Mechanism of Redox-mediated Guanine Nucleotide Exchange on Redox-active Rho GTPases*

Rho GTPases regulate multiple cellular processes including actin cytoskeletal rearrangements, transcrip-tional regulation, and oxidant production. The studies described herein demonstrate that small molecule redox agents, in addition to protein regulatory factors, can regulate the activity of redox-active Rho GTPases. A novel (G XXXX GK(S/T)C) motif, conserved in a number of Rho GTPases, appears critical for redox-mediated guanine nucleotide dissociation in vitro . A detailed molecular mechanism for redox regulation of G XXXX GK(S/ T)C motif-containing Rho GTPases is proposed. Rho GTPases branch of the Ras superfamily small guanine nucleotide-binding Cdc42 assays and sample preparations were performed in sealed cuvettes and vials to avoid diffusion of volatile redox agents into the open atmo-sphere (10). To prevent transition metal-mediated conversion of the O 2 . radical into H 2 O 2 and OH (cid:1) (26) as well as NO into the nitrosonium ion, all buffers used for kinetic and biochemical assays were passed of Guanine Nucleotide Exchange on Redox-active Rho GTPases—

NOS as a NO-producing enzyme, several studies have implicated XO in the production of NO (33). Peroxynitrite is a precursor of both ROS and RNS, as it is formed by the reaction of NO with O 2 . , which can then be decomposed into ⅐ NO 2 and OH ⅐ (30).

EXPERIMENTAL PROCEDURES
Preparation of Chemicals and Proteins-The chemicals used for all experiments were of the highest grade unless otherwise noted. Radiolabeled guanine nucleotides ([ 3 H]GDP and [ 3 H]GTP) were diluted with unlabeled guanine nucleotides prior to use, giving ϳ1000 dpm/M nucleotide. Rac1-(1-177), Rac1 F28L, Rac1 C18S, RhoA-(1-181), and Cdc42-(1-188) proteins were expressed and purified as described previously (34). The final proteins were Ͼ95% pure as determined by SDS-PAGE. The protein concentration was determined by the Bradford method (35). XO was purchased from Sigma.
Experimental Conditions and Preparation of Redox Agents-All assays and sample preparations were performed in sealed cuvettes and vials to avoid diffusion of volatile redox agents into the open atmosphere (10). To prevent transition metal-mediated conversion of the O 2 . radical into H 2 O 2 and OH ⅐ (26) as well as NO into the nitrosonium ion, all buffers used for kinetic and biochemical assays were passed * This work was supported by National Institutes of Health Grants RO1CA89614-O1A1 and PO1 HL45100 (to S. L. C.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
¶ To whom correspondence should be addressed: Dept. of Biochemistry and Biophysics, University of North Carolina, 530 Mary Ellen Jones Bldg., Chapel Hill, NC 27599-7260. Tel.: 919-966-7139; Fax: 919-966-2852; E-mail: campbesl@med.unc.edu. 1 The abbreviations used are: GEF, guanine nucleotide exchange factor; ROS, reactive oxygen species; RNS, reactive nitrogen species; XO, xanthine oxidase; NOS, nitric-oxide synthase; MS, mass spectrometry; BSA, bovine serum albumin; WT, wild type; 8-DIm-DP, 8-diamino-4O-imidazolone ribose diphosphate; 5-DIm-DP, 5-diamino-4O-imidazolone ribose diphosphate; AIm-DP, 5-amino-4O-imidazolone ribose diphosphate; IIm-DP, 5-imino-4O-imidazolone ribose diphosphate; DMm-DP, 5-diamino-4O-2-methyl-imidazolone ribose diphosphate; NHE, normal hydrogen electrode; DTPA, diethylenetriaminepentaacetic acid. through a metal-chelating Bio-Rad Chelex 100 cation exchange column (10), and diethylenetriaminepentaacetic acid (DTPA, 0.1 mM) was added to the buffers prior to performing the experiments, unless otherwise noted. Detailed methods for generation and quantification of O 2 . using XO and a methylene blue-coupled assay system under aerobic conditions have been described previously (12). NO gas was purified by passing it through a scrubbing column with 5 M KOH, and quantification of NO in the assay solution was achieved by using a hemoglobincoupled assay under anaerobic conditions (10). Authentic ⅐ NO 2 was used for ⅐ NO 2 -mediated Rho GTPase guanine nucleotide dissociation assays, and quantification of ⅐ NO 2 in the assay solution was achieved by using the NO 2 /NO 3 assay kit C II (Dojindo) under anaerobic conditions. N 2 O 3 was prepared by mixing NO gas and authentic ⅐ NO 2 gas (1:1, mol:mol) under anaerobic conditions. Although the content of N 2 O 3 in the assay solution was not directly determined, the quantity of N 2 O 3 in the assay solution was estimated to be similar to the NO content (i.e. 5 M NO gas, which is mixed with 5 M ⅐ NO 2 gas) in the assay, based on the assumption that NO reacts with ⅐ NO 2 stoichiometrically 1: . To determine the effect of a radical-quenching agent on Rac1 GTP association, ascorbate (1 mM) was added to the assay mixture at assay time 500 s. As a control, the intrinsic rate of Rac1 GTP association was measured using experimental conditions identical to those described above, except that XO was not added. For all assays including both Rho GTPase GDP dissociation and GTP association, aliquots were withdrawn at specific time points and spotted onto nitrocellulose filters. The filters were then washed three times with assay buffer, and radioactivity was determined using a Beckman-Coulter scintillation counter. The rates of redox agent-mediated Rho GTPase GDP dissociation and GTP association were obtained by fitting the data to a simple exponential decay and exponential association, respectively. The regression values associated with the fit were r 2 Ͼ 0.7595. Mass Spectrometry Analysis-Mass spectrometry (MS) sample preparations and analyses of the nucleotide derivatives released from Rho GTPases were identical to those described previously (12), except the Rac1 and Rac1 variants, C18S and F28L (1 M, 500 l), were employed instead of Ras and Ras variants.

RESULTS AND DISCUSSION
Redox-active Rho Subfamily GTPases-We have identified a distinct class of redox-active GTPases conserved in nearly 50% of all Rho subfamily GTPases that contain a cysteine, Cys 18 (Rac1 numbering), located at the end of the P-loop (GXXXXGK(S/T), residues 10 to 17, Rac1 numbering). This cysteine-containing P-loop motif will be referred to herein as the GXXXXGK(S/T)C motif. For comparison, select members of the Ras subfamily of GTPases such as H-, K-, and N-Ras and Rap1A possess an NKCD motif that forms interactions with the guanine nucleotide base and contains a redox-active cysteine, Cys 118 (Ras numbering) (5, 10 -12). Inspection of the available NMR and x-ray crystal structures for the GXXXXGK(S/T)C motif-containing Rho GTPases Rac1, RhoA, and Cdc42 (36 -38) indicates that the Cys 18 thiol in the GXXXXGK(S/T)C motif is solvent-accessible, suggesting that ROS and RNS (e.g. O 2 . and ⅐ NO 2 , respectively) may be able to target the Cys 18 thiol. Consistent with this premise, results shown in Fig. 1A  The redox-active Cys 118 thiol in the NKCD motif of H-Ras faces the Phe 28 side chain and is ϳ12 Å away from the center of the Phe 28 side chain (11,12). Coupling between Cys 118 and Phe 28 appears critical for redox-mediated guanine nucleotide dissociation (11,12). Intriguingly, the Cys 18 thiol in the GXXXXGK(S/T)C motif of Rac1 also faces the Phe 28 side chain (Rac1 and Ras numbering are the same) and is only ϳ3.6 Å away from the center of the Phe 28 side chain (36 -38). We have previously proposed a mechanism whereby reaction of either NO/O 2 (via ⅐ NO 2 ) or O 2 . with Ras generates a Cys 118 -thiyl radical (Cys 118 ⅐ ), which in turn promotes formation of a guanine radical intermediate resulting in perturbation of Ras guanine nucleotide interactions and accelerated release of GDP as GDP derivatives from Ras (11,12). We further proposed that Phe 28 may act as a conduit between the Ras Cys 118 ⅐ and the guanine nucleotide base to facilitate electron transfer (11,12). If the mechanism of redox-mediated guanine nucleotide dissociation of GXXXXGK(S/T)C motif-containing Rho GTPases (see below) is similar to that of Ras GTPases (11,12), electronic coupling between the Rac1 Cys 18 -thiyl radical (Cys 18 ⅐ ) and the Phe 28 side chain is expected, with the Phe 28 side chain serving as an electron conduit between the Rac1 Cys 18 ⅐ and the bound GDP guanine base.
To evaluate whether Phe 28 is involved in redox-mediated regulation of Rho GTPase activity, we determined the rate of O 2 . -mediated guanine nucleotide dissociation associated with a Rac1 F28L variant. The Rac1 F28L variant shows a faster intrinsic rate of guanine nucleotide dissociation (Fig. 1A) relative to that of WT Rac1, similar to that observed previously for the Ras F28L mutant (40). However, the rate of guanine nucleotide dissociation was not further enhanced in the presence of O 2 . . The Cdc42 F28L variant has been characterized previously by NMR, and results from this study indicate that the structure of Cdc42 is not significantly altered by substitution of leucine for phenylalanine (41). Rather, loss of the packing interaction between Phe 28 and the guanine nucleotide base alters dynamics of the Cdc42 protein guanine nucleotide-bind-ing site, resulting in reduced guanine nucleotide-binding affinity and enhanced guanine nucleotide dissociation. Given the conservation of this residue and the structural homology between Cdc42 and Rac1 (36,38), it is anticipated that the Rac1  Table I. expect that the shorter distance between the Phe 28 side chain and the Cys 18 thiol contained within the GXXXXGK(S/T)C motif will result in more efficient redox coupling and consequently enhanced redox-mediated guanine nucleotide dissociation from GXXXXGK(S/T)C motif-containing GTPases relative to NKCD motif-containing Ras GTPases (10 -12). In support of this premise, the rates of O 2 . -enhanced GDP dissociation from both Rac1 and Cdc42 (ϳ6 ϫ 10 Ϫ3 s Ϫ1 ) (Table I) are at least 2-fold faster compared with that observed for H-Ras (ϳ3 ϫ 10 Ϫ3 s Ϫ1 ) (12), yet it is not clear whether the faster rate of redox-mediated guanine nucleotide dissociation of Rac1 and Cdc42 compared with H-Ras observed in vitro has implications for their differential regulation by redox agents in vivo. Unlike Rac1 and Cdc42 (36,38), RhoA possesses another cysteine (Cys 18 RhoA numbering; Lys 16 Rac1 numbering) in addition to the redox-active cysteine in the GXXXXGK(S/T)C motif (37). Although this additional cysteine faces the redox-active Cys 18 with a distance of ϳ10 Å, it does not face either the Phe 28 side chain or the bound GDP guanine base (37). Hence, it is likely that this additional cysteine interacts electronically with the redox-active Cys 18 (10), can stimulate dissociation of GDP from Rac1 (Fig. 1C), RhoA, and Cdc42 (Table I). However, Rac1 guanine nucleotide dissociation is not enhanced in the presence of NO alone (in the absence of O 2 ) or a stoichiometric mixture of NO and authentic ⅐ NO 2 (to produce N 2 O 3 ) (Fig. 1C). These results indicate that the activity of GXXXXGK(S/T)C motifcontaining Rho GTPases is sensitive to ⅐ NO 2 , which may facilitate guanine nucleotide dissociation from Rac1 (Fig. 1C), RhoA, and Cdc42 (  (13)(14)(15)(16)(17). Rac1 has been shown to regulate the assembly and activation of NADPH oxidase in phagocytes and is implicated in regulation of NADPH oxidase-related enzymes in non-phagocytic cells to generate bactericidal O 2 . from NADPH oxidase as well as modify gene expression, cell proliferation, adhesion, and many cell-specific functions, respectively (2,27). It has been shown also that the Rac1 GEF kalirin, as well as Rac1 and Rac2, associates with inducible NOS (42). Moreover, Rac1 colocalizes with inducible NOS in activated macrophages (42). Hence, Rac GTPases may be localized and exposed to levels of O 2 . and ⅐ NO 2 that regulate its activity in situ.  Fig. 1 show that various redox agents facilitate guanine nucleotide dissociation of GXXXXGK(S/T)C motif-containing Rho GTPases. Cellular activation of Rho GTPases requires exchange of bound GDP for GTP. However, to produce the active GTP-bound state of Rho GTPases, GTP association should occur following redox-mediated GDP dissociation from the Rho GTPase to complete guanine nucleotide exchange.

Redox Activation of GXXXXGK(S/T)C Motif-containing Rho GTPases-Results in
As shown in Fig. 2  are present at high levels relative to radical-scavenging agents. This, in turn, may cause down-regulation of NADPH oxidase activity, thereby providing a mechanism for negative regulation of Rac-mediated NADPH oxidase activity. Further studies are required to delineate GTPase signaling mediated by redox agents with respect to the action of radical scavengers in cells.  Fig. 3, we postulate that the mechanism of ROS-and RNS-mediated guanine nucleotide dissociation of GXXXXGK(S/T)C motif-containing GTPases is similar to that of NKCD motif-containing Ras GTPases (11,12), as our kinetic results, the redox-coupling partners (GXXXXGK(S/T)C, Phe 28 , and bound GDP), and their structural architecture show a similarity to NKCD motif-containing Ras GTPases (43).

Potential Mechanism for Redox Activation of GXXXXGK(S/ T)C Motif-containing Rho GTPases-As shown in
Given that Rac GTPases co-localize with O 2 . -producing NADPH oxidases and regulate their activity (27), Rac1 activity may be regulated by O 2 . , possibly via a feedback mechanism.
Although this mechanism is shown for O 2 . , our data indicate ( Fig. 1 and Table I) that other redox agents, such as ⅐ NO 2 , can modulate Rac1 activity and possibly other GXXXXGK(S/T)C motif-containing Rho GTPases. Consistent with our data, Rac GTPases have been shown to co-localize and interact with inducible NOS and may facilitate peroxynitrite production (42). Given the data presented here and the mechanism proposed for redox regulation of H-Ras, we postulate that a Rac1 Cys 18 ⅐ intermediate, formed by the reaction of the Rac1 Cys 18 thiol and O 2 . (Fig. 3, step i), withdraws an electron from the Rac1bound GDP guanine base via the Rac1 Phe 28 side chain to produce a Rac1-GDP cation radical (G . ϩ -DP) (step ii). G . ϩ -DP is then converted to G ⅐ -DP by elimination of a H ϩ from the N-1 of G .
ϩ -DP, resulting in disruption of key Rac1 guanine nucleotide hydrogen-bond interactions, including the guanine N-1 interaction with the Asp 118 side chain (Rac1 numbering) resulting in release of GDP and formation of a GDP-deficient form of Rac1 (apoRac1) (Fig. 3, step iii) (Fig. 3, step iv), and formation of GDP-O 2 H is likely to interfere with additional Rac1-GDP interactions thereby triggering dissociation of the bound GDP-O 2 H from Rac1. GDP-O 2 H can be then decomposed to an unstable 5-diamino-4O-imidazolone ribose diphosphate (5-DIm-DP) (Fig. 3,  step v). If reaction of G ⅐ -DP with O 2 . occurs prior to release of G ⅐ -DP from Rac1, 5-DIm-DP will be the dominant product formed over 8-diamino-4O-imidazolone ribose diphosphate (8-DIm-DP), as the C-5 site of Rac1-bound G ⅐ -DP is exposed (hence O 2 . is accessible to the C-5 site of Rac1-bound G ⅐ -DP), whereas the C-8 site of Rac1-bound G ⅐ -DP is not solvent-accessible (36). However, if reaction of G ⅐ -DP with ⅐ NO 2 occurs prior to the release of G ⅐ -DP from Rac1, the 5-DIm-DP is likely to predominate, and this adduct is then released from Rac1 (Fig.  3, step v). Depending on the experimental conditions, the Rac1released 5-oxo-GDP is further degraded into oxygenated nucleotide derivatives. If 5-oxo-GDP is treated with acid (e.g. formic acid), 5-imino-4O-imidazolone ribose diphosphate (IIm-DP) and (Fig. 3, step vi 1 ) 5-amino-4O-imidazolone ribose diphosphate (AIm-DP) (Fig. 3, step vi 2 ) can be formed (12). If 5-oxo-GDP is treated with methanol prior to acid, 5-diamino-4O-2methyl-imidazolone ribose diphosphate (DMm-DP) (Fig. 3, step  vi 3 ) can be produced. The release of GDP-oxygenated adduct(s) from Rac1 produces a GDP-deficient form of Rac1 (apoRac1). ApoRac1 can be converted to an active guanine nucleotide binding state by the addition of a radical scavenger (i.e. ascorbate) in the presence of GDP or GTP. Given the cellular abundance of GTP, by mass action, the active guanine nucleotide binding form of Rac1 is likely to become GTP-bound leading to Rac1 activation in situ.
According to the proposed mechanism (Fig. 3), the Cys 18 and the Phe 28 side chain play a key role in radical-based perturbation of Rac1 GDP-binding interactions and generation of oxygenated GDP adduct(s) released from Rac1. Thus, mutation of these key residues should hinder O 2 . -mediated Rac1 guanine nucleotide dissociation. Consistent with the proposed mechanism as shown in Fig. 1A (Fig. 4A). Given that the predicted molecular masses of AIm-DP, IIm-DP, and DMm-DP are 415.26, 417.26, and 465.26 Da, respectively, we have putatively assigned the three Rac1-released GDP-O 2 H derivatives as AIm-DP, IIm-DP, and DMm-DP, respectively. We were unable to detect MS peaks corresponding to the molecular weight of 8-DIm-GDP derivatives. The inability to detect these derivatives is consistent with our premise that the C 8 site of G . ϩ -DP is not exposed to solvent, so that the reaction product of the G ⅐ -DP C-8 with O 2 . , 8-DIm-GDP, is unlikely to be produced and detected. The Rac1-derived oxygenated GDP derivatives were not detected when the Rac1 variants, C18S and F28L, were treated with O 2 . (Fig. 4, B and C), providing further support that the Rac1 residues Cys 18 and Phe 28 play a key role in the production of the 5-DIm-GDP adduct. Additionally, when NO/O 2 ( ⅐ NO 2 ) was used instead of O 2 . to promote Rac1 guanine nucleotide dissociation, a nitrated GDP derivative (putatively assigned as 5-guanidino-4-nitroimidazole diphosphate (11)) was detected (not shown). The generation of this adduct is consistent with the proposed mechanism in Fig. 3, where reaction of the redox agent ⅐ NO 2 with the Rac1 Cys 18 thiol can also produce Cys 18 ⅐ , and the resultant GDP radical then reacts with ⅐ NO 2 to produce the nitrated GDP adduct (11).
The key feature of the proposed mechanism (Fig. 3) is that the redox agents react with the Cys 18 thiol in the GXXXXGK(S/ T)C motif to produce Cys 18  . (or RNS, ⅐ NO 2 ), promotes guanine nucleotide dissociation (Fig. 1) giving rise to apoRac1 and a modified GDP adduct (Fig. 4)  Rho GTPase Redox Regulation and Biological Relevance-Based on the results described here, we postulate that there exist two distinct mechanisms by which Rho GTPase activity is modulated by stimulation of Rho GTPase guanine nucleotide dissociation, small molecule and protein modulatory agents, i.e. redox agents and GEFs. Activation by one or both of these mechanisms is likely to depend on cellular conditions such as co-localization with a GEF or redox agent, GEF activation state, level of redox agents, and redox environment. Although a defined role for redox agents in direct cellular regulation of Rho GTPases has not yet been established, recent studies provide evidence for in vivo redox regulation of Rho GTPase activity. For example, it has been shown that the activation state of the Rho GTPases, Rac1 and RhoA, responds rapidly to changes in oxygen tension in a phosphatidylinositol 3-kinase and NADPH oxidase-dependent manner, and their coordinated actions regulate endothelial barrier function in endothelial cells of conduit pulmonary arteries (22). It has also been shown that Caki-1 cells exposed to hypoxia (1% O 2 , low oxygen tension) exhibit increased Cdc42, Rac1, and RhoA protein expression and activity (23). Overexpression and activation of Rho proteins is downstream of and dependent on the production of ROS, because specific inhibition of ROS-producing NADPH oxidase down-regulates GTPase activity (23). In contrast, previous studies indicate that blocking NADPH oxidase did not inhibit Rac1 activity under certain hypoxia-induced conditions in endothelial cells of conduit pulmonary arteries (22). Moreover, human neutrophil adherence to extracellular matrix proteins has been shown to induce an initial inhibition of stimulated ROS formation, and this is believed to result from inhibition of GEF-mediated Rac2 activation (47). Understanding the basic mechanism of GEF and redox-mediated regulation of Rho GTPase activity in vitro should help to delineate the complexities associated with in vivo regulation.
Given our results, we would predict that high levels of redox agent(s) and oxidative conditions may result in Rho GTPase thiol modification and redox inactivation of Rho GTPases. Therefore, thiol-modified forms of Rho GTPases may be insensitive to ROS production by NADPH oxidase. However, high levels of redox agents and oxidative conditions may be more representative of oxidative stress and pathophysiological rather than physiological conditions or conditions that give rise to oxidative burst, for instance, in response to microbial invasion (48).
Other GXXXXGK(S/T)C Motif-containing GTPases-In addition to the redox-active GXXXXGK(S/T)C motif-containing Rho GTPases, we have found that a number of Rab GTPases also possess a GXXXXGK(S/T)C motif. Rab GTPases comprise the largest branch of the Ras GTPase superfamily and are best known for their role in vesicular trafficking (49). Hence, we speculate that (i) redox agents, e.g. O 2 . , ⅐ NO 2 , and H 2 O 2 /Cu 2ϩ (OH ⅐ ), may modulate temporal and spatial cellular functions of the GXXXXGK(S/T)C motif-containing Rab GTPase subfamily GTPases; and (ii) the fundamental molecular mechanism of the redox-active Rab GTPases may be similar to that proposed for redox-active Rho GTPases.
In summary, we have identified and characterized a unique redox-active motif present in Rho family GTPases that is important for redox-mediated regulation of guanine nucleotide exchange activity in vitro. Moreover, the radical-based molecular mechanism of Rho GTPase guanine nucleotide exchange appears similar in nature to the mechanism characterized for Ras GTPases. The redox-active motif is present in other Ras superfamily GTPases (i.e. Rab GTPases), suggesting that redox regulation of GTPase signaling is more widespread than previously envisioned. The studies described here form the basis for the investigation of cellular regulation of Ras superfamily GTPases by redox agents.