Redox Control of GTPases: From Molecular Mechanisms to Functional Significance in Health and Disease

1. Protein regulator-mediated activity modulation of Ras superfamily GTPases

These small GTPases function by cycling between active guanine nucleoside diphosphate (GDP)-bound and inactive GTP-bound states (Fig. 2) (74). As illustrated in Figure 2, guanine nucleotide-exchange factors increase the slow intrinsic rate of guanine-nucleotide exchange (GNE) to populate the proteins in their biologically active GTP-bound states (74, 135, 258, 259). The activated GTPases, in turn, interact with a variety of downstream effector proteins that modulate numerous cellular signaling processes, including cellular proliferation, actin cytoskeletal organization, and gene expression (74, 97). The signal is downregulated through the action of GTPase-activating proteins that increase the low intrinsic rate of GTP hydrolysis for most Ras family GTPases and thereby terminate downstream signaling events (Fig. 2) (74, 201, 259). The combined action of these regulators determines the relative proportion of active and inactive GTPases present in the cell. Protein regulators for Ras, Rab, Rho, and Ran proteins are known, but there are no known protein regulator(s) specific to dexamethasone-induced Ras protein1 (Dexras1) or Ras homolog enriched in striatum (Rhes).

OPEN IN VIEWER

FIG. 2.

Regulation cycle of small GTPase activity. GAPs, GTPase-activating proteins; GEFs, guanine nucleotide exchange factors; RNS, reactive nitrogen species; ROS, reactive oxygen species. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article at www.liebertonline.com/ars).

2. Redox-mediated activity modulation of Ras superfamily GTPases

Many of these small GTPases are redox sensitive, and their known conserved redox-sensitive sequences have been termed the NKCD, GXXXXGK(S/T)C, and CGNKXD motifs (vide infra). The action of redox agents on these redox-sensitive GTPases is similar to that of guanine nucleotide-exchange factors in that they perturb GTPase nucleotide-binding interactions that result in the enhancement of the GNE of small GTPases.

1. Reactive nitrogen species

The heme-containing enzyme nitric oxide synthase (NOS) catalyzes the oxidation of L-arginine to NO and L-citrulline via the intermediate N-hydroxyl-L-arginine (5, 70, 199). Several isoforms of NOS, such as neuronal NOS (nNOS or NOS1), inducible NOS (iNOS or NOS2), and endothelial NOS (eNOS or NOS3), are known (189, 205). Despite the overwhelming current interest in NOS proteins as NO-producing enzymes, several studies implicate xanthine oxidase in the production of NO (75, 191, 192, 315, 316).

a. Properties of RNS

NO has an E _m ⁷ value of ∼−1.7 V and ∼−0.8 V (vs. NHE), respectively, versus the one-electron reduced singlet excited and triplet ground state of the nitroxyl anion (NO⁻) (14). Other biologically significant RNS include a radical nitrogen dioxide (^•NO₂; E _m ⁷ = ∼1.0 V vs. NHE) and nonradical higher oxides (e.g., dinitrogen trioxide [N₂O₃]) (6, 267). NO is a precursor of these various RNS because it can react with O₂ to produce ^•NO₂; the reaction of ^•NO₂ with another NO results in production of higher oxides such as N₂O₃ (6). This N₂O₃ formation reaction is reversible because the formed N₂O₃ can be degraded into NO and ^•NO₂ (6). Given that cysteine is the key and common residue of all of the redox-sensitive motifs found in these small GTPases, the mechanisms of the reactivities and chemical reactions of RNS with thiol/thiolate (Fig. 3) are particularly important to comprehending the RNS-mediated regulation of the activity of these GTPases.

OPEN IN VIEWER

FIG. 3.

Reactions of thiol/thiolate with RNS. The number on the electron movement arrow indicates a potential order of reaction. Researchers have been unable to observe a direct reaction between an authentic NO and the RSH form of Cys-SH or GSH under in vitro anaerobic conditions (99, 295). However, a previous study has proposed that in the presence of an electron acceptor, NO reacts directly with Cys-SH or GSH to produce Cys-SNO and GSNO, respectively, via a radical intermediate (82). It also has been proposed that an O₂-independent oxidation of NO by coupling it with a reduction of a transition metal (i.e., Fe³⁺) produces a nitrosonium ion-metal adduct (A). This nitroso-metal adduct reacts with RS⁻, such Cys-S⁻ or GS⁻, respectively, to produce Cys-SNO or GSNO (A) (282). Studies have shown that N₂O₃ reacts directly with Cys-SH and GSH, respectively, to produce Cys-SNO and GSNO (138, 153, 295). It is possible that a lone pair of electrons on the sulfur atom of Cys-SH or GSH nucleophilic attack to N₂O₃ to produce Cys-SNO or GSNO, respectively, and a nitrite (NO₂ ⁻) (B). If so, because RS⁻ is a better nucleophilic agent than RSH, the probable S-nitrosation species with N₂O₃ is likely to be Cys-S⁻ and GS⁻ (C). A radical reaction path between ^•NO₂ and RS⁻ to produce NO₂ ⁻ and an RS^• also has been discerned. Studies show that ^•NO₂ reacts with one of the RS⁻ molecules, such as Cys-S⁻ or GS⁻, to produce Cys-S^• or GS^• (D) (69, 130). However, ^•NO₂ does not react with RSH, nor with Cys-SH and GSH, because the standard midpoint redox potential at pH 7 of RS^•/RSH (e.g., Cys-S^•/Cys-SH or GS^•/GSH) is higher than that of ^•NO₂/NO₂ ⁻ (6, 267). The RS^•, Cys-S^•, or GS^• thus formed can further react with NO to produce Cys-SNO or GSNO, respectively (D). Cys-SNO (or GSNO) can be homolytically cleaved, respectively, into Cys-S^•, and NO (or GS^• NO) (E). RS^• may react with another RS^• to produce an RSSR such as a cysteine disulfide, a glutathione disulfide, or a cysteine-glutathione disulfide. R₁S^• can oxidize R₂S⁻ to produce R₁S⁻ and R₂S^•. For example, Cys-S^• reacts with GS⁻ to produce Cys-S⁻ and GS^•, or vice versa. When NO is available, RS^• (e.g., GS^• or Cys-S^•) reacts with NO to produce RSNO (e.g., GSNO or Cys-SNO, respectively). A direct transfer of an NO group from Cys-SNO or GSNO to another RSH assisted by an acid/base pair is possible (F) (107, 221, 260). This NO group transfer mediated by an acid/base pair may be relevant to S-nitrosation of a protein that has a pair of acid and base residues vicinal to a cysteine that is to be S-nitrosated. It also is possible that an RS⁻, such as Cys-S⁻, nucleophiles the NO moiety of GSNO and results in transfer of the NO group from GSNO to Cys-S⁻ to produce GS⁻ and Cys-SNO (G). An NO group transfer from Cys-SNO to GS⁻ to produce Cys-S⁻ and GSNO also is possible. Cys-S^•, cysteinyl radical; Cys-S⁻, cysteinate; Cys-SH, cysteine; Cys-SNO, S-nitrosocysteine; GS^•, glutathionyl radical; GS⁻, glutathiolate; GSH, glutathione; GSNO, S-nitrosoglutathione; NO, nitric oxide; NO₂, nitrogen dioxide; RS^•, thiyl radical; RS⁻, thiolate; N₂O₃, dinitrogen trioxide; RSH, thiol.

2. Reactive oxygen species

ROS include the O₂ ^•−, a hydroxyl radical (OH^•), and a hydrogen peroxide (H₂O₂) (6). All of these ROS are produced by various metabolic processes as well as by cellular enzymes, including NAD(P)H oxidases, xanthine oxidase, and cyclooxygenases (6, 21, 108, 116, 234).

The O₂ ^•−-generating multimeric enzyme complex NAD(P)H oxidase consists of a membrane-bound heterodimer of cytochrome b ₅₅₈ with a large glycoprotein gp91 ^phox and a small protein p22 ^phox in complex with Rap1A (8, 26, 38, 73, 142, 238). Homologs of the gp91 ^phox subunit have been termed Nox (Nox isomers, including Nox1 and Nox2) and their occurrences differ in organs (155). The other cytosolic components of the NAD(P)H oxidase include p67 ^phox , p47 ^phox , p40 ^phox , and Rac1 (1). Xanthine oxidase is primarily considered the terminal enzyme of purine (e.g., hypoxanthine and xanthine) catabolism in the cell, catalyzing the hydroxylation of hypoxanthine to xanthine and xanthine to urate; the processes are coupled with reduction of O₂ to O₂ ^•− (94, 108, 187). Cyclooxygenases (also known as prostaglandin-endoperoxide synthases), including isoforms of a constitutively active Cox1 as well as inducible forms of Cox2 and Cox3, generate O₂ ^•− (43, 68, 234). In addition, when the substrate L-arginine is not present, NOS produces O₂ ^•− (222, 223, 301, 302, 311).

OH^• has been shown to be formed in various subcellular organelles, including endoplasmic reticulum, mitochondria, nucleus, lysosomes, and peroxisomes (81, 140). H₂O₂ associated with Fenton chemistry is considered to be a main source of OH^• in these organelles (vide infra).

a. Properties of ROS

The E _m ⁷ of O₂ ^•− (∼0.9 V vs. NHE) is close to that of ^•NO₂. However, the E _m ⁷ of OH^• (∼2.3 V vs. NHE) is much higher than other RNS and ROS (6). Some ROS are interconvertible; for example, H₂O₂ can be converted into OH⁻ and into OH^• when coupled with oxidation of a transition metal ion (e.g., Fe²⁺ to Fe³⁺) (91). Although this reaction is commonly referred to as a type of Fenton reaction, it was first proposed by Harber and Weiss (287). H₂O₂ can also be decomposed into H⁺ and a perhydroxyl radical (^•OOH; a protonated form of O₂ ^•−) when coupled with reduction of a transition metal ion (e.g., Fe³⁺ to Fe²⁺), and this process is termed a Fenton reaction (287). Given that the pK _a of ^•OOH is ∼4.8 (20), ^•OOH can be deprotonated to O₂ ^•− at a physiological pH (i.e., pH 7.4). A reverse process, generation of H₂O₂ from O₂ ^•−, can be facilitated by superoxide dismutase (186) or the transition metal-catalyzed Harber-Weiss reaction (137). As with RNS, the reactivities and mechanisms of the reaction between ROS and thiol/thiolate (Fig. 4) must be comprehended to understand the redox-sensitive, motif-targeting, and ROS-mediated regulation of the activity of these GTPases.

OPEN IN VIEWER

FIG. 4.

Reactions of thiol/thiolate with ROS. Although H₂O₂ is a strong oxidant, its nonradical two-electron nucleophilic reaction is slow in targeting RSH (e.g., GSH or Cys-SH) to produce a form of RSOH (e.g., glutathionyl sulfenic acid; or cysteinyl sulfenic acid, Cys-SOH, respectively) (204). It is, however, possible to speculate that a nucleophilic attack of an RS⁻ on the oxygen atom of H₂O₂ yields RSOH and H₂O (A). As suggested for Ras Cys-SNO formation, if an acid/base pair of residues is present vicinal to the target Cys-SH of a protein, sulfenation of the protein Cys-SH with H₂O₂ is likely to be enhanced as shown in B. Given that the redox potential of OH^• is typically higher than in cellular components (267), OH^• can react with virtually all biomolecules. Ordinarily, this would exclude OH^• from a potential role of as a cellular signaling molecule. However, one study shows that the O₂-dependent formation of OH^• by a Fenton reaction was found to be relatively highly localized in perinuclear endoplasmic reticulum compared with other subcellular organelles such as mitochondria, Golgi apparatus, peroxisomes, and lysosomes (179). This OH^• in the perinuclear endoplasmic reticulum has been shown to play a role in expression and transactivation of hypoxia-regulated genes such as the hypoxia-inducible factor-1α and in activation of hypoxia-inducible factor-1α target gene expression (179). Hence, in certain circumstances, OH^• may be considered to be a cellular signaling molecule. Also, because OH^• is a strong one-electron oxidant, the potential OH^•-mediated formation of RSOH via the RS^• intermediate should be considered as shown in C (228). RSOH, formed by either a two-electron process (A or B) or a radical-based process (C), is thought to be an intermediate in the formation of the sulfur nitrogen bond (S-N). A lone pair of electrons of a nitrogen atom of the amine group attacks the sulfur atom of RSOH to form an S-N bond and H₂O (C). Although a reaction between two RSOH molecules is thought to generate thiosulfinate (228), RSOH can be oxidized by H₂O₂ to yield a form of a sulfinic acid that can be further oxidized by H₂O₂ to produce a sulfonic acid (204). Because the redox potential of O₂ ^•− is higher than that of RS⁻ (e.g., Cys-S⁻ or GS⁻) but lower than that of RSH (e.g., Cys-SH or GSH) (267), O₂ ^•− can oxidize RS⁻ but not RSH (D). Several studies have consistently shown that the reaction rate between O₂ ^•− and Cys-SH or GSH is relatively slow (17, 296, 297). The RS^• (e.g., Cys-S^• or GS^•) formed by the reaction between O₂ ^•− and RS⁻ can further react with RS⁻ (E, Path 1) or O₂ (E, Path 2), respectively, to produce an RSSR^•− (e.g., a cysteinyl disulfide radical anion) or an RSOO^• (e.g., a cysteinyl peroxyl radical) (305). Because RSSR^•− is a strong reductant, it predominantly reacts with O₂ to produce O₂ ^•− and RSSR (e.g., cysteine disulfide) (Path 1). This RSSR can then be cleaved proteolytically to yield two RS^• molecules. Before being isomerized, RSOO^• (a product of E, Path 2) also can be reduced to an RSOOH (e.g., a cysteinyl hydroperoxide) via hydrogen abstraction (e.g., from RSH) (E, Path 3). RSOOH was thought to be a precursor of the formation of the higher sulfur oxides, including RSOH (34, 52), via reaction with water to produce RSOH and H₂O₂ (E, Path 3) (305). RSOO^• can be isomerized to the sulfonyl radical (Path 4) (250) and then react with RSOOH (or RSH) to give sulfinic acid and RSOO^• (or RS^•) (E, Path 5) (250). Sulfonyl radical can react with O₂ to produce the sulfonyl peroxyl radical (e.g., a cysteine-sulfonyl peroxyl radical) (E, Path 6) (314). Cysteine-sulfonyl peroxyl radical reacts with H₂O to yield sulfonic acid (e.g., a cysteine-sulfonic acid, Cys-SO₃H) and perhydroxyl radical (314), which can be deprotonated to O₂ ^•− at pH 7.4 (E, Path 6). H₂O₂, hydrogen peroxide; O₂ ^•−, superoxide anion radical; OH^•, hydroxyl radical; RSOH, sulfenic acid; RSOO^•, thiyl peroxyl radical; RSOOH, thiyl hydroperoxide; RSSR, disulfide; RSSR^•−, disulfide radical anion.

3. Peroxynitrite

Although peroxynitrite (ONOO⁻) can be classified as an RNS, ONOO⁻ is classified in a separate section for convenience. The nonradical RNS, ONOO⁻, can be formed by the reaction of NO with O₂ ^•− (77). This is an example of the reactivity between RNS and ROS to produce another set of RNS and/or ROS. ONOO⁻ is known to be a powerful and destructive oxidant (152, 233), most likely because of its high reduction potential (294). ONOO⁻ can be decomposed into ^•NO₂ and OH^• (77). ONOO⁻ can also react with CO₂ to produce ^•NO₂ and carbonate radical (CO₃ ^•−) (E _m ⁷ = ∼1.8 V vs. NHE) (22).

The E _m ⁷ of CO₃ ^•− predicts that CO₃ ^•− can react with Cys-SH/GSH or Cys-S⁻/GS⁻ to produce Cys-S^• or GS^•. This fuels speculation that CO₃ ^•− has a role in cellular redox signaling via targeting cysteine-based redox-sensitive proteins. However, because the distribution of CO₂ is limited in tissues and organs, the role of CO₃ ^•− may be restricted or localized. Additional studies are required to examine what, if any, role CO₃ ^•− has in cellular redox signaling cascades.

4. Cellular redox conditions associated with oxygen content

Because O₂ often plays a role in the generation of and interconversion of biologically relevant RNS and ROS, the in vivo and in vitro studies of redox-mediated cellular signaling events also should take O₂ content into account.

Hypoxia is a pathophysiological condition in which the whole body or a part of it is deprived of adequate O₂. Hypoxia occurs because of a mismatch between the O₂ supply and the demand for it at the cellular level (291). Unlike under normal conditions—the nonhypoxic condition—a typical experimental hypoxic condition is a low O₂ concentration (typically ∼1–5% O₂) in solution. O₂ is still present in cells under a hypoxic condition, though in relatively low concentrations. However, although there is some debate about it and a possibility as well that what transpires depends on cellular compartments (e.g., endoplasmic reticulum, mitochondria, lysosomes, and peroxisomes) and cell types, hypoxic cells do not necessarily generate lesser ROS or RNS than cells at normal O₂ concentrations. Anoxia, an extreme form of hypoxia, also is a pathophysiological condition in which an organ's tissues are deprived of O₂. Therefore, the term “anoxic condition” is theoretically matched with the term “anaerobic condition,” which is defined as an absence of O₂ during the course of in vitro experiments (99). The term “aerobic condition” refers to in vitro experimental conditions in an open atmosphere, which corresponds to a “nonhypoxic condition” in pathophysiology. Nevertheless, anoxia and hypoxia are often used interchangeably—without regard to their specific meanings—in pathophysiological conditions to describe a situation that occurs when there is a diminished supply of O₂ to an organ's tissues. In this article, if necessary, usage of the terms “hypoxic” and “anoxic” is confined to in vivo experiments; the terms “anaerobic” and “aerobic” are used exclusively for in vitro experimental conditions.

1. Structural architecture of the bound nucleotide base and chemistry

Because of the presence of C₆ oxygen on the purine ring, the guanine base has the highest number of conjugations among the purine (including adenine, guanine, and inosine) and pyrimidine (including cytosine, uracil, and thymine) bases. This ring conjugation complexity of the guanine base could endow it with the lowest redox potential of the bases. The E _m ⁷ of the free guanine base (∼1.3 V vs. NHE) is lower than that of CO₃ ^•− and OH^• but higher than that of ^•NO₂ and O₂ ^•− (261). Also, because the stacking interactions of nucleotide bases in DNA are minimal, the redox potential of the guanine base in DNA is unperturbed (125). Thus, the DNA guanine base can be a target of CO₃ ^•− and of OH^• but not of ^•NO₂ or O₂ ^•−.

A guanine nucleotide (GDP or GTP) binds to small GTPases with a high affinity (Fig. 5) (106, 126, 127, 171, 229). The pH-dependent amide (NH) backbone chemical shift of Ras Asp¹¹⁹ in the NKXD motif was negligible, and the solvent exchange rates of the Ras Asp¹¹⁹ NH chemical shift associated with the pH change (from pH 5.9 to 8.0) were slow (103). The results suggest the presence of robust high affinity-binding interactions between the guanine nucleotide base N₁-H and C₂-NH₂ and the Ras Asp¹¹⁹ carboxyl side chain. The mutagenesis of these residues consistently and significantly decreases the guanine nucleotide-binding affinity of Ras (47, 57, 65, 317). In addition, the hydrogen-bonding interaction between the bound nucleotide base C₆-oxygen and the Ala¹⁴⁶ NH of the SAK motif also plays a role in high affinity-binding interactions between Ras and guanine nucleotide (190, 211). An equivalent interaction also is observed in Rho GTPases (66, 109, 115).

The high affinity-binding interactions between the guanine nucleotide base and the NKXD/SAK motif predict resonance states for one-electron oxidized form of the GDP base (a guanine nucleotide cationic radical [G^•+-DP]) bound to small GTPases, including Ras, Rho, and Rab. When these high affinity-binding interactions are absent, the existence of a number of resonance states of G^•+-DP can be expected (Fig. 6). However, introduction of the hydrogen bonds between the N₁-H/C₂-NH₂ of the GDP base and the carboxy side chain of the aspartic acid of the NKCD motif (i.e., Asp¹¹⁹ Ras numbering) stabilizes several resonance forms of the bound GDP base (Fig. 7). Resonance stabilization carries the possibility of lowering the redox potential of molecules (i.e., the G^•+-DP/GDP couple). These analyses suggest that the redox potential of the small GTPase-bound GDP coupled with G^•+-DP is lower than that of the free GDP form or of DNA. The resonance state analysis with GDP can be immediately applicable to the one-electron oxidized form of GTP (i.e., G^•+-TP). This is because the γ nucleotide phosphate with the p-loop binding interactions is unlikely to influence the resonance state of the nucleotide base.

OPEN IN VIEWER

FIG. 6.

Cationic radical resonance states of a free guanine nucleotide base. A number on the electron movement arrow guides a resonance product. P-P indicates a diphosphate group.

OPEN IN VIEWER

FIG. 7.

Resonance stabilizations of the small GTPase-bound guanine nucleotide base. P-P indicates a diphosphate group.

2. Structural architecture of the region of the redox-sensitive motif of small GTPases and redox chemistry

Because the P-loop GXXXXGK(S/T) and the NKXD motifs play key roles in binding interactions between small GTPases and their ligand nucleotide, a mechanical perturbation of the binding interactions between these small GTPase motifs and a nucleotide is likely necessary to release the bound nucleotide (i.e., GDP).

1. Redox-sensitive cysteine in the NKCD motif

Because the solvent-exposed sulfur atom of the Cys¹¹⁸ side chain in the NKC¹¹⁸D motif is vicinal to several electron-withdrawing oxygen atoms of protein residues such as Glu¹⁴³ and Thr¹⁴⁴ (190, 211), the pK _a value of the Cys¹¹⁸ side chain is likely to be smaller than that of Cys-SH and GSH (pK _a = ∼8.3). This analysis suggests that under the physiological pH (e.g., pH 7.4), the Cys¹¹⁸ side chain is likely to be presented, at least in part, as a form of RS⁻. If so, because the redox potential of RS^•/RS⁻ is lower than that of the ^•NO₂/NO₂ ⁻ couple (267), the Ras Cys¹¹⁸ thiolate side chain (Ras Cys-S⁻) is a suitable target for ^•NO₂. Because the redox potential of O₂ ^•− is similar to that of ^•NO₂ (267), this Ras Cys-S⁻ also can be a target of O₂ ^•−. Also, the nucleophilicity of the sulfur atom of Ras Cys-S⁻ is likely to be diminished because the lone pairs of electrons on the sulfur atom are likely to be occupied by the interaction with these multiple electron-withdrawing groups (vide supra). This analysis does not support the potential Ras S-nitrosation (Ras-SNO) mechanisms suggested in Figure 3A and C, in which a nucleophilic attack by a lone pair of electrons of the sulfur atom of a cysteine residue toward the nitrogen atom either of a metal nitrosonium ion adduct or of N₂O₃ will be blocked. This analysis also excludes the possibility of nonradical-based formation of Ras Cys¹¹⁸ sulfenic acid via a reaction between Ras Cys-S⁻ and H₂O₂, because the lone pair of electrons of the sulfur atom that is necessary for a nucleophilic attack toward H₂O₂ to produce Ras Cys¹¹⁸ sulfenic acid is unavailable (Fig. 4A).

2. Phenylalanine vicinal to the NKCD motif

A Ras Switch I residue, Phe²⁸, is conserved in nearly all other small GTPases. The distance between the phenyl group para position (the C₄ carbon atom) of Phe²⁸ side chain and the center of Ras-bound GDP guanine base is ∼3.6 Å (Fig. 5) (29, 119, 190, 211, 244, 276). The perpendicular interface between the Phe²⁸ phenyl side chain and the guanine base results in a n-π stacking interaction (Fig. 5) (29, 39, 119, 188, 190, 211, 244, 276, 318).

The close n-π interaction leads to a suspicion that the plus charge on the para position of the phenyl side chain is close to the π electron-enriched guanine side chain, thereby stabilizing one of the cationic radical resonance states of the Phe²⁸ phenyl side chain (Fig. 8A). Moreover, because of this cationic radical resonance stabilization, a reduction can be expected in the redox potential of the Phe²⁸ phenyl side chain (coupled with the cationic radical form of Phe²⁸ side chain). Alternatively, a resonance state with a positive charge at the meta position, and thus a radical at the para position, was postulated as present (Fig. 8B) (104). However, a radical at the para position of the Phe²⁸ side chain could clash with the π electrons of the guanine base. This electronic state is therefore likely to be less favorable than that of a plus charge at the para position (Fig. 8A). This analysis corrects the misstatement in the previous suggestion (104) as following: Because the radical on the para position of the Phe²⁸ side chain (the C₄ atom of the phenyl side chain) associated with the bound guanine base is in an unfavorable state, the proposed spin transition (sp² → sp³) of the C₄ atom Phe²⁸ side chain is likely to be forbidden.

OPEN IN VIEWER

FIG. 8.

Stabilization and perturbation of the Phe²⁸ side-chain cationic radical associated with the bound guanine nucleotide base. For convenience, although multiple resonance states can exist, only one additional resonance state is shown for each of the cationic resonance forms A and B of the Phe²⁸ side chain. The perpendicular aromatic-aromatic n-π interaction between the Phe²⁸ side chain and the guanine nucleotide base is shown as ≈.

The spatial arrangement among the Phe²⁸ side chain, the redox-sensitive Cys¹¹⁸ side chain, and the bound nucleotide guanine base also is intriguing (Fig. 9). The phenyl side chain of the Phe²⁸ faces the sulfur atom of the Cys¹¹⁸ side chain at an angle of 0°–5°, and no residue is present between these two side chains. The distance between the center of the Phe²⁸ phenyl side chain and the sulfur atom of the Cys¹¹⁸ is ∼12 Å (Fig. 9) (190, 211), which minimizes ionic, hydrogen-bonding, and hydrophobic interactions. However, this distance (∼12 Å) does not limit electron transport between two atoms (210).

OPEN IN VIEWER

FIG. 9.

Spatial architecture of the Phe²⁸ side chain and the redox-sensitive cysteine in the NKCD motif-containing proteins with respect to a bound nucleotide. This figure was generated by using RASMOL with PDB 121P. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article at www.liebertonline.com/ars).

1. Redox-sensitive cysteine(s) associated with the GXXXXGK(S/T)C motif

Similar to the redox-sensitive Cys¹¹⁸ side chain in Ras proteins, the sulfur atom of the side chain of the redox-sensitive cysteine (Cys¹⁸; Rac1 numbering) in the GXXXXGK(S/T)C motif is accessible by solvent and located in an acidic environment (109). With only slight distance variations, all of these interactions with the sulfur atom of the cysteine in the GXXXXGK(S/T)C motif are well conserved in other Rho GTPases such as RhoA and Cdc42 (66, 115). The structural architecture suggests that, as with the sulfur atom of the Ras Cys¹¹⁸ side chain, the sulfur atom of the side chain of the cysteine in the GXXXXGK(S/T)C motif is likely present as an anionic thiolate form (Cys¹⁸ thiolate). Hence, this acidic environment makes the sulfur atom of the Rac1 Cys¹⁸ side chain susceptible to redox agents such as ^•NO₂ and O₂ ^•−. A subset of Rho family GTPases, such as RhoA, RhoB, and RhoC, has additional cysteines, Cys¹⁶, Cys⁶³, and Cys¹⁵⁹ (RhoA numbering; they are not present in Rac1 and Cdc42), close to the redox-sensitive Cys²⁰ (RhoA numbering; equivalent to Rac1 Cys¹⁸) in the GXXXXGK(S/T)C motif (115). Among them, the action of one cysteine residue, Cys¹⁶ (not present in Rac1 and Cdc42), is characterized as a redox sensitive residue in vitro (105).

2. Phenylalanine vicinal to the GXXXXGK(S/T)C motif

Phe²⁸ (Rac1 numbering is the same as Ras numbering) is well conserved throughout all GXXXXGK(S/T)C motif-containing proteins. The Rac1 Phe²⁸ phenyl side chain also is poised perpendicularly with the bound guanine nucleotide base at a distance of ∼3.6 Å (Fig. 10A). The sulfur atom of the Rac1 Cys¹⁸ side chain faces the center of the Phe²⁸ side chain but lies a remote ∼3.6 Å from it (Fig. 10A) (109), whereas the Ras Cys¹¹⁸ side chain is a remote ∼12 Å from the center of the Phe²⁸ side chain (Fig. 9).

OPEN IN VIEWER

FIG. 10.

Spatial arrangement of the Phe²⁸ side chain, a bound nucleotide, and the redox-sensitive cysteine in the GXXXXGK(S/T)C motif-containing GTPases. A and B were generated by using RASMOL with PDB 1MH1 and 1A2B, respectively. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article at www.liebertonline.com/ars).

1. Redox-sensitive cysteines associated with the CGNKXD motif

The Ran crystal structure (266) has revealed that the sulfur atom of the Cys¹²⁰ side chain in the C¹²⁰GNKXD motif can be accessed from the outside. This also suggests that the redox-sensitive Cys¹²⁰ side chain in the C¹²⁰GNKXD motif can be a direct target of redox agents. Although remote from the CGNKXD motif, Ran has another redox-sensitive cysteine Cys⁸⁵ (98). The sulfur atom of the Cys⁸⁵ side chain is exposed to solvent and close to several oxygen atoms of Ran residues, including Lys¹² and Gln⁸⁴ (266). However, only a few functional groups interact with the Cys¹²⁰ side chain (266). Hence, the sulfur atom of the Cys⁸⁵ side chain is likely to be poised in a relatively acidic environment. This environment, in turn, encourages speculation that the redox potential of the Cys⁸⁵ side chain is less than that of the Cys¹²⁰ side chain. Also, the sulfur atom of the cysteine Cys¹²⁰ side chain is ∼11.8 Å from the sulfur atom of the side chain of Cys⁸⁵ (266). Intriguingly, no residue is present between these sulfur atoms of the cysteine Cys¹²⁰ and Cys⁸⁵ side chains.

2. Multiple packed phenylalanines

Notably, the Ras and Rho GTPase Phe²⁸ side chain interacts perpendicularly with the bound-GDP guanine base that plays a role in Ras nucleotide-binding interactions (104). However, the Ran Phe²⁶ (sequentially equivalent to the Ras Phe²⁸) phenyl side chain is located away from the bound-GDP guanine base (Fig. 11) (266). Unlike the Ras, Rho, and Rab proteins, the Ran GTPase has an unusually high number of phenylalanines (Fig. 11). Uniquely, in an arrangement that also includes this Ran Phe²⁶ side chain, a series of Ran phenylalanine residues are stacked one with another (Fig. 11). These stacked Ran phenylalanine residues are present between Ran Cys¹²⁰ and Cys⁸⁵ side chains: (i) The Phe²⁶ side chain is vicinal to the redox-sensitive Cys¹²⁰ side chain in the CGNKXD motif at a distance of ∼5.5 Å; (ii) this Phe²⁶ side chain also packs against the Phe¹⁵⁷ side chain in a distorted coplanar arrangement; (iii) the Phe¹⁵⁷ side chain packs against the side chain of Phe¹⁶¹, which then faces another Ran phenylalanine residue, the Phe⁵² side chain; (iv) this Phe⁵² side chain stacks with the side chain of Phe¹⁷⁶, which also faces the side chain of Phe⁶¹; (v) the Phe⁶¹ side chain packs against the Phe¹¹ side chain; and (vi) finally the Phe¹¹ side chain faces a sulfur atom of the side chain of Cys⁸⁵, another redox-sensitive Ran cysteine. Including Ran Phe²⁶, the multiple-stacked Ran phenylalanines are termed herein as a Ran aromatic cluster, which consists of seven phenylalanines. For convenience, the seven phenylalanines in the Ran aromatic cluster are numbered from 1 to 7 as Phe²⁶ ₁, Phe¹⁵⁷ ₂, Phe¹⁶¹ ₃, Phe⁵² ₄, Phe¹⁷⁶ ₅, Phe⁶¹ ₆, and Phe¹¹ ₇ (Fig. 11). The average distance between each Ran phenylalanine residue side chain is ∼5 Å. A sequence analysis suggests that although some Ran phenylalanines are matched to tyrosines, the residues assigned for the Ran aromatic cluster are virtually conserved in Dexras1 and Rhes proteins (98). Considering the structural features of the Ran aromatic-stacking cluster (Fig. 11) in conjunction with the packing distances, the main force held in the stacked-aromatic cluster is likely to be a stacked π − π interaction. In fact, such a π − π stacking interaction has been observed in hemoglobin in which the Tyr²⁴ and His²⁰ side chains are contained in a coplanar arrangement within an hemoglobin α chain pack at a distance of 4.8 Å (58).

OPEN IN VIEWER

FIG. 11.

The structural architecture of CGNKXD motif-containing GTPase with respect to its bound nucleotide. This figure was generated by using RASMOL with PDB 1BYU. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article at www.liebertonline.com/ars).

3. Structural relevance of Ran GTPases redox motifs with respect to Dexras1 and Rhes GTPases

Studies have identified a redox-sensitive cysteine, Cys¹¹ (Dexras1 numbering; equivalent to Rhes Cys¹⁰), in Dexras1 (63, 120). A sequence analysis shows that both Dexras1 and Rhes GTPases possess the CGNKXD motif (98). However, the composition of the aromatic cluster of Dexras1 and Rhes proteins differs slightly from that of the Ran aromatic cluster (98): Many tyrosines, in lieu of phenylalanines, are present in the Dexras1 and Rhes GTPases. The structures of Dexras1 and Rhes are unknown. However, given that the Dexras1 and Rhes protein sequences are similar to that of Ran GTPase (98), the structure of Dexras1 and Rhes can be predicted by using SWISS-MODEL (88, 246) with the Ran PDB coordinate 1BYU as a reference (266). A comparison of the Ran crystal structure and the SWISS-MODEL analysis for Dexras1/Rhes further suggests that: (i) The putative redox-sensitive Dexras1 Cys¹¹ side chain (same as the Rhes Cys¹⁰ side chain) is structurally equivalent to the Ran Cys⁸⁵ side chain; (ii) the Ran aromatic-stacked cluster also is conserved, with some variations, in Dexras1 and Rhes GTPases; and (iii) the order of the packed-aromatic cluster of Dexras1 and Rhes proteins does not completely match with Ran.

1. Redox-induced chemical modification-based protein conformational change mechanisms for the NKCD motif-containing GTPases

The redox-sensitive Ras NKCD motif was discovered before identification of the GXXXXGK(S/T)C and CGNKXD motifs. Consequently, the first attempt to suggest a redox regulation mechanism for GTPases was made for a Ras protein that contains the NKCD motif. Because the redox architecture of GTPases is virtually identical within its motif type (i.e., the NKCD motif-containing GTPases), the mechanism proposed for the Ras protein can be applied to other proteins that contain the NKCD motif.

2. Chemical reaction process-based mechanism for NKCD or GXXXXGK(S/T)C motif-containing GTPases

An in vitro kinetic analysis shows that O₂ is essentially required for NO-mediated Ras nucleotide dissociation (104). Because NO reacts with O₂ to produce ^•NO₂, this result was interpreted to mean that ^•NO₂ is the active redox agent for the facilitation of the dissociation of nucleotide from Ras. This conclusion was further confirmed by using authentic ^•NO₂ gas that facilitates Ras nucleotide dissociation in the absence of O₂ (101). N₂O₃, which is another NO reaction product with O₂, was eliminated from the list of possibly active Ras-targeting redox agents because N₂O₃-mediated Ras nucleotide dissociation, under anaerobic conditions, was minimal (99, 104). The action of the NO/O₂ mixture (containing the active redox agent ^•NO₂) on the Ras mutant C118S was ineffective, suggesting that Ras Cys¹¹⁸ in an NKCD motif is the target site for ^•NO₂. Notably, ^•NO₂ reacts with the sulfur atom of Cys-S⁻ or GS⁻, but not with Cys-SH or GSH, respectively, to produce Cys-S^• or GS^• (see Section I). Hence, identification of ^•NO₂ as an agent that targets the Ras NKCD motif suggests that Ras Cys¹¹⁸ thiyl radical (Cys¹¹⁸-S^•) is likely to be formed by the reaction between ^•NO₂ and Ras Cys¹¹⁸-S⁻. A prediction on the formation of Ras Cys¹¹⁸-S^• also is consistent with the Ras structural analysis and suggests a hypothesis that the Ras Cys¹¹⁸ side chain is in a deprotonated state (see Section II). However, this analysis leads one to correct the previous statement that the state of Cys¹¹⁸-S⁻, but not the Cys¹¹⁸-SH state, is a target of ^•NO₂ (99, 104). This analysis also leads to a hypothesis that the action of the ^•NO₂-mediated Ras nucleotide dissociation is associated with the formation of Ras Cys¹¹⁸-S^• and thus is a radical-based process.

An unexpected but important factor to consider for the redox regulation mechanism of Ras activity is that a chemically modified nucleotide 5-guanidino-4-nitroimidazole ribose diphosphate (NIm-DP) or a product of 5-oxo-GDP degradation, but not an intact nucleotide, was detected from the Ras sample treated with NO/O₂ or O₂ ^•−, respectively (101, 102, 104). The result indicates that the bound nucleotide base is involved in the dissociation process of the redox-mediated Ras nucleotide. The redox potential of a free guanine base, coupled with G^•+-DP, is ∼1.3 V (vs. NHE) (261). Only if the redox potential of the bound GDP base is lower than that of Ras Cys¹¹⁸-S^•/Ras Cys¹¹⁸-S⁻, can Ras Cys¹¹⁸-S^• abstract an electron from the bound GDP base to produce Ras Cys¹¹⁸-S⁻ and G^•+-DP. Therefore, as analyzed in Section II, it can be postulated that the tight hydrogen bonding interactions between the bound nucleotide base and the NKCD/SAK residues lower the redox potential of the bound nucleotide base below the redox potential of the sulfur atom of the Cys¹¹⁸ side chain.

The sulfur atom of the Cys¹¹⁸ side chain does not face the nucleotide base (Fig. 9) (29, 119, 211, 244, 276). Moreover, the Asp¹¹⁹ carboxy side chain lies between the sulfur atom of the Cys¹¹⁸ side chain and the bound nucleotide base (Fig. 9). These structural architectures suggest that, although oxidation of the bound nucleotide base coupled with a reduction of Ras Cys¹¹⁸-S^• is feasible in terms of redox potential, a direct electron transport from the bound nucleotide base to the sulfur atom of the Cys¹¹⁸ side chain is impossible. Hence, an indirect interaction between the bound nucleotide base and the formed Ras Cys¹¹⁸-S^• also requires consideration. As analyzed in Section II, the Phe²⁸ phenyl side chain interacts perpendicularly with the bound nucleotide base and faces the sulfur atom of the Cys¹¹⁸ side chain (Fig. 9). Therefore, the Phe²⁸ phenyl side chain is likely a mediator of a possible electronic interaction between the bound nucleotide base and Ras Cys¹¹⁸-S^•. Ras-released NIm-DP was not detected from the NO/O₂-treated mutant Ras F28L, which lacks the Phe²⁸ side chain (Ras F28L) (104). The result of the mutant study, in conjunction with the structure of the Phe²⁸ phenyl side chain in the NKCD motif-containing GTPase, suggests a hypothesis that the π electron-enriched Phe²⁸ phenyl side chain serves as a conduit for transfer of an electron from the bound nucleotide base to Ras Cys¹¹⁸-S^• to produce G^•+-DP and Ras Cys¹¹⁸-S⁻.

a. RNS-mediated molecular mechanism of the NKCD motif-containing GTPases

On the basis of the results of these several studies and analyses (vide supra), Heo et al. have proposed a distinctive radical-based molecular mechanism to explain RNS-mediated Ras activity (Fig. 12): A reaction of the solvent-exposed sulfur atom of Ras Cys¹¹⁸-S⁻ with ^•NO₂ produces Ras Cys¹¹⁸-S^• (Fig. 12, Step 1). The Ras Cys¹¹⁸-S^• formed in this reaction abstracts an electron from the bound nucleotide base via the Phe²⁸ side chain to produce Ras Cys¹¹⁸-S⁻ and G^•+-DP (Fig. 12, Steps 2 and 3). On the basis of the resonance stabilization analysis in Section II, a cationic radical intermediate state is proposed for the Phe²⁸ side chain in which a plus charge and a radical, respectively, are localized at the para and meta positions of the Phe²⁸ side chain. As with the DNA cation radical G^•+ (125, 194, 251), release of a proton leads to production of a relatively stable form of G^•-DP (Fig. 12, Steps 4, 5, and 6). Another ^•NO₂ reacts with G^•-DP to produce a GDP-NO₂ adduct (Fig. 12, Step 7). Formation of G^•-DP perturbs the hydrogen bonding interaction between the Ras NKCD/SAK motif and the bound nucleotide base. This perturbation results in release of the bound nucleotide as a form of GDP-NO₂ and production of the nucleotide-deficient Ras that has Cys¹¹⁸-S⁻. As with the guanine-NO₂ adduct in DNA (87, 206), the free GDP-NO₂ is hydrated and then decarboxylated to produce NIm-DP (Fig. 12, Steps 8, 9, and 10). The nucleotide-deficient Ras that is formed, in turn, binds with abundant GTP to produce a GTP-bound active Ras in cells. However, if GDP, instead of GTP, is populated, the active form of Ras will not be formed. Hence, this radical-based molecular mechanism does not dictate the state of active or inactive Ras but only facilitates nucleotide dissociation from Ras.

OPEN IN VIEWER

FIG. 12.

Proposed mechanism for the ^•NO₂-mediated dissociation of a bound nucleotide from the NKCD motif-containing proteins. The proposed mechanism includes detailed electronic states and chemical reaction steps associated with radical-based nucleotide modification and dissociation from the NKCD motif-containing GTPases. The mechanism illustrated here also corrects an earlier description (104) in which the state of the redox-sensitive cysteine for the ^•NO₂-mediated nucleotide dissociation was shown as a thiol (Cys¹¹⁸-SH) instead of as a thiolate (Cys¹¹⁸-S⁻). Hydrogen-bond interactions between protein residues and GDP are shown with dotted lines. GDP, guanine nucleoside diphosphate.

A certain ROS, O₂ ^•−, also enhances wild-type (wt) Ras nucleotide dissociation (101, 102). Like ^•NO₂, O₂ ^•− can react with the sulfur atom of Cys-S⁻ or GS⁻, but not with either Cys-SH or GSH, to produce Cys-S^• or GS^• (see Section I). As with ^•NO₂, the O₂ ^•−-mediated Ras C118S nucleotide dissociation was limited, suggesting that the Ras Cys¹¹⁸-S⁻ also is a target of O₂ ^•− to produce Ras Cys¹¹⁸-S^•. Because this Ras Cys¹¹⁸-S^• is the same radical species that is generated by ^•NO₂ with Ras, the O₂ ^•−-mediated Ras activity modulation also is likely to be based on the radical Ras Cys¹¹⁸-S^•.

1. Relative redox potential of the Phe²⁸ phenyl side chain as well as the bound nucleotide base and Ras Cys¹¹⁸-S⁻

Although the redox potential of a free phenylalanine side chain is unknown, this potential has been predicted to be close to or less than 2.1 V (vs. NHE) (145, 283). The redox potentials of Ras Cys¹¹⁸-S⁻ and the bound nucleotide base have been predicted to be Ras Cys¹¹⁸-S^•/Ras Cys¹¹⁸-S⁻ (∼0.8 V vs. NHE) >G ^•+-DP/GDP (<∼0.8 V vs. NHE; vide supra). Hence, it is apparent that electron transfer through the Phe²⁸ side chain is kinetically unfavorable. As analyzed in Section II, the close n-π interaction between the Phe²⁸ phenyl side chain and the bound nucleotide base could lower the redox potential of the Phe²⁸ phenyl side chain via stabilization and localization of one of the cationic radical states of the Phe²⁸ phenyl side chain (plus a charge on the C₄ atom of the phenyl side chain). However, it is unclear how much the redox potential of the Phe²⁸ phenyl side chain declines because of the close n-π interaction with the bound nucleotide base. Nevertheless, it is predictable that the redox potential of the Phe²⁸ side chain of the GTPase-bound NKCD motif-containing GTPases may not be as high as in a free phenylalanine phenyl side chain (>∼2.1 V vs. NHE) (145, 283). Even if the redox potential of a Phe²⁸ side chain were lower than that of a free phenylalanine side chain, the redox potential of the Phe²⁸ side chain still could exceed that of a Cys¹¹⁸ side chain and thus be higher than that of the bound nucleotide base. If so, an electron transfer from the bound GDP base to Ras Cys¹¹⁸-S^• via the Phe²⁸ side chain includes an uphill step of Phe²⁸ to Ras Cys¹¹⁸-S^• to produce a Phe²⁸ side chain cationic radical (Phe^28•+) and Ras Cys¹¹⁸-S⁻ (Phe²⁸ → Ras Cys¹¹⁸-S^•), respectively, followed by a downhill process from the bound GDP base to Phe^28•+ to produce Phe²⁸ and G ^•+-DP (GDP base → Phe^28•+). This brings up the question of how this uphill electron transfer proceeds.

2. Application of Marcus theory to the Phe²⁸ side chain-mediated electron transfer between the bound nucleotide base and thiyl radical of Ras GTPases

A theory has been developed that contends that electron transfer in proteins is regulated by pathways that are optimal combinations of through-bond, hydrogen bond, and through-space (19). The driving force of a reaction, as predicted by this Marcus theory (183, 184), also affects the rate of electron transfer in and between proteins (134). Studies such as these allow evaluation of reorganization energies, which strongly influence reaction rates. Briefly, the redox event associated with GTPases that contain the NKCD motif can be expressed as an energy diagram (Fig. 13). In accordance with Marcus theory (129, 183, 184), the endergonic tunneling steps (¹ k _et ^en and ³ k _et ^en; Steps 1 and 3, respectively, in Fig. 13) are concomitant with the exergonic electron transfer events (² k _et ^ex and ⁴ k _et ^ex; Steps 2 and 4, respectively, in Fig. 13). In physiological reactions, endergonic tunneling associated with ¹ k _et ^en and ³ k _et ^en (particularly ¹ k _et ^en in Fig. 13) has come to be considered a biologically relevant process, but only with some reluctance. However, thermally assisted tunneling (quantum probability) up to ∼0.5 eV uphill has been shown in photosynthetic systems (210, 300). As discussed above, the value of the redox potential of the Phe²⁸ side chain is expected to be less than 2.1 V (vs. NHE). Hence, ¹ k _et ^en between the Phe²⁸ side chain and Cys¹¹⁸-S^• would not be very endergonic. Besides, a previous study has explicitly noted that “although the overall reactions are usually modestly favorable, these chains often include VERY endergonic steps” (210). A number of uphill electron transfer events in various proteins, including azurin, myoglobin, and cytochrome complexes, also have been reported (51, 85). For example, although the redox potential of an aromatic amino acid Trp¹⁴² in cytochrome b is expected to be much higher than that of the two redox centers ubiquinol and cytochrome c ₁ (267), Trp¹⁴² intervenes in electron transfers between these two redox centers (28).

OPEN IN VIEWER

FIG. 13.

The energy surface diagram of the NKCD motif-containing GTPases for electron transfer. The far left and far right of the energy diagram represent the electronic states of the Cys¹¹⁸-S^•-Phe²⁸-GDP base and Cys¹¹⁸-S⁻-Phe²⁸-GDP^•+, respectively. The bottom middle parabola curve (between these two adiabatic curves) denotes the state of the Cys¹¹⁸-S⁻-Phe²⁸ side chain cationic radical-GDP base (H _J). The top middle curve (nonadiabatic region) represents the state of the Cys¹¹⁸-S^•-GDP base, which lacks the Phe²⁸ side chain energy surface.

In Marcus theory (183, 184), the electron transfer rate (k _et; a combination of ¹ k _et ^en, ² k _et ^ex, ³ k _et ^en, and ⁴ k _et ^ex) between redox centers is almost inversely proportional to the edge-to-edge distance and roughly proportional to the protein packing density (ρ) in the volume between redox centers; and ρ is nearly inversely proportional to the square root of the barrier height (Fig. 13) (210). A study also has shown that ligands with π systems provide easy passage of electrons for outer-sphere redox reactions, where k _et for the [Co(phe)₃]³⁺/[Co(phe)₃]²⁺ system is 40 M/s, which is many orders of magnitude faster than that for [Co(NH₃)₃]³⁺/[Co(NH₃)₃]²⁺ (64). Electron tunneling between redox centers via aromatic side chains (ρ = 0.77 ± 0.09) is consistently faster than that for electron tunneling through a vacuum (ρ = 0) (85, 210). These results lead one to suggest that the enriched π electrons in the Phe²⁸ side chain function via an electron tunneling mechanism as an electron transfer-intervening medium to mediate highly directional uphill electron transfer between the Cys¹¹⁸ side chain and the bound nucleotide base (Fig. 13).

In summary, the molecular mechanism from the bound nucleotide base to Ras Cys¹¹⁸-S^• via a Phe²⁸ side chain (Fig. 12) can be explicitly tethered with the corresponding energy diagram (Fig. 13): (i) Although the first endergonic and exergonic tunneling steps (¹ k _et ^en and ² k _et ^ex; Steps 1 and 2, respectively, in Fig. 13) correlate with the reaction Step 2 in Figure 12, the transient energy minimized state H _J located between adiabatic curvatures (Fig. 13) couples with the product of Step 2 in Figure 12; and (ii) the second endergonic and exergonic processes in the energy diagram (³ k _et ^en and ⁴ k _et ^ex; Steps 3 and 4, respectively, in Fig. 13) correspond to Step 3 in Figure 12.

3. Potential mechanism for the action of redox agents on GTPases that possess the CGNKXD motif

A recent study shows that Ran GTPase can be activated by ^•NO₂, one of the RNS, but not by either NO or N₂O₃ (98). The target Ran redox sites are identified as cysteine residues (the solvent-exposed Cys⁸⁵ and Cys¹²⁰ in the CGNKXD motif ) (105). It must also be taken into account that an intact nucleotide is released as an end product of ^•NO₂-mediated Ran nucleotide dissociation (98). This result suggests that the bound guanine nucleotide may not be involved in the ^•NO₂-mediated Ran nucleotide dissociation. This also strongly implies that the redox-regulation mechanism for Ran is likely to differ from the radical-based GDP modification mechanism for Ras and Rho GTPases.

Intriguingly, although the Ran Phe²⁶ side chain is away from the bound nucleotide (Fig. 11) (266), the Ran Phe²⁶ plays a role in the ^•NO₂-mediated Ran nucleotide dissociation (105). Also, results from CD spectra suggest that, although treatment of wt Ran with ^•NO₂ alters its secondary structure, its treatment with ^•NO₂ had no effect on the CD spectrum of Ran F161L (unpublished data). Given that the side chains of Ras Phe²⁶ and the Phe¹⁶¹ belong to the packed Ran phenylalanine residues (see Section II), it can be hypothesized that the Ran aromatic cluster is involved in the redox response of the Ran protein. Noticeably, the stacked aromatic cluster is placed between these two redox-sensitive cysteines, Cys⁸⁵ and Cys¹²⁰ (see Section II).

a. Potential molecular mechanism of the redox regulation of GTPases that contain the CGNKXD motif

These results in conjunction with the structural and sequence analyses of Ran, speculation about the nature of a potential molecular mechanism is possible (Fig. 14): A redox-competent RNS (e.g., ^•NO₂) or ROS (e.g., O₂ ^•−) reacts with the solvent-accessible Ran Cys¹²⁰ thiolate side chain (Cys¹²⁰-S⁻) located in the CGNKXD motif to produce the Ran Cys¹²⁰ thiyl radical (Cys¹²⁰-S^•). In turn, this Cys¹²⁰-S^• couples with the Ran Cys⁸⁵ thiolate (Cys⁸⁵-S⁻) via the Ran aromatic cluster to produce Cys¹²⁰-S⁻ and Cys⁸⁵-S^•. The Cys⁸⁵-S^• thus formed can be quenched by NO (if NO is present) to produce an S-nitrosated Ran (44). The electronic coupling between these two Ran redox centers generates a temporal Ran aromatic cluster cationic radical (aromatic cluster^•+). The formed transient aromatic^•+ subsequently perturbs the π-π stacking interactions in the Ran aromatic cluster. The result is alteration of the Ran guanine nucleotide-binding site. The altered conformation of the Ran guanine nucleotide-binding site perturbs the Ran guanine nucleotide-binding interaction and leads to the release of an unmodified nucleotide that results in formation of a nucleotide-deficient Ran. Cellular abundant GTP binds to the nucleotide-deficient Ran to produce an active Ran (Fig. 14).

OPEN IN VIEWER

FIG. 14.

Proposed redox-regulation mechanism for Ras-related nuclear protein GTPase.

This postulated mechanism requires an assumption that the relative redox potentials among three Ran redox components are ordered as follows: Cys¹²⁰-S^•/Cys¹²⁰-S⁻ > Cys⁸⁵-S^•/Cys⁸⁵-S⁻. This assumption of the relative redox potential between the sulfur atom of Cys¹²⁰ and the Cys⁸⁵ side chain is based upon an analysis of the local structural environments of the Ran protein (see Section II).

Because the sulfur atoms of Ran Cys¹²⁰ and Cys⁸⁵ face each other from a distance of ∼11.8 Å, a direct electron transfer from Cys⁸⁵-S⁻ to Cys¹²⁰-S^• (or vice versa; from Cys¹²⁰-S⁻ to Cys⁸⁵-S^•) is possible without involvement of the Ran aromatic cluster. However, this direct electron transfer is a less attractive possibility than a transfer via the aromatic cluster because electron transfer through a vacuum is less favorable than through an electron-rich medium (i.e., via a π electron-enriched medium) (48).

1. Mechanistic properties of the Ran aromatic cluster and Marcus theory

Although a detailed experiment should be performed to elucidate a true molecular mechanism and/or test the potential proposed mechanism, Marcus theory (183, 184) and the electron tunneling hypothesis (19, 84, 134, 162, 202, 210) can partly answer a few remaining questions associated with the electron transfer mechanism and the redox-induced Ran protein conformational change. One of the key unanswered questions is, how does the packed aromatic cluster serve as an electron conduit if its redox potential is higher than those of the Cys¹²⁰ and Cys⁸⁵ side chains (i.e., aromatic cluster^•+/aromatic cluster >Cys¹²⁰-S^•/Cys¹²⁰-S⁻ > Cys⁸⁵-S^•/Cys⁸⁵-S⁻)? Accounting for the postulation of the relative redox potentials of the Ran redox-sensitive cysteines and the aromatic cluster (vide supra), the Ran redox components (i) Cys¹²⁰ (redox agent target) and (ii) Cys⁸⁵ (electron donor) can be matched to the redox components in the NKCD motif-containing GTPases. These components are, respectively, (i) Cys¹¹⁸ (the redox agent target) and (ii) the bound nucleotide (the electron donor). Although the electron mediator for the NKCD motif-containing GTPases is a single phenylalanine residue (Fig. 12), Ran, in contrast, has packed multiple phenylalanines. However, an earlier study has clearly noted that tunneling often includes multiple endergonic steps (210). Hence, the Ran aromatic cluster can be considered to be functionally matched to the Ras Phe²⁸ (electron transfer mediator). Thus, the energy diagram for these redox components for the NKCD motif-containing GTPases (Fig. 13) also can be readily applicable to the Ran protein electron transfer. The energy diagram (Fig. 13) illustrates that, overall, the electron transfer process from the Cys⁸⁵ side chain to the Cys¹²⁰ side chain is thermodynamically favorable (if Cys¹²⁰-S^•/Cys¹²⁰-S⁻ < Cys⁸⁵-S^•/Cys⁸⁵-S⁻, the reverse is true). However, as with the electron transfer in GTPases that contain the NKCD motif, the predicted path of the electron transfer from the Ran Cys⁸⁵ side chain to the Cys¹²⁰ side chain via the aromatic cluster includes a series of endergonic and exergonic electron transfer events (Steps 1, 2, 3, and 4 in Fig. 13). Notably, significantly larger ρ-values can be found in the π systems of aromatic residues (64, 85, 210). For this reason, by taking Marcus theory into account (183, 184), electron tunneling between two Ran cysteine redox centers via aromatic cluster π systems is likely to be feasible.

2. Redox-mediated activity regulation of Dexras1 and Rhes GTPases

The SWISS-MODEL analysis (see Section II), in combination with redox chemistry, predicts that the redox architectures of Dexras1 and Rhes GTPases are similar to that of the Ran protein. Hence, it can be speculated that the mechanism proposed for the Ran protein is applicable to the action of RNS and/or ROS on Dexras1 and Rhes proteins. Further rigorous studies are essential to elucidate the molecular mechanism for the RNS/ROS-mediated activation of Dexras1 and Rhes GTPases.

1. Upregulation of Ras GTPases by redox agents

Lander et al. showed that NO activates Ras by enhancing Ras GNE in human T cells (158). Another study has shown that treatment with N-methyl-d-aspartate (NMDA)-activated nNOS in primary cortical neuronal rodent cultures resulted in activation of Ras (312).

Several Ras-dependent cellular signaling cascades are implicated in the redox responses of Ras (Fig. 15). For example, various oxidative agents (i.e., H₂O₂ and hemin) also could activate Ras by enhancing Ras GNE and thus upregulate the mitogen-activated protein kinase (MAPK) pathway in Jurkat cells (Fig. 15) (159). This study also showed that expression of a dominant negative Ras mutant in Jurkat cells blocked signaling by these oxidative agents. This result, which was assessed by Ras-dependent activation of the nuclear factor κ-light-chain-enhancer of activated B cells (NF-κB), suggests that direct activation of Ras (Fig. 15) may be a central mechanism by which a variety of oxidative redox stress stimuli transmit their signal to the nucleus. Moreover, the NO-mediated activated Ras results in tyrosine phosphorylation of various cellular proteins (i.e., focal adhesion kinase and Src kinase) through stimulation of MAPK pathway (Fig. 15) (200). NO-mediated Ras activation also stimulated extracellular signal-regulated kinases, c-Jun terminal kinase (JNK) and extracellular signal-regulated kinase (ERK), respectively, via the phosphatidylinositol 3′-kinase (PI3K) and Raf signal transduction pathways (Fig. 15) (55, 56). A more recent study has shown that exogenous addition of H₂O₂ induces chemical modification of the Ras redox site; this modification facilitates activation of the antiapoptotic serine threonine kinase (Akt) but not activation of ERK (2).

OPEN IN VIEWER

FIG. 15.

Ras downstream signaling transduction cascades. Solid arrows indicate Ras-dependent stimulatory transduction paths between cellular signaling proteins. Direct redox stimuli on Ras GTPase are represented with dotted arrows. RNS- or ROS-mediated upregulation of Ras cascades various biological responses via its downstream pathways as indicated. *Raf, a Ras downstream effector protein, is one of the mitogen-activated protein kinase kinases (MAPKKKs, MAP3Ks or MKKKs). ^†MEK1/2 and MKK4/7 belong to MAPKKs (also termed MAP2Ks or MKKs). ^‡ERK1/2 and JNK1/2/3 are members of MAPKs. ERK, extracellular signal-regulated kinase; JNK, c-Jun terminal kinase; ROS, reactive oxygen species.

1. RNS

The hypoxia-inducible transcription factors that bind to hypoxia-responsive elements (HREs) are key players in cellular responses to changes in oxygen tension as well as their responses to ROS (141). One early study suggested that HREs were associated with Ras and MEK1 in human prostate cells (LNCaP and PC3), and NO was a major signaling intermediate responsible for the HRE induction in these cells (253). Overexpression of iNOS has consistently been shown to be directly involved in deregulation of the growth of hepatocellular carcinoma cell specimens from rats, mice, and humans via activation of NF-κB, which couples with the HRas-dependent ERK pathway (Fig. 15) (31). Further, cell compartment-dependent Ras redox regulation coupled with the ERK pathway has been reported (114). This study showed that upregulation of eNOS activates NRas but not KRas on the Golgi complex of T cells. This NRas activation was due to the formation of Ras-SNO (see Section II) at the redox-sensitive Ras NKCD motif residue Cys¹¹⁸, which induces T cell death. In certain human breast cancer cells (MDA-MB-231, MCF-7, and MDA-MB-468), not only the Raf-MEK1/2-ERK1/2 cascade but also the PI3K/Akt pathway was shown to be upregulated via Ras activation (Fig. 15), and this action of NO was independent of guanosine 3′,5′-cyclic monophosphate (cGMP) (217). Lim et al. showed that a constitutively active oncogenic Ras (e.g., KRas G12V) in mice stimulates the PI3K-Akt signaling cascade, which in turn activates eNOS by phosphorylation (174). This activated eNOS then further activates other wt Ras isoforms, such as H and KRas GTPases, via formation of Ras-SNO at their redox-sensitive site Cys¹¹⁸ in the NKCD motif (174). The study suggests that this oncogenic Ras-PI3K-Akt pathway coupled with the eNOS-Ras cascade is, at least in part, responsible for initiation and maintenance of oncogenic Ras-mediated growth of tumors. Moreover, a study using a rat intestinal epithelial cell line (IEC-6) shows that an activating mutation of KRas leads to enhancement of the iNOS expression via various transcription factors, including NF-κB, and the endogenously produced NO contributes to tumorigenesis and continuation of the tumor growth of KRas-transformed cells (273). In addition, the MAPK pathway (Fig. 15) also participates in redox-mediated tumorigenesis associated with Ras activation. A study using human melanoma cell lines (WM793 and WM35) shows that activation of mutations of NRas enhances iNOS expression in human melanoma via activation of the MAPK pathway (62). Intriguingly, a potentially reversible process, induction of Ras-activating mutation by upregulation of NOS, also has been investigated. Patients with acute myeloid leukemia had a high expression of iNOS, but there was no correlation between expression of iNOS and expression of K, H, and NRas mutation and expression (25). However, another study reported that treatment of Sencar mice with a topical application of an NO releasing agent, (±)-(E)-4-methyl-2-[(E)-hydroxyimino]-5-nitro-6-methoxy-3-hexenamide, enhances tumor-initiating activity via induction of mutation in HRas at codon 61 or 13 to produce a constitutively active oncogenic HRas GTPase (241).

2. ROS

Irani et al. shows that NIH 3T3 cells transformed with the constitutively active HRas G12V produce large amounts of ROS (118). This study also shows that the MAPK activity was decreased and JNK activity was unchanged in HRas G12V-transformed cells. This suggests that the mitogenic activity of cells induced by HRas G12V associated with ROS was independent of the MAPK and PI3K-JNK pathways (Fig. 15). The same research group also shows that one ROS, O₂ ^•−, plays a role as a signaling molecule in phagocytic cells, such as modulating the activity of Ras and Rac (117). Another study shows that, in human gastric adenocarcinoma cells, the Ras-dependent ERK1/2 pathway is a common path the response for the response to oxidative stress associated with ROS (Fig. 15) (111). However, given that ROS can activate ERK1/2 signaling cascades via inhibition of protein phosphatases, ROS likely activates the ERK1/2 signaling pathway at multiple points (167, 227, 292). For example, Ras expression is not necessarily required for ROS-mediated activation of the ERK1/2 signaling pathway: ROS induces the activation of the ERK1/2 signaling pathway in cells prepared from ventricles of 1-day-old Wistar rats and overexpressed with a dominant negative Ras mutant (320). This may be due to in part to the c-Src-mediated phosphorylation and activation of phospholipase Cγ (286). Also, blockage of MEK1/2, an upstream protein of ERK, by its inhibitor (U0126 or PD98059) results in inhibition of oxidative stress-induced ERK1/2 activation (166, 169). This result suggests that oxidative stress signaling is not exerted directly on ERK1/2 but is targeted at upstream cascades such as Ras.

ROS are shown to modulate cellular oxidative stress at the level of transcription. A recent study using fibroblast and osteosarcoma cells shows an explicit role in cancer by Ras associated with ROS in which Ras induces expression of an oncogenic transcription factor, the forkhead box protein M1, and the action of Ras was O₂ ^•− dependent (212). The forkhead box protein M1 is overexpressed in most human carcinomas (178, 220, 281). The Ras-dependent protein kinase C (PKC) pathway in gastric cancer (AGS-B) cells also has been shown to be responsible for oxidative stress-mediated activation of the histidine decarboxylase promoter (111, 112).

One study has shown that a constitutively active oncogenic Ras variant upregulates expression of Nox1 in NIH 3T3 cells (268) via the Ras-dependent MAPK pathway (195). Conversely, this study also showed that use of small interfering RNAs to inhibit Nox expression blocked transformation of phenotypes (anchorage-independent growth, morphological changes, and production of tumors in athymic mice) associated with Ras upregulation (195). The result of this study suggests that the production of ROS by Nox is required for oncogenic Ras transformation. Nox1 consistently plays a key role in Ras-induced tumor angiogenesis, possibly via activating the Raf-MEK1/2-ERK1/2 pathway-dependent transcription factor Sp1 in transformed normal rat kidney cells and gastric cancer (150, 243). Suppression of Nox activity has been shown to be able to block the oncogenic HRas-mediated growth of NIH 3T3 cells (42). Transcriptional regulation of Nox1 by Ras also has been investigated with CaCO-2 cells derived from a certain human colon adenocarcinoma (72, 136). The presence of active Ras in CaCO-2 cells also correlates with expression of Nox1 mRNA (3). An in vivo study using transgenic mice shows a correlation between Nox1 mRNA expression and activation of mutations in Gly¹² and Gly¹³ of KRas (164). This correlation is tightly coupled with incidences of colorectal cancer.

1. NOS

An in vivo study using rats has shown that blockage of NO synthesis by inhibiting NOS results in inhibition of Ras signaling that attenuates end-organ damage during severe hypertension and endothelial dysfunction (18). Oliveira et al. showed that a Ras-dependent Raf-MEK1/2-ERK1/2 signaling pathway (Fig. 15) in rabbit aortic endothelial cells was stimulated by sodium nitroprusside (SNP) or SNAP as an NO source and a cGMP analog, 8Br-cGMP (208, 209). Importantly, this study also suggested a Ras activation mechanism in which the transamination of the NO moiety of S-nitrosoglutathione or SNAP onto the Cys¹¹⁸ site Ras produces active Ras-SNO. Apurinic/apyrmidinic endonuclease/redox factor-1 (APE1/ref-1) plays a key role in DNA repair and also governs the reductive activation of many redox-sensitive transcription factors (285). Using APE1/ref-1^+/–mice, it was shown that APE1/ref-1 also regulates endothelium-dependent NO production and vasomotor tone via HRas (122). Interestingly, the action of ONOO⁻ for the activation of ERK in cardiomyocytes derived from a rat heart is not through Ras but via an unusual signaling cascade involving Raf and MEK1 (218). This study also suggests that PI3K and PKC may be responsible for phosphorylation and activation of Raf (41, 147) because both PI3K and PKC were shown to be targets of ONOO⁻ (10, 11, 90, 193, 319).

2. ROS

H₂O₂-mediated activation of ERK via Ras plays a key role in protecting cardiac myocytes from apoptotic death in cardiac myocytes prepared from ventricles of 1-day-old Wistar rats (4). This oxidative stress-induced ERK1/2 activation is reported in a variety of cell types. These sources are pleural mesothelial cells isolated from the parietal pleura of Fischer 344 rats (123), astrocytes obtained from a 2-day-old Sprague-Dawley rat brain (277), T cells (86), pulmonary epithelial cell cultures (30), chinese hamster lung fibroblast (V79) cells (143), ventricular myocytes isolated from hearts of adult male Sprague-Dawley rats (304), lung microvascular endothelial cells (280), and hepatocytes derived from rats (46).

In human umbilical vein endothelial cells, Ras has been shown to mediate the Raf-MEK1/2-ERK1/2 pathway (Fig. 15) leading to apoptosis after oxidative injury (50, 214). However, in PC12 cells, H₂O₂ activates ERK through Ras, and this activation has been shown to play a key role in preventing apoptosis (89, 313). One study shows that Angiotensin II (AII) induces hypertrophy in vascular smooth muscle cells (VMCs) derived from mice via ROS-mediated Ras activation that may contribute to atherosclerosis or restenosis (13). It also shows that ROS from NAD(P)H oxidase induces hypertrophy via activation of the Raf-MEK1/2-ERK1/2 pathway (Fig. 15) via Ras activation in ventricular myocytes derived from the hearts of adult rats (304).

As seen in the action of eNOS on H, N, and KRas (114, 174), the Ras Cys¹¹⁸ site can be chemically modified by ROS but in a different manner. Adachi et al. showed that the Ras Cys¹¹⁸ side chain in the NKCD motif was S-gluathionylated (Ras Cys-SSG) in cultured Rat VMCs poised with exogenous H₂O₂ or overexpressed with NAD(P)H oxidase (2). Intriguingly, this study also showed that AII-induced hypertrophy is associated with Ras activation by formation of Ras Cys-SSG. An overexpression of glutaredoxin-1 caused effective deglutathionylation of Ras that was coupled with inactivation of Ras and an attenuation of AII-induced hypertrophy (2). Also, overexpression of Trx1 in cells derived from adult rat ventricular myocytes alleviates hypertrophy that is mediated by α-adrenergic receptor-stimulated Ras activation (154) via the Raf-MEK1/2-ERK1/2 signaling pathways (Fig. 5) (303, 304). An in vivo study tested transgenic mice with cardiac-specific overexpression of Trx1 for development of hypertrophy (306). These mice did not develop hypertrophy. However, antisense inhibition of Trx1 results in hypertrophy in cultured cardiac myocytes via activation of the Ras-dependent Raf-MEK1/2-ERK1/2 cascade (306).

1. RNS

An in vivo study using a postnatal day-7-rat model of perinatal hypoxia-ischemia showed that Ras is activated in both the hippocampus and cortex within 2 h after hypoxia-ischemia (284). This study also shows that, although the Ras activation is NOS-dependent, its contribution to the pathophysiologic NO-dependent mechanisms of neurologic injury was minimal. Importantly, although expression of nNOS was detected in a subtype of the cortical interneurons of the brains of normal humans and those with Alzheimer's disease, nNOS was also detected in a subset of the pyramidal neurons of the patients with Alzheimer's disease, and this nNOS in the pyramidal neurons is highly colocalized with expression of Ras GTPase (180). One study has shown that NO-induced Ras-mediated MAPK activation is required for oxygen-glucose deprivation tolerance in cortical neurons (78). In addition, a study using human embryonic kidney 293 cells has shown that NO stimulates ATP-sensitive potassium (K_ATP) channels via targeting a redox-sensitive Ras Cys¹¹⁸ site that leads to activation of the MAPK pathway but not activation of the PI3K pathway (175). This NO-Ras-MAPK-K_ATP pathway in primary hippocampal cells prepared from an embryonic day-19 rat has been shown to be responsible for neuroprotection mediated by neuronal ischemic preconditioning (175). A recent study shows that the Ras-dependent Raf-MEK1/2-ERK1/2 pathway is involved in NO-induced neural protection in cortical neurons derived from the brains of 16-day-old Wistar rat embryos (247).

2. ROS

In PC cells, constitutively active Ras has been shown to enhance ROS production (76) that activates ERK via Ras (249), suggesting that the Ras-dependent Raf-MEK1/2-ERK1/2 pathway route is involved in the neuronal signaling response. Creatine may enhance survival signaling via activation of the Ras-NF-κB cascade in primary cerebrocortical cultures (132). This in turn may be coupled with the protective effects of creatine against glutamate cytotoxicity in neuronal cells and in animal models of neurodegenerative diseases.

2. Downregulation of Ras GTPases by redox agents

In some cases, endogenously released NO or an exogenous treatment of an NO donor inactivates Ras, which may link to certain diseases. In N293 cells (human embryonic kidney 293 transfected with nNOS), Ras was inactivated via S-nitrosation at the HRas Cys¹¹⁸ site in the NKCD motif by NO generated by nNOS; the Raf-MEK1/2-ERK1/2 cascade coupled with Ras was blocked (225, 226).

Treatment of an NO donor, SNAP or DETA/NO, inhibits activation of MAPK via inactivation of the Ras and Raf cascades in human pulmonary arterial smooth muscle cells (198). However, one study has reported that expression of iNOS results in upregulation of Ras that blocks VMCs proliferation via MAPK activation and operates independent of cGMP (139).

Intriguingly, in T cells, an NO donor, 2,2-(hydroxynitrosohydrazono)-bis-ethanamine, decreases expression of Ras; this decrease is linked to the role of NO as an inhibitor of lymphocyte cytotoxicity (67). This study also shows that NO neither activates nor inhibits the expressed Ras.

3. Modulation of Rap1A GTPase activity by redox agents

Unlike the Ras proteins, the redox regulation of Rap1A and its signaling consequences has not been studied extensively. RNS- or ROS-mediated inactivation of Rap1A has not been observed. However, several studies using human T-cell Jurkat cells have shown that Rap1A activity is sensitive to NO, because SNP and SNAP can promote Rap1A-GTP loading thereby leading to NO-induced activation of Rap1A (196, 236), which colocalizes with NAD(P)H oxidase (172). Moreover, expression of an activated Rap1A mutant in Wistar rat thyroid cells treated with NO-releasing agent SNP enhances sensitivity to apoptosis, whereas cells expressing wt Rap1A resisted NO-initiated cell death (236). Hence, it is possible that Rap1A activity also may be regulated by NAD(P)H oxidase-produced ROS. These in vivo results support the intriguing possibility that ROS-mediates regulation of the activity of Ras-related proteins.

1. Upregulation of Rho proteins by redox agents

Although a definite role for redox agents in direct cellular regulation of Rho proteins remains to be established, one study has shown that Caki-1 cells exposed to hypoxia exhibit increased Cdc42, Rac1, and RhoA protein expression and activity (278). This study also found that overexpression and activation of these Rho proteins was downstream of, and dependent on, the production of ROS because specific inhibition of ROS-producing Nox downregulated GTPases activity (Fig. 16). To date, however, unlike with the Ras and Rho proteins, neither an in-cell nor an in vivo study of the redox regulation of Rab GTPases has been made.

1. RNS

A body of evidence suggests that RhoA regulates via the ROCK pathway (Fig. 16). NO production from eNOS and that misregulation of this process results in the pathogenesis of several cardiovascular disorders, including hypertension and atherosclerosis. Studies show that RhoA inhibits eNOS expression (163), and inhibition of ROCK in human endothelial cells leads to the activation of the PI3K-Akt-eNOS pathway (Fig. 15) and cardiovascular protection (299). The Rho-ROCK pathway coupled with eNOS is linked significantly to the development of erectile dysfunction and the pelvic atherosclerosis of rats (213). A study using the rat intestinal cell line (IEC-6) shows that NO impairs mucosal healing by inhibiting enterocyte migration via activation of RhoA coupled with the SH2-containing protein tyrosine phosphatase (36). A recent study also shows that de novo nNOS expression was sufficient to induce, via the RhoA-ROCK pathway (Fig. 16), synaptic withdrawal in adult rat motoneurons (271). Isolated rat aortic rings (endothelium denuded) treated with ROS released from an xanthine oxidase and its substrate xanthine mixture were shown to induce Ca²⁺ sensitization via activation of the RhoA-ROCK signaling pathway but not via the Ca²⁺-independent PKC (124). Also, short-term air pollution increases hypertension in a rat artery via O₂ ^•−-mediated upregulation of the RhoA-ROCK pathway (270). One study also has shown that NO plays an antagonistic role against O₂ ^•− via Rac GTPase in smooth muscle cells of transgenic mice overexpressed with human cDNA of the constitutively active mutant of Rac1 (96). NO released from 2,2-(hydroxynitrosohydrazono)-bis-ethanamine consistently inhibits the activity of NAD(P)H oxidase in human VMCs via inhibition of Rac1 translocation (203). Conversely, the formation of O₂ ^•− by the NAD(P)H oxidase scavenges endothelial NO and thus is responsible for the development of renovascular hypertension and endothelial dysfunction (131).

2. ROS

In HeLa cells, the endogenously released ROS associated with activation of Rac further upregulates a signal transduction pathway associated with NF-κB (269). A pull-down assay has revealed that in retinal pigment epithelial cells, the small GTPase activated by H₂O₂ is Rac1 (110). The activation state of Rac1 and RhoA responds rapidly to changes in oxygen tension via PI3K and Nox-dependent signaling pathways. Their coordinated actions regulate the endothelial barrier function in the endothelial cells of conduit porcine pulmonary artery endothelial cells (298).

An in vivo study has shown that cardiomyocyte-specific Rac1, but not other Rac isoforms (Rac2 and Rac3), is critical to generating oxidative stress via interaction with gp91 ^phox and other cytosolic components of NAD(P)H oxidase (i.e., p47 ^phox ) that are linked to development of cardiac hypertrophy in an adult mice heart in response to vasoactive substances such as AII (240, 242, 274). The action of AII can be attenuated by statin treatment, resulting in downregulation of the Rac1-mediated NAD(P)H oxidase activation and reduction of ROS production in SMCs (288). Notably, ROS released from upregulation of NAD(P)H oxidase coupled with increased Rac1 activity often results in failure of the left ventricular myocardium of patients with ischemic cardiomyopathy or dilated cardiomyopathy, and oral statin treatment permits downregulation of Rac1 activity in the human heart (181). Intriguingly, a study using the Cos7 cell line shows that Cdc42 can serve as a competitor of Rac for cytochrome b ₅₅₈ (61).

2. Downregulation of Rho protein by redox agents

Most research results suggest that a redox agent activates Rho proteins by an increase in Rho GNE. However, an opposite effect of RNS on regulation of the activity of RhoA proteins has been reported (256). In that report, iNOS was not only involved in the increased expression of RhoA but also in prevention of the upregulation of RhoA in diabetic rat hearts. Although indirect, an explanation has been proposed for the mechanism in the ROS-dependent downregulation of RhoA (207): An activation of Rac GTPase stimulates release of O₂ ^•− from NAD(P)H oxidase, which in turn inhibits the low-molecular-weight protein tyrosine phosphatase (LMW-PTP). Because p190Rho-GTPase-activating protein (p190Rho-GAP) is a substrate of LMW-PTP, inactivation of LMW-PTP results in accumulation of the active phosphorylated form of p190Rho-GAP. This activated p190Rho-GAP stimulates the hydrolysis of bound GTP to produce inactive GDP-bound RhoA. This downregulated RhoA contributes to the spread of integrin-dependent cells.

1. Upregulation of Ran and Dexras1 by redox agents

A body of physiological study results on the cellular redox stress associated with Ran has been reported (92, 148, 170, 197, 224, 307, 310). A H₂O₂-induced oxidative stress-induced perturbation of the Ran nucleo-cytoplasmic concentration gradient has been reported (148, 310). In HeLa cells, the oxidant H₂O₂ inhibits classical nuclear import, including perturbation of the Ran nucleo-cytoplasmic gradient (148). The Ran gradient in HeLa cells dissipates in response to a stress-induced depletion of GTP-bound Ran, diminishing the efficiency of Ran nuclear import (148, 310). The study also showed that the H₂O₂-mediated oxidative stress induces a relocation of the nucleoporin (Nup153) as well as the nuclear carrier importin-β. The result is to reduce docking of the importin-α/β/cargo complex at the nuclear envelope. Moreover, the H₂O₂-induced oxidative stress causes Ran, importin-β, and Nup153 to undergo a caspase/proteasome-dependent (but ubiquitin-independent) proteolysis (148). The nuclear accumulation of importin-α appears to be triggered by a collapse in the Ran gradient, causing suppression of the nuclear export of importin-α (197).

Dexras1, which also possesses the CGNKXD motif, is mainly involved in redox regulation of neural cells (173, 252). However, unlike with Dexras1, no in vivo or cell studies have been reported on redox regulation of the activity of the Rhes protein.

1. RNS

The pathophysiological effect of misregulation of RNS-mediated Ran signaling has not been investigated.

2. ROS

Although it is unclear whether the effect of ROS on Ran is direct or indirect, one study showed that Ran/TC4 was upregulated by H₂O₂ in nonmalignant breast epithelial cells but not in malignant cells (307). Further, a proteomic approach has identified in plants a thioredoxin-dependent activation of the Ran/TC4 protein (170).

1. RNS

Dexras1 forms a ternary complex with the NO-producing nNOS as well as the carboxy-terminal PDZ ligand of nNOS in rat brains (63). Formation of this complex enhances the ability of nNOS to activate Dexras1 in rat brains. In a separate study, colocalization of Dexras1 with carboxy-terminal PDZ ligand of nNOS and nNOS also was observed in the neurons and glial cells of rats (173). In vitro and cell studies also show that the Dexras1 Cys¹¹ site can be S-nitrosated (120). A subsequent study using HEK293T and PC12 cells has shown that stimulation of the NMDA receptors activates nNOS, leading to S-nitrosation at the Dexras1 Cys¹¹ site and activation of Dexras1, which is linked to NMDA-mediated neurotoxicity (37).

2. ROS

The pathophysiological relevance of misregulation of Dexras1 associated with ROS has not been investigated.

2. Nonredox response of Ran under certain conditions

As discussed in Section III, these redox-sensitive GTPases such as Ras and Rho proteins sometimes can be downregulated by RNS and/or ROS or they can be insensitive to one or the other or to both. A recent study shows that a mild oxidative stress (conditions that do not induce death in the majority of cells as defined by the author) induced by diethyl maleate has, without exception, little effect on the nucleocytoplasmic concentration gradient of Ran (149).