC/MS/MS. After treatment with low H 2 O 2 concentrations (0.1 mM), the thiol groups of ERK2 Cys 38

ERK-MEK interaction and ERK shuttle to nuclei depend on mitochondrial oxidation of the redox-sensitive cysteines
In order to assess the role of oxidizable cysteines on ERK activation and redistribution, ERK2 mutants C38A, C214A, and C214E were transfected onto LP07 cells to search for their interaction with MEK1/2. H 2 O 2 oxidation enhanced wild type ERK2-MEK1/2 interaction (Fig. 8B), as was previously observed in Fig. 6B. However, H 2 O 2 had no effect on the interaction of MEK1/2 with ERK2 when Cys 214 was substituted by an Ala (C214A) (Fig. 8B). ERK2-MEK1/2 interaction was otherwise enhanced by the replacement of Cys 214 with a Glu (C214E), even in the absence of H 2 O 2 (Fig. 8B).
Wild type ERK2 translocated to mitochondria and was afterwards retained in nuclei after stimulation with 1 mM H 2 O 2 , as endogenous ERK (Fig. 8C and see Fig. 3). In contrast, ERK2 mutants C38A and C214A were retained in mitochondria in detriment of nuclear entrance (Fig. 8C). p-MEK1/2 was as well retained in mitochondria after transfection with ERK2 mutants C38A and C214A, which indicates that defective oxidation impedes ERK-MEK complex exit from mitochondria (Fig. 8D).
ERK redox-sensitive cysteine domains are well conserved in all MAPKs as well as in other kinases ( Table 2). Noteworthy is the fact that both oxidable Cys 38 in ERK2 and Cys 41 in JNK2 are in the same domain and were oxidized to -SO 3 H, but at different H 2 O 2 levels. JNK2 Cys 213 and p38 Cys 48 and Cys 231 all present in these domains, were not oxidized.

Discussion
In the present study we provide evidence that supports that H 2 O 2 drives proliferation and cell cycle arrest by specific MAPKs activation in LP07 cells. Opposite effects elicited by low and high H 2 O 2 concentrations have been previously observed in various tumor cell lines [10], as well as in normal tissues [6]; this supports the notion that increases in H 2 O 2 concentrations change cell phenotype by eliciting sequential and opposite responses [3]. This process relies on the alternative activation of either ERK1/2 or JNK1/2 and p38 MAPKs by modulation of the interaction with their cognate MAPKKs. We also confirm the presence of MAPKs and MAPKKs in mitochondria and that these kinases redistribute to the cytosolic and nuclear compartments. The mechanism of MAPK import onto the outer membrane and intermembrane space still awaits elucidation [36]. MAPKs could enter and exit the organelle without previous mitochondrial damage, alike Bcl-2 or Bcl-x L , both mitochondrial proteins that translocate to nuclei [37,38].
We show for the first time that modulation of MAPKs interaction with their upstream kinases relies on the oxidation of  (Table 2). A remarkable finding is that selective kinase activation is based on a differential sensitivity to oxidants, particularly H 2 O 2 . Furthermore, efficient ERK2 binding to MEK is achieved by oxidation of two of the five ERK2 cysteine thiols, Cys 38 and Cys 214 , to cysteine sulfinic (-SOH 2 ) and sulfonic (-SOH 3  ERK2 cysteine oxidation by H 2 O 2 proceeds outside its catalytic site, increases binding to MEK1/2 by three to four-fold, modulates activity, and, more importantly, it does not occur at every H 2 O 2 concentration, for it can only be achieved at low oxidant level. Cysteine post-translational modifications appear to be critical for the ERK activation pathway and kinase redistribution, and  Modulation of enzyme activity by cysteine oxidation to sulfenic acid has been reported: thiol oxidation has been shown to participate in the catalytic mechanism of peroxiredoxins (Prx) isoforms I-VI, which reduce H 2 O 2 by forming disulphide [39], sulfenic acid (in Cys 51 in mammals) [40,41]. Prxs are further oxidized to sulfinic acid and inactivated [42], the same that MAPK phophatase-3 [43] and phosphatase-1B [44]. The cysteine sulfinic acid of Prxs can be reduced to cysteine by ATP-dependent sulfiredoxin [45] or sestrins-Hi95 or sestrin2 [46]. Together, these facts argue for the reversible oxidation of cysteines as a mechanism of protein activation.
The domains surrounding the cysteines susceptible to oxidation in the former enzymes share an Arg residue (Cys+6 to Cys+11) [47] as well as those in MAPKs. Thiol oxidation requires a low pKa for the cysteine group [48]. When located in position Cys+10 or Cys+9, Arg helps decrease pKa below 5.8 (normal Cys pKa = 8.5), to dissociate Cys-SH to the thiolate (Cys-S -) at physiological pH, and to stabilize cysteine sulfinic oxidation. Because mutation of Cys 214 to glutamic acid enhances ERK-MEK interaction, negative charges must be involved in the interaction and thus, we postulate that cysteine oxidation introduces negative with wild type (WT) ERK2, or ERK2 mutants H230R or Y261N, both with reduced binding capacity to MEK1/2 [33]. After incubation with H 2 O 2 , mitochondrial and nuclear fractions were isolated, and ERK2 distribution was detected by western blot using an anti Flag antibody. A western blot representative of 3 independent experiments is shown. Complex IV and PCNA were used as loading control. doi:10.1371/journal.pone.0002379.g007 charges with very low pKa (,2) that allow the interaction to occur [48]. An attractive idea is that charged cysteines lead MEK or other ligands as they ''walk'' through arginines to Asp 316 and Asp 319 (Fig. 9A and B), the two essential ERK acidic residues that integrate the D domain for binding to ligands [32]. p38 Cys 162 negative charges might be attracted to the arginines of upstream ligands, like Arg 104 of MKK3 [49]. With respect to JNK, the probable mechanism of redox binding is controversial, for  oxidation proceeds in several Cys domains. This argues for the possibility that oxidation of multiple cysteines may stabilize the hydrophobic interactions involved in the binding capacity of proteins and also suggests that more than one model of regulation by oxidation could apply for the different biological systems and molecular pathways. We conclude that proliferation of LP07 cells, and presumably in other cellular systems, depends on sustained oxidation of ERK2 thiols to sulfinic or sulfonic acid achieved at low oxidant yield. If dysfunctional mitochondria are incapable of increasing the oxidative level in cells [10], then they cannot contribute to cell cycle arrest either by oxidation of p38 Cys 162 or by impeding Cys 38 and Cys 214 oxidation in ERK2 and, thus, drive to an uncontrolled cell division and ultimately to cancer.

Materials and Methods
Cell line, culture conditions and treatments LP07 cell line was derived from P07 lung tumor spontaneously arisen in a BALB/c mouse, and extensively characterized [10,50]. Cells were maintained in Dulbecco's modified Eagle's medium nutrient mixture F-12 HAM (D-MEM) with 10% fetal bovine serum (FBS) and 50 mg/ml gentamycin. For treatment, cells were 24 h serum starved and then stimulated with H 2 O 2 and/or the MAPKs inhibitors (2 h prior to H 2 O 2 treatment). When appropriate, cells were seeded onto a 22.1-mm diameter well and transiently transfected with wild-type 3XFlag-CMV7-ERK2 vector, the ERK2 mutants H230R and Y261N (kindly provided by Dr. M. Cobb) or ERK2 mutants C38A, C214A and C214E (Genscript, Piscataway, NJ, USA), with Lipofectamine 2000 (Invitrogen) according to manufacturer's instructions. After transfection, cells were stimulated and harvested, and subcellular fractions were recovered for western blot.
Proliferation assays LP07 cells were seeded onto a 96-well plate (8.10 4 cells/well) and treated as described above for 48 h in the presence of 0.8 mCi/well [ 3 H] thymidine (specific activity, 70 to 90 Ci/mmol; NEN/Dupont, Boston, MA). Then cells were harvested and radioactivity measured in a liquid scintillation counter (Wallac 1414, Turku, Finland). Alternatively, LP07 cells seeded onto a 24well plate (1.10 5 cells/well) and treated as above, were harvested by trypsinization and counted in a Neubauer chamber in the consecutive days after stimulation.
Cell growth assay LP07 were seeded onto a 24-well plate, treated as above, and then harvested and counted in a Neubauer chamber in the consecutive days after stimulation.

Cell cycle and apoptosis assays
Treated cells were harvested and incubated with (i) 100 mg/mL propidium iodide in 0.1% sodium citrate, 0.1% Triton X-100 at 4uC overnight in the darkness [51] or (ii) Annexin V-FITC (Immunotech) according to manufacturer's instructions. Cells were run on a FACScalibur flow cytometer (Becton-Dickinson, Mountain View, CA) and analyzed with WinMDI software for windows.

Fluorescence labeling and confocal microscopy
Cells were grown on cover slides and stained with a specific mitochondrial marker, MitoTracker Deep Red 633 FM (Molecular probes) (100 nM, 45 min at 37uC), fixed in 4% paraformaldehyde, blocked in 1% BSA, 0.3% Triton X-100, PBS, pH 7.4, in a humidified chamber for 1 h, and incubated with primary (anti ERK1/2, JNK1/2 or p38) antibodies and secondary antibodies conjugated with Cy3 for 1h at RT in the same buffer. Cover slides were mounted in Fluorsave mounting media (Calbiochem). For in vivo imaging, cells were grown on Lab-Tek Chambered Borosilicate Coverglass System (Nunc), and transfected with GFP-hERK2 or JNK1-GFP, and stained with Mitotracker as described above. Confocal laser scanning microscopy was performed with an Olympus FV1000 using a 6361.35 NA oil immersion objective. Excitation filters and emission detected with a PDA device were as follows: GFP, 488 nm excitation, 500-560 nm emission; Cy3, 532 nm excitation, 580610 nm emission; MitoTracker Deep Red, 633 nm excitation, 650-750 nm emission. Images were acquired with Olympus Fluoview FV10-ASW software and analyzed with DIP image software for MATLAB (TNO, Delft). Images were analyzed by intensity correlation analysis (ICA) [51]; if two structures are part of the same complex or are present at the same place, then their staining intensities should vary in synchrony, whereas if they are on different complexes or structures they will exhibit asynchronous staining. ICA was done with MATLAB software as already developed, studied and reported [52]. Live images were analyzed as follows: a mask was done over the nucleus or over mitochondria when MitoTracker intensity was over 20, and GFP fluorescence was followed in these areas as well as in the whole cell. Nuclear and mitochondrial GFP fluorescence intensity was corrected for total photobleaching.

Immune-electron microscopy
Cells were fixed in 4% paraformaldehyde and 0.5% glutaraldehyde in phosphate buffer 0.2 M pH 7.4 for 4 h. After washing in 0.1% glycine the samples were dehydrated in ethanol and embedded in LR White as described [6]. Samples were labeled overnight with anti p-ERK1/2 antibody diluted 1:10, and then washed in 1% BSA and 0.05% Tween in PBS. Then they were incubated with a secondary antibody (1:50) conjugated with 10 nm colloidal gold particles (Sigma anti-rabbit IgG) for 1 h at RT. Grids were washed, fixed in 2% glutaraldehyde for 10 minutes and counterstained in 2% uranyl acetate. Samples were analyzed on a Zeiss EM-C10 transmission electron microscope (506 and 1006 objective) at 80k and photographed with 35 mm Kodak electron film.

Pull down assay
Cytosolic or mitochondrial fractions were incubated with human recombinant ERK2-GST, p38-GST or JNK2-GST (Stressgen) oxidized with H 2 O 2 and bound to agarose in lysing buffer for 2 h at 4uC. After incubation, agarose beads were washed in lysing buffer and cracked in loading buffer. Finally, samples were run on SDS-PAGE and detection of MAPKKs was assessed. To block the cysteines, proteins were incubated with 4vinylpyridine (Sigma-Aldrich) for 1 h at RT, and then washed and oxidized. In mock experiments recombinant MAPKs kinases were absent.

Mass spectrometry
Human ERK2, p38 or JNK2 proteins (Upstate) oxidized with 0.1-20 mM H 2 O 2 were analyzed by mass spectrometry. We used the mutant non-phosphorylatable ERK K52R to avoid autophosphorylation and its putative effect on ERK oxidation. Tryptic digestion was performed with methylated trypsin to reduce autolysis (Promega, Madison, WI). Digestion products were dried in an APD SpeedVac (ThermoSavant), desalted by ZipTip (Millipore CB 18B , Millipore, Billerica, MA) and resuspended in 60% acetic acid for injection by autosampler (Surveyor, Thermo-Finnigan). Mass analysis was done in a ThermoFinnigan LCQ Deca XP Plus ion trap mass spectrometer equipped with a nanospray ion source (ThermoFinnigan). Proteins were identified with MS/MS Mascot (Matrix Science) search software using TurboSequest. Mascot and Sequest searches allowed for variable modifications as methionine, histidine or tryptophan oxidation (+16 Da) and cysteic acid reduction (+48 Da), and tyrosine and tryptophan nitrosylation (+45 Da). Sequest search also included variable mono and di-oxidation of cysteine (+16 and +32, respectively). Figure S1 Redistribution of MAPKKs upon redox stimulation. Cellular distribution of (A) MEK1/2 (B) MKK4, and (C) MKK3 was followed as in Fig. 3. In (B) and (C), squares represent nuclei and circles represent mitochondria; red and black symbols indicate 1 and 50 mM H 2 O 2 , respectively. (D) On the bases of the experimental findings, a representative scheme proposes a cycle of MPKKs to transport cognate kinases from mitochondria to nucleus. Green circles indicate that MAPK are preferentially bound to upstream MAPKKs at the different redox states, respectively Found at: doi:10.1371/journal.pone.0002379.s001 (7.40 MB TIF)