Redox regulation of Rac1 by thiol oxidation

Abstract

The Rac1 GTPase is an essential and ubiquitous protein that signals through numerous pathways to control critical cellular processes, including cell growth, morphology, and motility. Rac1 deletion is embryonic lethal, and its dysregulation or mutation can promote cancer, arthritis, cardiovascular disease, and neurological disorders. Rac1 activity is highly regulated by modulatory proteins and posttranslational modifications. Whereas much attention has been devoted to guanine nucleotide exchange factors that act on Rac1 to promote GTP loading and Rac1 activation, cellular oxidants may also regulate Rac1 activation by promoting guanine nucleotide exchange. Herein, we show that Rac1 contains a redox-sensitive cysteine (Cys¹⁸) that can be selectively oxidized at physiological pH because of its lowered pK_a. Consistent with these observations, we show that Rac1 is glutathiolated in primary chondrocytes. Oxidation of Cys¹⁸ by glutathione greatly perturbs Rac1 guanine nucleotide binding and promotes nucleotide exchange. As aspartate substitutions have been previously used to mimic cysteine oxidation, we characterized the biochemical properties of Rac1^C18D. We also evaluated Rac1^C18S as a redox-insensitive variant and found that it retains structural and biochemical properties similar to those of Rac1^WT but is resistant to thiol oxidation. In addition, Rac1^C18D, but not Rac1^C18S, shows greatly enhanced nucleotide exchange, similar to that observed for Rac1 oxidation by glutathione. We employed Rac1^C18D in cell-based studies to assess whether this fast-cycling variant, which mimics Rac1 oxidation by glutathione, affects Rac1 activity and function. Expression of Rac1^C18D in Swiss 3T3 cells showed greatly enhanced GTP-bound Rac1 relative to Rac1^WT and the redox-insensitive Rac1^C18S variant. Moreover, expression of Rac1^C18D in HEK-293T cells greatly promoted lamellipodia formation. Our results suggest that Rac1 oxidation at Cys¹⁸ is a novel posttranslational modification that upregulates Rac1 activity.

1. Introduction

Rac1 is a member of the Rho subclass of Ras superfamily GTPases. It functions as a molecular switch by cycling between active GTP- and inactive GDP-bound states to control the timing and specificity of cellular pathways that regulate diverse cell functions, including gene expression, cell motility, cell morphology, and cell cycling [1]. Owing to the slow intrinsic rates of nucleotide exchange and hydrolysis, temporal regulation of Rac1 activity requires modulatory factors, such as guanine nucleotide exchange factors (GEFs)², which facilitate exchange of GDP for GTP; GTPase-activating proteins (GAPs), which catalyze GTP hydrolysis; and guanine nucleotide dissociation inhibitors (GDIs), which prevent GDP dissociation and sequester Rac1 away from cell membranes [2,3]. In addition, Rac1 is spatially and temporally regulated by various posttranslational modifications, including C-terminal lipidation [4–6], phosphorylation [7], ubiquitination [8,9], and SUMOylation [10]. Rac1 is an essential protein [11] that plays a critical role in regulating multiple cellular processes. Its dysregulation is correlated with many diseases, including cancer [12,13], osteoarthritis [14], cardiovascular disease [15], and neurological disorders [16]. Until recently, upregulation of the activity of Rho family GTPases in cancer was believed to result solely from altered Rac1 expression levels and/or aberrant expression or regulation of Rac1 GEFs, GAPs, and GDIs [17,18]. However, oncogenic mutations have been discovered in Rac1, Rac2, and Cdc42 [19]. In particular, a Rac1^P29S mutant has recently been identified in 9% of primary melanomas [20], which drives oncogenic transformation in melanocytes due to increased nucleotide cycling [21–23], and Rac2^P29L has been sporadically observed in human melanoma and breast cancers [24]. Another mutant, Rac1^N92I, was identified in the fibrosarcoma cell line HT1080 [24]. Further, the Rac1 splice variant Rac1b also promotes fast nucleotide cycling and has been shown to promote cellular transformation [25] as well as sustaining tumor survival [26]. Whereas most oncogenic mutations identified in Ras family proteins cause chronic activation by impairing GAP-mediated GTP hydrolysis [27], oncogenic mutations in Rho family GTPases promote increased guanine nucleotide exchange in a GEF-independent manner, which results in constitutive activation [19].

In addition to GEF-mediated regulation and mutation, reactive oxygen and nitrogen species (ROS and RNS) can directly facilitate guanine nucleotide exchange and Rac1 activation [28]. Although ROS and RNS are best known for their role in oxidative stress, in which they induce DNA damage as well as oxidize lipids and proteins [29], ROS and RNS have been shown to regulate cellular signaling at physiological levels [30,31]. Cellular oxidants, such as nitrogen dioxide $\left({\text{NO}}_2^{•}\right)$, superoxide $\left({\mathrm{O}}_2^{•-}\right)$, hydrogen peroxide (H₂O₂), and peroxynitrite (ONOO⁻), have been shown to specifically react with protein thiols and alter protein activity [32]. In fact, we have previously shown that oxidants, such as nitrogen dioxide, superoxide, and peroxide, can react directly with Rac1 Cys¹⁸ and regulate Rac1 activity by promoting guanine nucleotide exchange in vitro [28]. Moreover, Rac1 can be activated by exogenous addition of peroxide to HeLa cells [33].

Osteoarthritis is a degenerative joint disease that is characterized by high levels of oxidative stress [34], and increased Rac1 activity has been shown to regulate disease progression [14]. Rac1 activity has also been found to be elevated in human osteoarthritis cells [14,35]. Moreover, in primary chondrocytes, Rac1 activation increases the expression of matrix metalloproteinase 13 (MMP-13) [35], an enzyme known to play a role in cartilage matrix degradation. In both of these studies, ectopic expression of inactive Rac1 or use of a chemical inhibitor to Rac1 reduced the level of cartilage destruction and disease progression, indicating a direct role for Rac1 activity in osteoarthritis.

In addition to direct regulation of Rac1 activity by ROS and RNS, Rac1 associates with and can regulate enzymes that produce ROS and RNS. Rac1 binds to and activates NADPH oxidase (NOX) isoforms (Nox1, Nox2, and Nox3) [36,37]. Once activated, the NOX complex produces superoxide, a common cellular ROS. Superoxide has a short half-life and can be reduced by superoxide dismutase 1 (SOD1) to peroxide, a less reactive ROS [38]. Rac1 has also been shown to directly interact with SOD1 in a redox- and nucleotide-dependent manner [39]. In a study by Harraz et al. [39], dithiothreitol (DTT)-reduced GTPγS-loaded Rac1 associated with SOD1, whereas exposure of GTPγS-bound Rac1 to peroxide at concentrations as low as 50 pM reduced association with SOD1 in vitro. These data suggest that oxidation of Rac1 can regulate its activity and alter interactions with regulatory proteins and effectors. Rac1 can also associate with and directly regulate the activity of endothelial and neuronal nitric oxide synthases (eNOS and nNOS) [40]. Nitric oxide (NO^•) generated from NOS is involved in numerous physiological processes [41], including vascular function [42], neurotransmission [43], and pathogen defense [44]. As Rac1 can regulate both nitric oxide production from NOS enzymes and superoxide production from NOX complexes, Rac1 likely modulates peroxynitrite generation in cells. Peroxynitrite is a potent oxidant that can easily cross membranes and directly oxidize thiols and iron–sulfur centers in proteins via direct (nonradical-mediated oxidation) or indirect (radical-mediated oxidation) mechanisms [45].

Given that Rac1 colocalizes with and modulates the activity of several redox enzymes, including Nox1/2/3, SOD1, eNOS, and nNOS, we investigated the effects of oxidative modification on Rac1 activity in vitro and in cells. We found that Rac1 has a redox-sensitive cysteine in the phosphoryl-binding loop (p-loop) that can be selectively oxidized by glutathione in vitro, and Rac1 was observed to be glutathiolated in primary chondrocytes. Further, Rac1 Cys¹⁸ glutathiolation perturbs guanine nucleotide binding and promotes guanine nucleotide cycling. We prepared a Rac1^C18D variant designed to place a negative charge in the nucleotide-binding pocket, similar to sulfinic/sulfonic acid oxidation and glutathiolation, and found that the Rac1^C18D variant shows greatly increased nucleotide dissociation rates in vitro, similar to glutathiolated Rac1 (Rac1^S–SG). Further, the Rac1^C18D variant shows increased activation in HEK-293T cells and enhances lamellipodia formation in Swiss 3T3 cells. We have also generated and char-acterized Rac1^C18S, a redox-insensitive Rac1 variant that is resistant to glutathiolation. Taken together, our findings suggest that Rac1 oxidation may promote enhanced nucleotide cycling in a GEF-independent manner, which could lead to increased Rac1 cellular activation. Thus, given the role of Rac1 in regulating cellular oxidant production, dysregulation of Rac1 via oxidative posttranslational modifications may contribute to a variety of disease states in which the redox state is altered.

2. Materials and methods

2.1. Rac1 glutathiolation in primary chondrocytes

Human articular chondrocytes were isolated from normal articular cartilage obtained from tissue donors and cultured as previously described [35]. Confluent cultures were made serum-free overnight before treatment with 25 μM menadione for 0, 10, or 30 min to induce ROS production with or without 100 ng/ml insulin-like growth factor-1 (IGF-1) for 10 min. Cell lysates were prepared as previously described [35] with the addition of 20 mM iodoacetamide and 200 U/ml bovine liver catalase to the lysis buffer. Cell lysates with equal amounts of total protein were immunoprecipitated with antibodies to Rac1 (clone 23A8 from EMD Millipore; Darmstadt, Germany) using the Pierce Classic IP Kit (Thermo Scientific; San Jose, CA, USA). Immunoprecipitated proteins were separated by SDS–PAGE under nonreducing conditions and then immunoblotted with a mouse monoclonal antibody (Arbor Assays; Ann Arbor, MI, USA) to detect protein glutathiolation, which was followed by stripping the blot and reprobing for Rac1.

2.2. Expression and purification of recombinant proteins

Human Rac1^WT (1–188, C178S) and the Rac1 Cys¹⁸ variants were cloned into pET15b (EMD Millipore), transformed into Escherichia coli BL21 (DE3) Rosetta2 cells (Stratagene; La Jolla, CA, USA), and grown at 37 °C to 0.6 OD₆₀₀. Rac1 expression was induced upon adding 1 mM isopropyl-β-D-1-thiogalactopyrano-side. The cells were grown for 4 h at 37 °C before lysis in 50 mM KH₂PO₄ (pH 7.5), 150 mM NaCl, 1 mM MgCl₂, 10 μM GDP, and 5 mM β-mercaptoethanol (βME). All Rac1 and Rac1 Cys¹⁸ variants were purified using a Ni–NTA column (Qiagen; Venlo, Limburg, The Netherlands) with a linear elution gradient from 0 to 300 mM imidazole. For longer term storage, purified Rac1 proteins were stored in 50% glycerol at −20 °C. The RhoGAP domain (residues 244–431) was cloned into the pQlinkH vector (Addgene), and human Tiam1 (GEF domain containing the DH/PH domain, residues 1040–1397) was cloned into pET28a. Similar to Rac1 expression and purification, these constructs were transformed into E. coli BL21 (DE3) Rosetta2 cells. The cells were lysed in 20 mM Na₂HPO₄ (pH 7.4), 150 mM NaCl, 20 mM imidazole, and 5 mM βME and purified using Ni–NTA agarose affinity chromatography (Qiagen).

2.3. Rac1 glutathiolation

Oxidized glutathione (GSSG) was added to Rac1 at 1000-fold excess for 15–60 min at 25 or 37 °C (time and temperature were varied to increase yield) in glutathiolation buffer (50 mM KH₂PO₄ (pH 6.5), 150 mM NaCl, 5 mM MgCl₂, 50 μM GDP, and 0.1 mM diethylenetriaminepentaacetic acid (DTPA)). The sample was dialyzed against prechilled buffer (20 mM KH₂PO₄ (pH 6.5), 50 mM NaCl, 5 mM MgCl₂, 10 μM GDP, and 0.1 mM DTPA) overnight.

2.4. Mass spectrometry of unmodified Rac1, glutathiolated Rac1, and ABD-modified Rac1

Rac1 mass measurements were performed on an LTQ Orbitrap Velos mass spectrometer (Thermo Scientific). The mass analysis of intact Rac1 samples was achieved in full-MS, single-ion monitoring, and electron transfer dissociation–tandem mass spectrometry (ETD–MS/MS) modes with a resolution of 120,000 at m/z 400 Da. The intact MS spectra were mass deconvoluted using ProMass, and ETD–MS/MS product ion spectra were processed manually by assigning sequence ions to theoretical masses corresponding to glutathiolated Rac1 or ABD-modified Rac1.

2.5. GDP dissociation assay

Rac1 was preloaded with 2′-/3′-O-(N′-methylanthraniloyl)guanosine-5′-O-diphosphate (mant-GDP; Biolog; Bremen, Germany) in 20 mM Tris (pH 7.5), 50 mM NaCl, 200 mM (NH₄)₂SO₄, and 0.1 mM ethylenediaminetetraacetic acid (EDTA) for 1 h at 37 °C. The reaction was incubated on ice for at least 1 h upon adding 20 mM MgCl₂. The excess nucleotide was removed, and Rac1 was buffer exchanged into 20 mM Tris (pH 7.5), 150 mM NaCl, and 5 mM MgCl₂. Nucleotide dissociation was initiated by addition of a 1000-fold excess of unlabeled GDP, and the nucleotide dissociation rate was determined by monitoring the fluorescence emission at 435 nm (excitation 365 nm) using an LS50B spectrophotometer (PerkinElmer; Waltham, MA, USA). Where indicated, the minimal catalytic fragment of the Rac1 GEF Tiam1 containing the DH/PH domain was used to stimulate Rac1 nucleotide dissociation [46]. All dissociation experiments were performed in triplicate. The fluorescent nucleotide dissociation curves were fit to a one-phase exponential decay equation using GraphPad Prism 5.0.

2.6. GTP hydrolysis

Rac1 single-turnover GTP hydrolysis rates were determined in the presence and absence of the RhoGAP domain of p50 rhoGAP (1:1000 GAP:Rac1) essentially as described [47]. Inorganic phosphate was removed from all buffers using the “phosphate mop,” consisting of 0.5 mM inosine in each of the following buffers and dialyzing the buffer in the presence of 1 unit of nucleoside phosphorylase [48]. Rac1 was loaded with GTP by incubating with 10-fold excess GTP at 37 °C for 1 min in 20 mM HEPES (pH 8), 20 mM (NH₄)₂SO₄, 1 mM EDTA, and 0.1 mM DTPA. Excess GTP was removed using a PD-10 column (GE Healthcare). The reaction was performed in triplicate using a buffer containing 20 mM Tris (ph 7.4), 50 mM NaCl, 0.1 mM EDTA, and 0.5 mM inosine. GTP hydrolysis was initiated by adding 100 μM MgCl₂ to a sample containing 50 μM Rac1 and FlipPi 5U (Addgene). FlipPi undergoes a conformational change upon binding inorganic phosphate, which alters its intrinsic fluorescence resonance energy transfer (FRET), as previously described [49]. Phosphate released from the hydrolysis of GTP to GDP was quantified to determine Rac1 GTP hydrolysis rates by monitoring the FlipPi FRET signal change using a Spec-tramax M5e spectrometer (excitation 415 nm and the emission ratio for 475 and 515 nm). The data were normalized and fit to a single association exponential (Prism 5.0; GraphPad; San Diego, CA, USA; n=2).

2.7. Rac1 Cys¹⁸ thiol pK_a measurements

Rac1^WT and Rac1^C18S were dialyzed into reducing buffer (15 mM HEPES (pH 8.0), 50 mM NaCl, 5 mM MgCl₂, 10 μM GDP, and 10 mM DTT), and the protein was incubated under reducing conditions for 30 min. Rac1 was then buffer exchanged using a Centricon concentrator (15-kDa molecular weight cut-off, Millipore) into nonreducing buffer (15 mM MES (pH 6.5), 30 mM NaCl, 5 mM MgCl₂, 200 μM DTPA, and 10 μM GDP). Rac1 was diluted into a mixed buffer system with pH values ranging from 5.5 to 8.5; each buffer contained 30 mM MES, 30 mM MOPS, 30 mM Tricine, 5 mM MgCl₂, 50 mM KCl, and 200 μM DTPA. ABD-f (4-(aminosul-fonyl)-7-fluoro-2,1,3-benzoxadiazole; Anaspec; Fremont, CA, USA) was added (1 mM) to initiate cysteine thiol modification. The reaction rate was determined by monitoring ABD-f fluorescence (excitation 389 nm, emission 513 nm) using a Spectromax M5e spectrometer (Molecular Devices; Sunnyvale, CA, USA). The initial reaction rates were fitted using linear regression analysis (Prism 5.0; GraphPad).

2.8. Rac1 circular dichroism and thermal stability

Circular dichroism data were collected on a JASCO J-815 CD spectrometer (Oklahoma City, OK, USA) with a JASCO Peltier device and water bath to control the temperature. Experiments were performed in a 1-mm cuvette at a protein concentration of 15 μM in 10 mM potassium phosphate (pH 6.5). Far-UV scans were from 200 to 280 nm. Thermal denaturation of Rac1 and Rac1 variants were monitored at 222 nm to estimate the protein melting temperature. The temperature ramp rate was 1 °C/min and data points were collected every 1 °C. All data are reported in units of mean residue ellipticity, which was calculated as follows: [θ]_MRE = (θ_raw × MRW)/(10 × c × l), where θ_raw is the ellipticity in degrees, MRW is (molecular mass in kilodaltons)/(No. of residues – 1), c is the protein concentration in g/ml, and l is the pathlength of the cuvette in cm, according to [50].

2.9. NMR experiments

Rac1 was expressed and purified as described above except that the cells were grown in ¹⁵N-enriched M-9 minimal medium. Two-dimensional (2D) ¹H–¹⁵N HSQC (heteronuclear single-quantum coherence spectroscopy) NMR experiments were performed using a Varian Inova 700-MHz spectrometer with a cryoprobe. The sample contained 200 μM Rac1, Rac1^C18S, or Rac1^C18A at 25 °C in 50 mM Tris maleate (pH 6.8), 50 mM NaCl, 5 mM MgCl₂, 50 μM GDP, 0.1 mM DTPA, and 1 mM DTT. The Rac1^C18D variant was collected on a Bruker 700-MHz spectrometer (Billerica, MA, USA). Two-dimensional ¹H–¹⁵N HSQC NMR spectra were collected and recorded using a 2500-Hz ¹⁵N spectral width and 512 complex points. The NMR data were processed using NMR Pipe and NMRViewJ [51,52].

2.10. Cell lines, plasmids, and reagents

HEK-293T cells (from the American Type Culture Collection) and Swiss 3T3 cells (a gift from Alan Hall, Memorial Sloan Kettering Cancer Center) were grown in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% fetal bovine serum (Sigma; St. Louis, MO, USA) and maintained at 37 °C in 5% CO₂ [53]. Keith Burridge (University of North Carolina) provided full-length human Rac1 and the full-length Rac1^C18S variant, which were cloned into the pCMVJ3 vector for mammalian expression; pCMVJ3-Rac1^C18D was generated from Rac1^WT using PCR-based mutagenesis.

2.11. PAK pull-down assays for Rac1-GTP in HEK-293T cells

Levels of active, GTP-bound Rac1 were assessed using pull-down assays with the PAK1 p21-binding domain (GST-bound PAK-PBD, a gift from Keith Burridge) as described previously [54]. Briefly, HEK-293T cells were transiently transfected with Rac1 expression plasmids using the TransIT transfection reagent (Mirus; Madison, WI, USA) according to the manufacturer’s instructions. The next day or at 80–90% confluence, the cells were starved in serum-free DMEM for 3 h. Next, the cells were washed twice with ice-cold phosphate-buffered saline (PBS; 5.4 mM KCl, 1.7 mM KH₂PO₄, 13 mM NaCl, and 5.4 mM Na₂HPO₄ (pH 7.4)) and lysed in magnesium lysis buffer (50 mM Tris (pH 7.5), 10 mM MgCl₂, 150 mM NaCl, 1% NP-40, 10% glycerol, and 0.25% Na deoxycholate) with protease inhibitors. Equal volumes were removed from each lysate for total protein analysis. Glutathione (GST) agarose beads containing 20 μg of GST–PAK-PBD were added to each lysate and incubated at 4 °C for 60 min. Agarose–GST–PAK-PBD and associated Rac1 were pelleted and washed three times with 500 μl of wash buffer (25 mM Tris (pH 7.5), 40 mM NaCl, and 30 mM MgCl₂). The final pellets were suspended in 1× protein sample buffer and resolved using SDS–PAGE. Rac1 protein and variants were detected through immunoblotting using an anti-Rac antibody (Millipore). Equal protein loading was confirmed using anti-tubulin (Sigma–Aldrich), and the Rac1^C18S and Rac1^C18D data were normalized to the Rac1^WT data.

2.12. Cytoskeleton assays in Swiss 3T3 cells

Swiss 3T3 cells were plated (5000 cells/well in 12-well plates) on coverslips previously coated with fibronectin (2.5 μg/ml) for 30 min at room temperature. The next day, Myc-tagged Rac1 constructs were transiently transfected into the cells using TransIT (Mirus) according to the manufacturer’s instructions. Twenty-four hours after transfection, the cells were fixed and stained with phalloidin and for the Myc-tag, as previously described [55]. Briefly, the cells were fixed in 4% paraformaldehyde (Electron Microscopy Sciences; Hatfield, PA, USA) overnight at 4 °C, permeabilized in 0.2% Triton X-100 (Sigma) in PBS for 5 min, and incubated with anti-Myc 911B antibody (1:500, Cell Signaling) for 1 h, followed by secondary AlexaFluor 488 anti-mouse antibody (1:500, Invitrogen) for 2 h and AlexaFluor 568 phalloidin (Invitrogen, 1:40 in PBS) for 30 min. The cells were incubated in the dark and rinsed in PBS three to five times between each step. Coverslips were mounted using 6 μl Vectashield with DAPI (Vector Laboratories; Burlingame, CA, USA). Cells were visualized and counted blindly for lamellipodia using a Nikon Eclipse TS100 microscope with a 40× objective. Representative images were recorded using a Zeiss LSM 710 confocal laser-scanning microscope with a 63× oil objective.

3. Results

We have previously shown that Rac1 activity can be directly modulated by ROS and RNS [28]. Our earlier studies focused on radical-mediated regulation of Rac1 activity, in which we found that oxidants ( ${\text{NO}}_2^{•}$ and ${\mathrm{O}}_2^{•-}$) capable of generating a Cys¹⁸ thiyl radical caused guanine base oxidation and nucleotide dissociation in Rac1; however, we also observed that peroxide, a nonradical oxidant, increased the intrinsic rate of exchange by ~10-fold [28]. As S-glutathiolation is commonly observed after thiyl radical formation because of the high cellular glutathione concentration [56], we expanded on these studies by examining whether glutathione modification can alter Rac1 activity. Given that Rac1 Cys¹⁸ is located in the nucleotide-binding pocket and forms multiple interactions with the guanine nucleotide ligand, we postulated that Cys¹⁸ is redox-sensitive, and oxidation of this thiol could alter guanine nucleotide binding and the Rac1 activation state.

3.1. Rac1 is glutathiolated in primary chondrocytes and is specifically glutathiolated at Cys¹⁸ in vitro

Rac1 has previously been shown to contribute to the development of osteoarthritis, probably because of its ability to upregulate MMP-13 expression in articular chondrocytes [14,35]. Therefore, we used primary human chondrocytes to examine whether endogenous Rac1 is glutathiolated. The cells were treated with menadione, which induces ROS production, in the absence and presence of the growth factor IGF-1, to examine the effects of ROS on IGF signaling. As shown in Fig. 1, menadione treatment, but not IGF-1, increased the level of Rac1 glutathiolation. Detection of glutathiolated Rac1 was greatest at 10 min after menadione addition and declined at 30 min.

Fig. 1.

Rac1 glutathiolation in human chondrocytes. Primary human chondrocytes were treated for 10 or 30 min with 25 μM menadione to induce oxidative stress in the absence and presence of 100 ng/ml IGF-1. Cell lysates immunoprecipitated with a monoclonal antibody to Rac1 were run on a nonreducing gel and immunoblotted with a monoclonal antibody to glutathione. The blot was stripped and reprobed with the Rac1 antibody.

Given that Rac1 Cys¹⁸ is solvent accessible (Supplementary Table 1) and sensitive to ROS and RNS [28], we explored whether Cys¹⁸ in Rac1 can be specifically glutathiolated in vitro. To generate glutathiolated Rac1 (Rac1^S–SG), Rac1^WT was treated with 1000× GSSG. The reaction products were analyzed by MS. Fig. 2 shows the intact mass analysis of Rac1^WT treated with GSSG at pH 6 and 7.5, as well as the subsequent characterization of Rac1 with a single glutathione modification by top-down mass spectrometry. We were able to semi-quantitatively assess the amount of adduct formed by comparing the normalized peak intensity of unmodified Rac1 to the intensity of the Rac1 glutathiolated peak. These analyses also show that the amount of glutathiolated Rac1 is dependent on the pH of the reaction. Fig. 2A shows the deconvoluted MS of Rac1 treated with GSSG at pH 7.5 that resulted in ~25% glutathione adduct (normalized to base peak of Rac1), and Fig. 2B shows the deconvoluted MS of Rac1 treated with GSSG at pH 6 that resulted in ~10% glutathione adduct (normalized to base peak of Rac1). The twofold drop in glutathiolation observed in the MS data of intact Rac1 suggests that a single cysteine residue in Rac1 is modified in a pH-dependent manner. These observations are consistent with data in Supplementary Fig. 1, in which ESI–MS performed on Rac1^WT and Rac1^C18S show glutathiolation for Rac1^WT but not Rac1^C18S, indicating that Cys¹⁸ is the primary site of glutathiolation in Rac1.

Fig. 2.

Mass spectrometry characterization of Rac1^WT glutathione adducts. (A and B) Mass deconvoluted zero-charge spectra of the ESI full-MS spectra (Supplementary Fig. 1) associated with intact Rac1 treated with 1000× glutathione at pH 6 and 7.5, respectively. Note: A single glutathione adduct is seen at pH 6.0 and 7.5, with more adducts observed at higher pH. (C and D) Top-down product ion MS/MS spectra resulting from the ETD of [M+16H+G]¹⁶⁺ (glutathiolated Rac1, Z = 16+, m/z = 1259.6 Da) with fluoranthene anions. Note: c- and z-type ions, nondissociated precursors (electron transfer with no dissociation), and precursors with fluoranthene (m/z = 202 Da) adducts (+F) of Rac1 were observed. As glutathione is prone to cleavage at the disulfide bond during ETD, some neutral loss glutathione ions were detected. In addition, z-type ions are prone to radical-mediated oxide adduction in ETD, which were detected. (E) ETD sequence ions mapped to Rac1 with putative glutathione sites highlighted by asterisks. Confident assignment of c- and z-type ions z₁₅₄, z₁₄₅, z₉₉, and c₄₅+G leads to the localization of the glutathione adduct to either Cys⁶ or Cys¹⁸ of Rac1.

The site of Rac1 glutathiolation was characterized by top-down mass spectrometry by subjecting the [M+16H+G]¹⁶⁺ ion (Z=16+ of Rac1 containing a single glutathione) to ETD. The resulting MS/MS product ion spectrum consists of the major c- and •z-type fragment ions, shown at two different mass ranges for clarity. Data in Fig. 2C show products spanning m/z 800–1500 Da, and Fig. 2D shows all product ions spanning m/z 1500–2000 Da. The lower m/z region resulted in mostly singly charged c- and •z-type product ions from C-terminal residues, whereas the N-terminal fragment ion c₄₅+G contained the only glutathione-containing product ion. Most of the N-terminus lacked structurally informative fragments because of the lower number of basic amino acid residues. In addition, the neutral loss of glutathione during ETD was occasionally observed. Conversely, the higher m/z region produced structurally rich c-type ions, i.e., c₈₇+G, and a number of other larger multiply charged c-type ions that contained glutathione (Fig. 2E). Our ability to detect glutathiolated Rac1 fragments was probably due to preservation of the covalently attached disulfide bond of glutathione upon competitive fragmentation along the N–Cα backbone. These ions, along with complementary multiply charged •z-type fragment ions lacking glutathione (•z₁₅₄, •z₁₄₅, •z₉₉), lead to unambiguous localization of glutathiolation to either Cys⁶ or Cys¹⁸. In further support of site-specific glutathiolation at Cys¹⁸, a Rac1 variant that lacks a cysteine at position 18 (Rac1^C18S), but shows biochemical properties similar to Rac1^WT, is not covalently modified by glutathiolation (Supplementary Fig. 1). Taken together, our data indicate that treatment of Rac1^WT with GSSG results in specific glutathiolation of Rac1 Cys¹⁸.

3.2. Rac1 Cys¹⁸ glutathiolation perturbs guanine nucleotide binding and enhances the intrinsic GDP dissociation rate

As Cys¹⁸ is located in the p-loop of Rac1 and directly interacts with the bound guanine nucleotide, we hypothesized that glutathiolation at this site may interfere with nucleotide binding. To determine whether Rac1 glutathiolation alters guanine nucleotide dissociation, Rac1^S–SG was preloaded with fluorescent GDP (mant-GDP), and the rate of GDP dissociation was determined. As shown in Fig. 3, oxidative modification of Rac1 Cys¹⁸ with glutathione enhances the rate of GDP dissociation by 200-fold. Adding the GEF domain (DH/PH) of Tiam1 to Rac1^S–SG did not increase the GDP dissociation rate. However, as the intrinsic rate of GDP dissociation is rapid under our experimental conditions, it is unclear whether we have the dynamic range to detect GEF-mediated enhancement.

Fig.3.

Oxidation of Rac1 Cys¹⁸ and Rac1 Cys¹⁸ variants alters guanine nucleotide exchange but not GTP hydrolysis. (A) Intrinsic mant-GDP dissociation curves for Rac1^WT, Rac1^C18S, Rac1^C18D, and Rac1^S–SG. The data were fit to a single exponential and standard errors were determined using Prism 5.0 (n = 3). (B) Bar graph of rates of intrinsic nucleotide exchange with GEF-mediated dissociation data (using the DH/PH domains of Tiam1) where applicable. (C) Graph of intrinsic and GAP-mediated single-turnover hydrolysis for Rac1^WT, Rac1^C18S, and Rac1^C18D. The RhoGAP domain of p50 rhoGAP was used for the GTP hydrolysis measurements. The data were fit to a single exponential and standard errors were determined using Prism 5.0 (n = 2). (D) Bar graph of rates of nucleotide hydrolysis with GAP-mediated hydrolysis data (using p50 rhoGAP) where applicable. Rates of exchange and hydrolysis are presented in Table 1.

To better understand the enhanced nucleotide exchange properties associated with glutathiolated Rac1, we generated Rac1 C18D, C18S, and C18A variants. The Rac1^C18D variant was generated because a Cys→Asp substitution has previously been shown to mimic a cysteine-to-sulfenic/sulfinic acid modification in recoverin [57]. As both Rac1^C18D and Rac1^S–SG introduce a negative charge into the phosphoryl-binding pocket, we also made more conservative Cys¹⁸ substitutions, including Rac1^C18S and Rac1^C18A. All variants were preloaded with fluorescent GDP (mant-GDP), and Rac1 GDP dissociation was determined by measuring the decrease in mant-GDP fluorescence as a function of time (Fig. 3A). The GDP dissociation rates of Rac1 and the Cys¹⁸ variants were also measured in the presence of the GEF domain of Tiam1. Similar to glutathiolated Rac1, Rac1^C18D shows a greatly enhanced intrinsic rate of nucleotide dissociation that was approximately 200-fold faster than that of Rac1^WT and was not further increased by addition of Tiam 1. In contrast, Rac1^C18S shows a similar GDP dissociation rate compared to Rac1^WT, and Rac1^WT and Rac1^C18S were similarly responsive to GEF-mediated GDP dissociation. Consistent with previous studies of Cdc42^C18A [58], the Rac1^C18A variant enhances the intrinsic rate of GDP dissociation by 11-fold. These results indicate that the Cys¹⁸ thiol side chain plays a role in stabilizing nucleotide binding (Fig. 3B and Table 1).

Table 1.

Intrinsic and GEF-mediated GDP dissociation rates for Rac1^WT, Rac1^C18S, Rac1^C18A, Rac1^C18D, and Rac1^S–SG.

Rac1 construct	koff (s−1) intrinsic	Fold increase (relative to WT)	koff (s−1) with Tiam1
Rac1WT	0.71 ± 0.02 × 10−4	–	5.02 ± 0.01 × 10−4
Rac1C18S	0.81 ± 0.00 × 10−4	1.1	4.28 ± 0.03 × 10−4
Rac1C18A	7.58 ± 0.02 × 10−4	10.7	n/d
Rac1C18D	157.87 ± 1.38 × 10−4	222.4	114.73 ± 0.61 × 10−4
Rac1S–SG	167.50 ± 25.13 × 10−4	235.9	n/d

In-cell regulation of Rac1 activity requires exchange of GDP for GTP to activate Rac1 and GTP hydrolysis for inactivation. To characterize the effects of Rac1 Cys¹⁸ variants on intrinsic and GAP-mediated GTP hydrolysis, we determined the rates of GTP hydrolysis of Rac1^WT, Rac1^C18D, and Rac1^C18S in the presence and absence of the minimal catalytic domain of p50 rhoGAP. Single-turnover GTP hydrolysis rates were determined by monitoring phosphate release upon GTP hydrolysis by detecting the FRET change associated with the phosphate-binding protein FlipPi 5U. As shown in Fig. 3C and quantified in Table 2, mutating Rac1 Cys¹⁸ to either Asp or Ser did not significantly affect either the intrinsic or the GAP-mediated GTP hydrolysis rates. These data indicate that perturbation of Rac1 Cys¹⁸ by mutation, including the oxidation mimetic, does not alter GAP-mediated down regulation in vitro.

Table 2.

Intrinsic and GAP-mediated GTP hydrolysis rates for Rac1^WT, Rac1^C18S, and Rac1^C18D.

Rac1 construct	Intrinsic GTP hydrolysis (s−1)	p50 rhoGAP GTP hydrolysis (s−1)
Rac1WT	0.90 ± 0.07 × 10−3	4.8 ± 0.1 × 10−3
Rac1C18S	1.50 ± 0.11 × 10−3	5.5 ± 0.3 × 10−3
Rac1C18D	0.26 ± 0.01 × 10−3	4.9 ± 0.2 × 10−3

3.3. The Rac1 Cys¹⁸ thiol has a depressed pK_a and can be selectively modified by ABD-f at physiological pH

As Rac1 Cys¹⁸ can be selectively modified by oxidized glutathione at physiological pH, we investigated whether the pK_a of the Rac1 Cys¹⁸ thiol is altered relative to a typical free cysteine thiol by measuring the cysteine reactivity of Rac1^WT and Rac1^C18S to ABD-f. We used the thiol-modifying reagent ABD-f, which preferentially reacts with the thiolate form of cysteine [59], to measure thiol reactivity over a wide pH range, as described previously by us for Ras [60]. ABD-f fluorescence was measured from pH 5.5 to 8.5 for Rac1^WT and Rac1^C18S (Fig. 4A). The rate of ABD-f reactivity for Rac1^WT was normalized to produce a pH titration curve for Rac1 Cys¹⁸ (Fig. 4B) and resulted in a pK_a for the Rac1 Cys¹⁸ thiol of ~7.25, which is approximately 1.5 pH units lower than a typical pK_a for a free cysteine [61]. No significant reactivity was detected for Rac1^C18S.

Fig. 4.

Rac1 Cys¹⁸ has an altered pK_a and populates the thiolate state at physiological pH. (A) ABD-f reactivity data for Rac1^WT over a pH range of 5.5–8.5. (B) Using the initial rate of modification by ABD-f to Rac1^WT, the initial rate of modification was plotted vs the pH to determine the pK_a of the Cys¹⁸ thiol. The resulting curve was fit to a Boltzmann sigmoidal curve to determine the inflection point, indicative of the pK_a, using GraphPad Prism 5.0. Curve fitting resulted in an estimated pK_a of 7.25 for Rac1 Cys¹⁸.

The Rac1 construct (Rac1 1–188, C178S) contains five cysteines, some of which are partially solvent accessible (Supplementary Table 1). Fig. 5 shows the intact mass analysis of Rac1^WT and Rac1^C18S that were reacted with ABD-f under the conditions used for glutathione modification. A single ABD-f modification site was identified by top-down mass spectrometry. The deconvoluted mass spectra of Rac1^WT shown in Fig. 5A consists of major peaks corresponding to unmodified intact Rac1, Rac1 with a single ABD adduct mass, and Rac1 with a single deoxygenated adduct mass. Conversely, the deconvoluted mass spectra of the Rac1^C18S variant, in Fig. 5B, show peaks corresponding to unmodified intact Rac1^C18S, Rac1^C18S with a sodium adduct, and two unidentified peaks ([M+76] and [M+99]). However, an ABD adduct is not observed for Rac1^C18S, suggesting that the Rac1^WT ABD adduct occurs exclusively at Cys¹⁸. This result is consistent with Rac1^WT glutathiolation MS data (Fig. 2) in which a single glutathione adduct was observed at Cys¹⁸. The ESI–MS spectra for Figs. 5A and 5B are shown in Supplementary Figs. 2A and 2B, respectively. The site of the ABD adduct was further characterized by top-down mass spectrometry. Both [M+18H+ABD]¹⁸⁺ (Z = 18+ of Rac1 containing a single ABD) and [M + 18H]¹⁸⁺ (Z = 18+ of unmodified Rac1) were subjected to ETD to comprehensively assign peaks modified by ABD-f. Fig. 5C shows the ABD-modified Rac1 sequence map annotated with observed major c- and •z-type fragment ions. The c₂₇+ABD, c₅₄+ABD −H₂O, c₁₂₃+ABD, c₁₃₂+ABD, and c₁₃₃+ABD ions as well as the z₁₄₈ ion allow localization of the ABD adduct to either Cys¹⁸ or Cys⁶. The absence of the ABD adduct for the Rac1^C18S variant confirms that Cys¹⁸ is the most likely site of ABD-f modification.

Fig. 5.

MS of Rac1^WT and Rac1^C18S treated with ABD-f. (A) Mass deconvoluted spectrum of ABD-f-treated Rac1^WT. (B) Mass deconvoluted spectrum of ABD-f-treated Rac1^C18S. Note: Rac1^WT has a peak corresponding to an ABD adduct, whereas a peak at the predicted mass (shown by the arrow) for the corresponding ABD adduct of Rac1^C18S is not observed. The unassigned peaks +76 and +99 Da from Rac1^C18S are adducts of unknown origin. (C) ETD map of ABD-modified Rac1^WT. Product ions generated by ETD of ABD-f-treated Rac1 are presented in Supplementary Fig. 3. Note that the N-terminal three residues (GHM) are not part of the native Rac1^WT sequence. The c₂₇+ABD, c₅₄+ABD−H₂O, c₁₂₃+ABD, c₁₃₂+ABD, and c₁₃₃+ABD ions as well as the z₁₄₈ ion allow for the localization of ABD to either Cys¹⁸ or Cys⁶. The absence of the ABD adduct in the Rac1^C18S variant supports Cys¹⁸ as the site of ABD modification.

Supplementary Figure 3 shows the resulting ETD-MS/MS product ion spectra of Rac1 and Rac1+ABD as a tiled view to compare the fragmentation of unmodified and ABD-modified Rac1. Comparative analysis of peaks associated with top and bottom spectra show differences (difference spectral peaks) that assist in narrowing down the fragments containing ABD-f modification. The distinct product ion peaks along with all other observed product ions were used for the annotation of the sequence map shown in Fig. 5C. The major c- and •z-type fragment ions are shown at several different mass ranges for clarity.

Taken together, our data show that Cys¹⁸ is the only cysteine that appreciably reacts with ABD-f or GSSG at physiological pH values. These results indicate that Rac1 Cys¹⁸ has a depressed pK_a, which populates the more reactive thiolate state under physiological conditions.

3.4. Thermal stability and structural analysis

As a decrease in nucleotide binding affinity is correlated with a decrease in stability for the Ras GTPase [62], we used CD to determine the total secondary structure content and thermal stability of Rac1, Rac1^S–SG, and Rac1 variants. By comparing the overall secondary structure and thermal stability, we can evaluate whether Rac1^S–SG and the Rac1 Cys¹⁸ variant disrupts the overall protein fold. As shown in Fig. 6, the overall secondary structure content determined by CD spectroscopy was unchanged for Rac1^S–SG. In contrast, the thermal stability was decreased by 9 °C relative to Rac1^WT. As the nucleotide exchange rate was significantly increased in Rac1^S–SG relative to Rac1^WT, the change in T_m is probably a reflection of decreased nucleotide binding affinity. We also find that the secondary structure content and thermal stability of Rac1^C18D are similar to those of Rac1^S–SG. In contrast to mutations and oxidative modifications that promote fast cycling, the thermal denaturation of the redox-insensitive Rac1^C18S variant was slightly higher than that of Rac1^WT, with a T_m of 69 °C compared to 67 °C for Rac1^WT.

Fig.6.

Oxidation and mutation of Rac1 Cys¹⁸ does not perturb protein secondary structure or stability. (A) CD spectra (scan from 200 to 280 nm) measuring the secondary structure elements of Rac1^WT, Rac1^C18S, Rac1^C18D, and Rac1^S–SG. (B) Thermal denaturation of Rac1^WT, Rac1^C18S, Rac1^C18D, and Rac1^S–SG at 220 nm at temperatures ranging from 25 to 95 °C. T_m values were calculated by fitting the data to a Boltzmann sigmoidal curve. Data are representative of three thermal melts and were fit using GraphPad Prism 5.0.

To determine how mutations at Cys¹⁸ perturb Rac1 in a site-specific manner, we employed NMR on ¹⁵N-enriched Rac1-GDP. We performed a 2D ¹H–¹⁵N HSQC on Rac1^WT, Rac1^C18D, and Rac1^C18S and compared the variants to the HSQC spectrum of Rac1^WT, which has been previously assigned [63]. As we are able to detect one peak for every backbone amide with the exception of proline, this technique allows us to probe whether mutation at Cys¹⁸ causes localized or more global structural perturbations in Rac1. The 2D ¹H–¹⁵N HSQC overlay of Rac1^WT and Rac1^C18D is shown in Fig. 7A, with residues showing chemical shift changes mapped onto the Rac1 structure (pdb 1MH1) (Fig. 7B). Comparison of the NMR spectra of Rac1^C18D and Rac1^WT showed chemical shift differences (22) in a small percentage (14%) of the total amides (chemical shifts greater than one line width). In addition, two amide peaks showed changes in linewidth. As shown in Fig. 7B, residues that undergo chemical shift changes upon mutation of Cys¹⁸ to Asp are highlighted in purple and are primarily localized to the p-loop, switch I, and SAx motif of Rac1 (switch II was not detected in the Rac1^WT NMR spectrum).

Fig. 7.

2D NMR spectra of Rac1 and Rac1 variants. (A) 2D ¹H–¹⁵N HSQC spectra overlay of Rac1^WT (black) and Rac1^C18D (magenta). Peaks that showed line broadening are shown in cyan and peaks with a chemical shift greater than one linewidth are in black. (B) The major chemical shift perturbations (broadening, where the Rac1^C18D resonance linewidth is less than 50% of the corresponding Rac1^WT resonance (shown in light pink), or shifted, where there was no peak detected in Rac1^C18D within one linewidth of a peak in Rac1^WT (shown in magenta) are mapped onto a ribbon diagram of GTP-bound Rac1 (pdb 1MH1). (C) 2D ¹H–¹⁵N HSQC spectra overlay of Rac1^WT (black) and Rac1^C18S (red). Peaks that showed line broadening are shown in cyan and peaks with a chemical shift greater than one linewidth are in black. (D) The major chemical shift perturbations (broadening, where the Rac1^C18S resonance linewidth is less than 50% of the corresponding Rac1^WT resonance (shown in light red), or shifted, where there was no peak detected in Rac1^C18S within one linewidth of a peak in Rac1^WT (shown in red) are mapped onto a ribbon diagram of GTP-bound Rac1 (pdb 1MH1). In (B) and (D), residues with chemical shifts or line broadening are as indicated. The unassigned and undetected residues are indicated in black, and the remaining residues are gray. GTP is depicted as a multicolored stick structure, and Mg²⁺ is shown as a green sphere.

In contrast, only nine residues show chemical shift changes for Rac1^C18S compared to Rac1^WT. Moreover, the residues displaying chemical shift changes correspond to residues proximal to the site of mutation (Fig. 7C). In Fig. 7D, a ribbon diagram of the structure of Rac1 (pdb 1MH1) is presented with residues showing chemical shift perturbations highlighted. Most of the residues that show peak changes are near the site of mutation, indicating that mutation of Cys¹⁸ to Ser does not significantly perturb the structure of Rac1, consistent with our findings that the biochemical properties and secondary structure of Rac1^C18S are not significantly altered compared to Rac1^WT. Further, these data support the use of Rac1^C18S as a redox-insensitive variant of Rac1.

3.5. Rac1^C18D is hyperactivated in HEK-293T cells

Our data indicate that both the Rac1^C18D variant and glutathiolated Rac1 show greatly enhanced rates of GDP dissociation, similar to the Rac1 fast-cycling mutant (P29S) that promotes Rac1-mediated oncogenesis. The Rac1^P29S mutant has a significantly faster nucleotide exchange rate and induces an activated phenotype in COS-7 cells [21]. As Rac1^C18D shows greatly enhanced nucleotide dissociation (>200-fold), we postulated that this oxidative mimetic will populate Rac1 in the active, GTP-bound state in cells. Therefore, we examined the activation state of Rac1^WT, Rac1^C18D, and Rac1^C18S in HEK-293T cells using a PAK1 pull-down assay. As shown in Fig. 8, Rac1^WT and Rac1^C18S show similar levels of association with PAK1-PBD. In contrast, Rac1^C18D showed a 5-fold higher level of activation compared to Rac1^WT and Rac1^C18S, which is consistent with our observation that Rac1^C18D has a greatly enhanced in vitro nucleotide dissociation rate compared with Rac1^WT. As enhanced Rac1 activation has been previously observed in HeLa cells upon H₂O₂ addition [33], our data further support the hypothesis that oxidation at Rac1 Cys¹⁸ can modulate the Rac1 activation state.

Fig. 8.

The Rac1^C18D variant is hyperactivated in HEK-293T cells. Rac1^WT, Rac1^C18S, and Rac1^C18D were transiently expressed in HEK-293T cells and PAK pull-down assays were used to assess the levels of active, GTP-bound Rac1. Active Rac1 was pulled down from cell lysates with PAK-PBD (p21-binding domain) coupled to agarose beads and was detected by immunoblotting for Rac1. A representative pull-down assay (n = 6) is shown.

3.6. Rac1^C18D enhances Rac1-mediated lamellipodia formation in Swiss 3T3 cells

As we determined through PAK1 pull-down assays that Rac1^C18D is more populated in its GTP-bound state compared with Rac1^WT or Rac1^C18S (Fig. 8), we sought to determine if the increased activity of Rac1^C18D could induce a biological phenotype that indicates enhanced Rac1 signaling. A canonical function of active Rac1 is actin cytoskeleton changes that induce lamellipodia formation [64,65]. Therefore, we visualized the actin cytoskeleton through phalloidin staining in Swiss 3T3 cells expressing Rac1^WT, Rac1^C18D, or Rac1^C18S. Cells expressing the Myc-tagged Rac1 constructs were identified by immunofluorescence, and lamellipodia formation in the Rac1-expressing cells was quantified. We determined that Rac1^C18D expression increased lamellipodia formation 5- to 10-fold compared with Rac1^WT and Rac1^C18S (Fig. 9). Therefore, Rac1^C18D can function similar to activated Rac1^WT and promote lamellipodia formation.

Fig. 9.

Rac1^C18D enhances lamellipodia formation in Swiss 3T3 cells. To examine the effect of Rac1 oxidation on its ability to regulate the actin cytoskeleton, Swiss 3T3 cells were transfected with Rac1^WT, Rac1^C18S, or Rac1^C18D. The actin cytoskeleton was visualized using phalloidin staining. (A and B) Rac1^C18D induces lamellipodia formation, whereas Rac1^C18S does not. Twenty-four hours after transfection with Myc-tagged Rac1 constructs, cells were fixed and stained for Myc-tag and phalloidin. Using the Myc-tag antibody to identify cells that express the Rac1 constructs, cells were counted (blindly with respect to the specific Rac1 construct expressed) for the presence or absence of lamellipodia (n = 2, Rac1^WT, 88 cells total; Rac1^C18S, 78 cells; and Rac1^C18D, 92 cells). Representative images were collected using a Zeiss 710 microscope with a 63× oil objective and are shown in (A). The percentage of cells with lamellipodia was quantified (B). Statistical significance was determined using Student’s t test (*p = 0.05); error bars represent SEM.

4. Discussion

Among its many vital functions, the small GTPase Rac1 plays a key role in regulating redox enzymes, such as NOS and NOX, which generate nitric oxide and superoxide, respectively [66]. Moreover, Rac1 has been shown to interact directly with the antioxidant enzyme SOD1 in a redox- and nucleotide-dependent manner [39]. Rac1 has also been shown to be directly regulated by ROS and RNS oxidants in vitro [28] and by exposure to peroxide in REF52 and HeLa cells [33]. These findings suggest that ROS and RNS regulate both Rac1-mediated oxidant production and Rac1 signaling.

Redox regulation by modification of cysteine residues has been shown to regulate the function of diverse types of signaling proteins. For example, the transcription factor NF-κB, which is a key mediator of Rac1 function, is one prominent example of a cell signaling protein that is regulated in part by cysteine oxidation [67,68]. NF-κB can be regulated directly by S-glutathiolation of its own p50 and/or p65 subunit or indirectly by S-glutathiolation of its regulators IKK and IκBα [69]. In addition, p38 MAPK can be activated by reversible cysteine oxidation [70], whereas a number of protein tyrosine phosphatases become inactivated by oxidation of a catalytic cysteine [71]. In addition to these signaling proteins, we have previously reported that thiyl radical formation at Cys¹¹⁸ of the small GTPase Ras can regulate its activity [72,73]. This mechanism of activation appears to play a key role in Ras-mediated tumorigenesis [74]. However, we have previously shown that non-radical-mediated cysteine oxidation (i.e., nitrosation and glutathiolation) of this solvent-accessible cysteine does not alter Ras activity [75,76]. In contrast to Ras, the redox-sensitive cysteine in Rac1 (Cys¹⁸) has direct interactions with the bound nucleotide [77], and oxidation of this cysteine by both radical- and nonradical cysteine oxidation can impair guanine nucleotide binding, resulting in accelerated nucleotide exchange [28] and upregulation of Rac1 activity [33]. Given that cellular Rac1 colocalizes with enzymes that generate both ROS and RNS, Rac1 probably reacts with a variety of oxidants in the cell. Here, we show that cellular Rac1 is S-glutathiolated in chondrocytes. This oxidative modification leads to enhanced nucleotide exchange in vitro, which is a key step to promote Rac1 activation in cells. Consistent with these observations, we find that a Rac1 oxidative mimetic shows enhanced GTP-dependent effector binding and downstream sig-naling to the cytoskeleton.

The 305-Da S-glutathione modification is a common thiol oxidation product found under both basal conditions and oxidative stress [56,78] and has been observed in many proteins with redox-sensitive cysteine residues [69,79,80]. For example, the Ras GTPase undergoes S-glutathiolation at Cys¹¹⁸ upon exposure to peroxynitrite and DEA NONOate (an NO^• -donating agent) in bovine aortic endothelial cells [80,81]. We now show that Rac1 is S-glutathiolated in primary chondrocytes upon menadione-induced ROS formation. Osteoarthritis is a degenerative joint disease that is characterized by high levels of oxidative stress in chondrocytes [34], and Rac1 activity is enhanced in osteoarthritic chondrocytes compared to normal cells [35]. Moreover, inhibition of Rac1 blocked production of the degra-dative matrix metalloprotease MMP-13 in chondrocytes [35] and reduced the severity of arthritis in a mouse model [14]. As we show here that S-glutathiolation of in vitro-purified Rac1 Cys¹⁸ greatly accelerates nucleotide dissociation, a key mechanism of activating Rac1, we postulate that Rac1 oxidation may promote its activation in degenerative joint disease and may play a role in driving arthritis progression.

In addition to S-glutathiolation, oxidation of cellular thiols by ROS can also promote the formation of sulfenic, sulfinic, and sulfonic acids. However, generation of a single species of oxidized thiol in vitro can prove difficult if not impossible. Previously, Permyakov et al. [57] used a Cys→Asp variant to mimic thiol oxidation in recoverin because aspartic acid shows similarity in shape and charge to sulfinic acid (Supplementary Fig. 4). As oxidation by ROS can yield several different oxidation products [32], use of the oxidation mimetic Rac1^C18D allows for the study of a singly oxidized species. This variant shows greatly enhanced guanine nucleotide exchange, similar to glutathiolated Rac1, supporting the use of this variant as an oxidation mimetic. We also generated redox-insensitive Rac1^C18S and Rac1^C18A variants. Interestingly, we observe a trend toward increased nucleotide dissociation for Rac1^C18S<Rac1^C18A<Rac1^C18D. Whereas the Rac1^C18D variant has a dramatically enhanced (200-fold) GDP dissociation rate, similar to that observed for Rac1 glutathiolation, the Rac1^C18A variant possesses an approximately 11-fold increased rate of nucleotide exchange relative to Rac1^WT. In contrast, the Rac1^C18S variant has an intrinsic dissociation rate that is similar to that of Rac1^WT. Therefore, the effect of mutation at Cys¹⁸ on guanine nucleotide binding is dependent on the mutation type, with the less conservative negatively charged substitution (Rac1^C18D) producing the largest perturbation in guanine nucleotide binding. Given these observations, we postulate that different oxidation states alter Rac1 activity in distinct ways.

Rac1 is well known as an inducer of cytoskeletal reorganization, such as membrane ruffling and lamellipodia formation [64,82]. Consistent with our in vitro data demonstrating increased nucleotide exchange, we observed that Rac1^C18D is hyperactivated in HEK-293T cells and promotes lamellipodia formation in Swiss 3T3 cells, whereas Rac1^C18S showed levels of cellular activation similar to those of Rac1^WT. To the extent that the Rac1^C18D variant mimics oxidized Rac1 (e.g., Rac1^S–SG), these results suggest that oxidation of Rac1 at Cys¹⁸ can promote Rac1 activation through increased nucleotide exchange. Furthermore, our results indicate that Rac1^C18D retains the ability to bind to at least a subset of effector proteins that promote Rac1-mediated cytoskeletal changes. Thus, unlike other dynamic posttranslational modifications of Rac1, such as phosphorylation [83,84] and palmitoylation of the C-terminal membrane-targeting domain [6], which decrease Rac1 activity and/or alter Rac1 effector binding by altering its localization [85], oxidation of the N-terminal Cys¹⁸ primarily enhances the population of activated Rac1.

Rac1 is also well known as an inducer of ROS and RNS [86]. However, in contrast to the enhancing effects of oxidation on Rac1-mediated cytoskeletal organization, cysteine oxidation of Rac1 has been reported to decrease its association with SOD1 [39]. Therefore, it will be interesting to determine whether its interactions with other redox enzymes, such as NOX [36,37] and NOS [40], are also negatively affected by Rac1 cysteine oxidation. If so, this could indicate the presence of a negative feedback loop that would further add to the complexity of Rac1-mediated redox signaling. Conversely, a positive feedback loop could exist in which oxidation of Rac1 may facilitate interactions with other effectors. Future studies will investigate whether Rac1 oxidation alters effector interactions.

Finally, it is intriguing to note that Rac1 and Ras are differentially regulated by thiol oxidants. We have previously reported that thiyl radical formation at Ras Cys¹¹⁸ can promote guanine nucleotide exchange and Ras activation [72,73], whereas non-radical-mediated nitrosation and glutathiolation does not alter Ras nucleotide binding [75,76]. In contrast, we show herein that Rac1 can be activated by nonradical oxidation at Cys¹⁸. We propose that the distinct locations of the redox-sensitive cysteines within the nucleotide binding motifs of Rac1 and Ras explain this differential regulation. We therefore performed sequence and structural analyses of the redox-sensitive motifs in the context of the redox-sensitive thiols. Fig. 10A and 10B depict the redox-sensitive motifs contained within the protein structures ([77] and pdb 3GFT) for Rac1 and Ras, respectively. The redox-sensitive thiol in Rac1, Cys¹⁸, is within the GxxxxGK [S/T] motif, otherwise known as the p-loop, which places the thiol within hydrogen-bonding distance from the α-phosphate. However, in Ras, the redox-sensitive thiol lies within the [N/T] KxD motif and does not have direct interactions with other residues in Ras or with the bound nucleotide. Further, the thiol is located 6.6 Å away from the bound nucleotide and is solvent exposed. This difference in location can explain their distinct modes of regulation. For both Rac1 and Ras, thiyl radical formation at their respective cysteines can lead to oxidation of the bound nucleotide [28,72,73], which promotes nucleotide dissociation and can result in activation. We find by several complementary methods that thiol oxidation by nonradical oxidative mechanisms perturbs nucleotide binding only in Rac1 and not Ras ([75,76] and this study). Further, the altered pK_a of Rac1 Cys¹⁸ (this study) relative to Ras [61] renders Rac1 susceptible to thiol oxidation at physiological pH. Hence, Rac1 Cys¹⁸ is susceptible to a wider array of oxidative modifications, including both radical and nonradical oxidation. This differential oxidant-mediated regulation is depicted in Fig. 10C and 10D.

Fig. 10.

Differential modes of Rac1 and Ras activity regulation by thiol-mediated oxidants. (A) A linear depiction of Rac1 and residues critical for nucleotide binding. The top bar shows the consensus sequence for the nucleotide binding motifs with the GTPase-specific sequence highlighted within the box. Below, the structural depiction illustrates where the GxxxxGK[S/T] and [N/T]KxD motifs are located within the Rac1 structure (pdb 1MH1) and their proximity to the nucleotide. Cysteine 18 is approximately 3.5 Å from the α-phosphate (labeled PA) in Rac1 when bound to GDP or GTP. The coloring is as follows: blue, nitrogen; orange, phosphate; yellow, cysteine; red, oxygen; teal, carbon associated with Rac1; and the bound nucleotide (GMPPNP) carbons are colored magenta. (B) A linear depiction of Ras (pdb 3GFT) as in (A) with the structural depiction of the GxxxxGK[S/T] and [N/T]KxD motifs shown below. The redox-sensitive Cys¹¹⁸ is approximately 6.6 Å from the guanine ring of the bound GMPPNP nucleotide and does not make direct interactions with either the nucleotide or residues within Ras. In addition, the orientation of the Cys¹¹⁸ thiol is directed away from the nucleotide-binding pocket. (C) The redox-cycle diagram for Rac1 illustrates the effects of reactive oxygen and nitrogen species on Rac1 nucleotide binding. Rac1 is sensitive to thiol oxidation by radical oxidants through reaction with Cys¹⁸, resulting in enhanced nucleotide dissociation. In addition, reaction of Rac1 with nonradical oxidants impairs nucleotide binding by covalent modification of Cys¹⁸. Owing to the proximity of Cys¹⁸ to the bound nucleotide, oxidation at this site interferes with nucleotide binding, which is indicated by the oxidative modification triangle (Ox) that overlaps with the nucleotide, and results in increased nucleotide exchange. As oxidation does not affect GAP function, oxidized Rac1 can be inactivated by GAPs. Known nucleotide-dependent steps are labeled with diamonds with a nucleotide label and independent steps are shown as diamonds with no label. NO^•, nitric oxide; ${\text{NO}}_2^{•}$, nitrogen dioxide; ${\mathrm{O}}_2^{•-}$, superoxide; GSH, glutathione; GSSG, oxidized glutathione; H₂O₂, peroxide; GAP, GTPase-activating protein; GEF, guanine nucleotide exchange factor. (D) A redox-cycle diagram for Ras. The redox-sensitive thiol in Ras at Cys¹¹⁸ does not directly interact with either Ras residues or the bound nucleotide; consequently, the major difference between Ras and Rac1 is that covalent modification of Ras Cys¹¹⁸ does not alter nucleotide binding. As such, the oxidative modification does not overlap with the bound nucleotide, as depicted. Labeling is the same as in (C).

5. Conclusions

In summary, our in vitro and cell-based experiments suggest that oxidative modification of Rac1 can increase Rac1 activity and downstream signaling by enhancing nucleotide exchange. In addition, our findings indicate that Rac1^C18D may prove useful as an oxidation mimetic, whereas Rac1^C18S is a redox-insensitive variant that displays biochemical properties, structure, and interactions with Rac1 regulators similar to Rac1^WT under nonstressed conditions. Thus, the Rac1^C18S variant should serve as a valuable tool to determine whether cellular oxidants act directly or indirectly on Rac1 and will aid in elucidating the role of Rac1 oxidation in normal and disease states. Our data support a potential new layer of regulation for Rac1 activation through oxidative modification. Redox regulation of Rac1 may not only directly modulate Rac1 activity but also provide a feedback mechanism for regulating major ROS- and RNS-modulating enzymes, including NOS, NOX, and SOD1, that are, in turn, regulated by Rac1. Our observations are particularly interesting given that Rac1 plays a critical role in cellular redox regulation and that both Rac1 and ROS/RNS are involved in multiple common pathophysiological conditions. For example, our finding that Rac1 is S-glutathiolated in primary chondrocytes suggests a possible mechanism for regulating Rac1 activity levels in osteoarthritis. It will be interesting to determine the role of Rac1 oxidation in this and other diseases, such as cancer, in which high levels of oxidative stress are present.

Abbreviations

ABD-f

4-(aminosulfonyl)-7-fluoro-2,1,3-benzoxadiazole

βME

β-mercaptoethanol

CD

circular dichroism

DMEM

Dulbecco’s modified Eagle’s medium

DTPA

diethylenetriaminepentaacetic acid

DTT

dithiothreitol

EDTA

ethylenediaminetetraacetic acid

eNOS

endothelial nitric oxide synthase

ETD–MS/MS

electron transfer dissociation–tandem mass spectrometry

FRET

fluorescence resonance energy transfer

GAP

GTPase-activating protein

GDI

guanine nucleotide dissociation inhibitor

GEF

guanine nucleotide exchange factor

GSSG

oxidized glutathione

HSQC

heteronuclear single-quantum coherence spectroscopy

IκBα

nuclear factor of κ light polypeptide gene enhancer in B cells inhibitor, α

IKK

IκB kinase

IGF-1

insulin-like growth factor-1

MAPK

mitogen-activated kinase

mant-GDP

2′-/3′-O-(N′-(methylanthraniloyl)guanosine-5′-O-diphosphate

MMP

matrix metalloproteinase

NF-κB

nuclear factor-κB

nNOS

neuronal nitric oxide synthase

NOX

NADPH oxidase

PBD

p21-binding domain

PBS

phosphate-buffered saline

p-loop

phosphoryl-binding loop

RNS

reactive nitrogen species

ROS

reactive oxygen species

SOD1

superoxide dismutase 1

Appendix A. Supporting information

Supplementary data associated with this article can be found in the online version at http://dx.doi.org/10.1016/j.freeradbiomed.2014.09.027.