A Molecular Redox Switch on p21 ras STRUCTURAL BASIS FOR THE NITRIC OXIDE-p21 ras INTERACTION*

We have identified the site of molecular interaction between nitric oxide (NO) and p21 ras responsible for initiation of signal transduction. We found that p21 ras was singly S -nitrosylated and localized this modifica- tion to a fragment of p21 ras containing Cys 118 . A mutant form of p21 ras , in which Cys 118 was changed to a serine residue and termed p21 ras C118S, was not S -nitrosylated. NO-related species stimulated guanine nucleotide ex- change on wild-type p21 ras , resulting in an active form, but not on p21 ras C118S. Furthermore, in contrast to parental Jurkat T cells, NO-related species did not stimu- late mitogen-activated protein kinase activity in cells transfected with p21 ras C118S. These data indicate that Cys 118 is a critical site of redox regulation of p21 ras , and S -nitrosylation of this residue triggers guanine nucleo- tide exchange and downstream signaling.

It is well known that signal transduction pathways initiated by extracellular ligands are dependent on protein-protein interactions for propagation and amplification of their signal. Many of these interactions lead to phosphorylation events. For example, receptor-tyrosine kinases require a series of protein interactions utilizing SH2 and SH3 domains of adaptor proteins to generate an activated form of p21 ras , a critical signaling enzyme (1)(2)(3).
Recent studies have identified reactive free radicals as central participants in certain signaling events (4 -7). Enhancing free radical destruction, either enzymatically or chemically, prevented ligand-stimulated transcription factor (8) and mitogen-activated protein (MAP) 1 kinase (4) activation and also prevented smooth muscle cell mitogenesis and chemotaxis (4). Thus, a role is emerging for reactive free radicals in mediating signal transduction.
Among the many recently discovered functions of NO, a role in signaling has surfaced (9). Although soluble guanylyl cyclase is an important target of NO in mediating some of its physiologic functions such as the regulation of blood pressure (10,11), other signaling events, some culminating in transcriptional activation, may be cGMP-independent (12)(13)(14)(15). Our studies have focused on how NO initiates cGMP-independent signaling within cells (16,17). We have identified p21 ras as a critical target of NO and other redox modulators (17)(18)(19). Here, we sought an understanding of the structural basis of the NO-p21 ras interaction in the hope of gaining insight into how redox signaling is achieved.
Generation of p21 ras C118S cDNA Constructs-Codon 118 of truncated (codons 1-166) Ha-ras cDNA was mutated from TGT (cysteine) to TCT (serine) using the polymerase chain reaction. The generated cDNA fragment was then sequenced (Sequenase) and cloned into the pATras bacterial expression vector. To generate full-length p21 ras C118S an NcoI/BamHI fragment (encoding residues 111-166 of the ras(1-166) mutant) was exchanged for a 0.8-kilobase fragment encoding residues 111-189 plus 3Ј-noncoding region and the coding junction sequenced. A BglII/BamHI fragment encoding full-length Ras(C118S) was then subcloned into the BamHI site of the pCDNA3 mammalian expression plasmid and orientation confirmed by BstXI digestion.
CNBr Digestion and ESI-MS Analysis of p21 ras -One small crystal of CNBr (Fluka) was added to 100 pmol of p21 ras in 20 l of 0.1 N HCl in a 0.5-ml polypropylene tube. Digestion was carried out at room temperature for 10 min prior to analysis by ESI-MS. After 10 min, samples were directly electrosprayed into a Finnigan-MAT TSQ-700 triple quadrupole instrument for analysis of S-nitrosylation exactly as we described previously (21).
Preparation of NO Solutions-NO solutions were prepared as we described previously (17). Briefly, a solution of 20 mM ammonium bicarbonate solution, pH 8.0, in a rubber-stoppered tube was sparged for 15 min with N 2 and then 15 min with NO gas (Matheson Gas, East Rutherford, NJ). This resulted in a saturated solution of NO (1.25 mM). This solution also contained higher oxides of NO which were not quantified.
Cell Transfection and Culture-The human T cell line Jurkat was grown in RPMI 1640 containing 2% L-glutamine and 10% fetal calf serum. For transfection cells were washed in serum-free medium and resuspended to 1-5 ϫ 10 6 cells/ml in serum-free medium. Liposomes (50 l of Lipofectin, Life Technologies, Inc.) were mixed with a solution of 10 g of p21 ras C118S plasmid in 150 l of sterile water. This was then added to 1 ml of culture and placed into culture flasks. After incubation for 24 h at 37°C and 5% CO 2 , the cells were supplemented with 3 ml of selection medium (RPMI 1640 containing 10% fetal calf serum, 2% L-glutamine, and 1 mg/ml G418 (Life Technologies, Inc.)). After another 24 h, cells were resuspended and maintained in the selection medium. Mock transfected cells did not receive any DNA. After 3-4 weeks in selection medium, transfected cells were analyzed for p21 ras levels via Western blotting.  1 The abbreviations used are: MAP, mitogen-activated protein; NO, nitric oxide; ESI-MS, electrospray ionization-mass spectrometry; ERK, extracellular signal-regulated kinase; PDGF, platelet-derived growth factor.
MAP Kinase Activity-MAP kinase activity was measured in an in vitro kinase assay using myelin basic protein as a substrate, as we described previously (18). Briefly, serum-starved cells (24 h, 5 ϫ 10 6 ) were treated for 30 min at 37°C, pelleted, and then resuspended in 300 l of RIPA buffer containing 20 mM Tris, pH 7.4, 137 mM NaCl, 2 mM EDTA, 1% Triton X-100, 25 mM ␤-glycerophosphate, 1 mM NaVO 3 , 2 mM NaPP i , 20% glycerol, 1 mM phenylmethylsulfonyl fluoride, and 10 g/ml leupeptin. Samples were vortexed, left on ice for 15 min, and then microcentrifuged for 2 min. Protein A-Sepharose prebound to anti-ERK1 or ERK2 (Santa Cruz Biotechnology) was added to supernatants (5 g/sample). After 1 h at 4°C, samples were washed twice with RIPA buffer and twice with kinase buffer (25 mM Hepes, pH 7.4, 25 mM ␤-glcerophosphate, 25 mM MgCl 2 , 2 mM dithiothreitol, and 0.1 mM NaVO 3 ). After the final wash, samples were resuspended in 20 l of kinase buffer, and 1 g of myelin basic protein was added along with 22 l of 10 Ci/nmol [␥-32 P]ATP. After 20 min at 30°C, 4 l of 6 ϫ Laemmli sample buffer containing 100 mM dithiothreitol was added, and samples were boiled for 2 min. Samples were run on 15% sodium dodecyl sulfate-polyacrylamide gels and were analyzed via a Phosphor-Imager (Molecular Dynamics).

Localization of the Site of S-Nitrosylation on p21 ras -Our
earlier studies correlated a single S-nitrosylation event on fulllength p21 ras with enhanced guanine nucleotide exchange (17). In the present in vitro studies, p21 ras lacking the carboxylterminal 23 amino acids is used. This form of p21 ras is commonly used for in vitro studies and possesses biochemical activity identical to that of the wild-type enzyme (22). To identify the exact site of S-nitrosylation, we took advantage of the fact that cleavage of p21 ras at Met residues with CNBr yields three major fragments each containing a single Cys residue (Fig. 1). We monitored each of the Cys residues for S-nitrosylation by subjecting p21 ras to CNBr digestion followed by analysis using ESI-MS. In this ESI-MS assay, the molecular mass of samples treated with NO is compared with that of untreated samples. An increase in mass of 29 Ϯ 1 Da (the mass of NO, 30 Da, minus the mass of the substituted proton) and its lability to increased energy input are indicative of S-nitrosylation (21). As seen in Table I, CNBr digestion of p21 ras yielded a fragment with a molecular mass of 6,223 Ϯ 2 Da, corresponding to Fragment 3 ( Fig. 1). Upon treatment of p21 ras with NO and subsequent cleavage with CNBr, Fragment 3 had a new mass clearly indicative of S-nitrosylation (Table I); that is, the mass of Fragment 3 from NO-treated p21 ras was equal to that of the unmodified Fragment 3 (6,223 Da) plus that of NO (30 Da). This fragment contains Cys 118 , and thus this Cys residue is the likely target of NO. It is possible that Fragments 1 and 2 ( Fig.  1) were not observed because of inter-and intramolecular hydrophobic interactions that precluded their solubilization.
To confirm that Cys 118 was indeed the site of S-nitrosylation, we generated a form of p21 ras identical to the wild-type enzyme, except that Cys 118 was modified to a Ser residue (referred to as p21 ras C118S). This modification only changes the sulfur atom of Cys 118 to oxygen, thus reducing the mass of the enzyme by 16 Da to 18,836 Da. We treated wild-type p21 ras with NO under conditions in which we achieved approximately 50% S-nitrosylation. Analysis by ESI-MS revealed the parent enzyme (mass ϭ 18,852 Da) and a singly S-nitrosylated derivative (mass ϭ 18,882 Da, Table I). Treatment of p21 ras C118S with NO under identical conditions resulted in no S-nitrosylated product but only the parent enzyme (Table I). These data identify Cys 118 as the molecular target of NO on p21 ras .
Cys 118 as the Molecular Trigger for NO Signaling-Having established that Cys 118 is the site of interaction of NO on p21 ras , we examined whether this S-nitrosylation was responsible for NO-induced p21 ras activation and subsequent downstream signaling events. We have found previously that NO and other free radicals induce guanine nucleotide exchange on p21 ras in cells and in vitro (17)(18)(19). Therefore, we examined whether NO could induce nucleotide exchange on GDP-preloaded p21 ras C118S in vitro. Direct measurement of exchange relies on a filter binding assay to which our p21 ras (1-166) protein does not bind quantitatively (22,23). Therefore, to measure exchange, GDP-preloaded p21 ras is treated with NO in the presence of [␥-32 P]GTP, and hydrolyzed 32 P i is quantified. We have previously shown this to be a measure of exchange in our system (17). As seen in Fig. 2, NO potently stimulated [␥-32 P]GTP hydrolysis of GDP-preloaded wild-type p21 ras (open circles). In contrast, NO had almost no effect on the exchange rate of p21 ras C118S (Fig. 2, closed circles). The basal rates of hydrolysis for the two enzymes were similar (24.6 Ϯ 6 fmol of PO 4 Ϫ released/min/mg for the wild-type enzyme and 18.4 Ϯ 5 fmol/min/mg for p21 ras C118S). A clear biphasic curve of activation and subsequent inhibition by higher concentrations of NO was seen. We and others have previously seen this type of biphasic behavior of NO in many systems (15,17,24). The inhibitory component may be due to nonspecific noxious effects of high concentrations of NO and its higher oxides or due to quenching (25). These data indicate that interaction of NO with Cys 118 is required for NO-induced guanine nucleotide exchange on p21 ras .
Since NO and other redox modulators can stimulate several biochemical events downstream of p21 ras , such as nuclear factor B translocation and MAP kinase activity (15, 18, 26), we  examined whether Cys 118 on p21 ras is a target for NO in cells. Using Jurkat T cells mock transfected or stably transfected with an expression plasmid encoding a full-length version of p21 ras C118S (i.e. residues 1-189), we examined the ability of NO to activate MAP kinase activity. Immunoprecipitation of the MAP kinases ERK1 and ERK2 from wild-type cells treated with NO-generating compounds S-nitroso-N-acetylpenicillamine or sodium nitroprusside resulted in enhanced phosphorylation of the ERK substrate (Fig. 3A, open bars). In contrast, Jurkat T cells stably expressing p21 ras C118S(1-189) did not respond to NO in this in vitro kinase assay (Fig. 3A, hatched  bars). These transfected cells expressed 7-10-fold more p21 ras than the wild-type cells as determined by Western blotting with anti-p21 ras antibody, Y13-259 (Fig. 3B). This antibody cannot distinguish between wild-type and p21 ras C118S, suggesting that although endogenous p21 ras was not specifically inhibited, ectopic expression of high levels of mutant p21 ras apparently prevented its signaling. This dominant negative activity of p21 ras C118S toward NO action may be due to its high level of expression. The pool of effectors available for wild-type p21 ras to interact with may be reduced greatly by overexpression of p21 ras C118S, perhaps due to their sequestration. Another possibility is that MAP kinase activity is suppressed in the mutant cells. To test this, we treated parental and transfected cells with phorbol myristate acetate (100 ng/ ml) and the calcium ionophore A23187 (500 ng/ml) for 5 min. We found that these agents, which bypass p21 ras in activating MAP kinase, stimulated MAP kinase activity in both cell types (i.e. 221 Ϯ 8 versus 261 Ϯ 9% of control, parental versus transfected). Thus, NO donors did not stimulate MAP kinase activity in p21 ras C118S-transfected cells although these cell harbored a functional MAP kinase system. These data indicate that Cys 118 is indeed the target of NO on p21 ras responsible for triggering downstream signal transduction.

DISCUSSION
Redox regulation of signaling pathways currently presents several conceptual riddles. These include identifying the source of regulatory redox species, maintaining specificity, and identifying the redox target. Our earlier work demonstrated a functional requirement for p21 ras in NO signaling (17). Here, we have focused on identifying the molecular target of NO on p21 ras . We have identified Cys 118 as the critical site of Snitrosylation. The crystal structure of p21 ras is well defined (27,28), and modeling studies show that Cys 118 is the most surface accessible of the three Cys residues in our p21 ras preparation. Using the program X-PLOR (Molecular Simulations, Inc.) the solvent accessible surface of p21 ras complexed to GDP (coordinates obtained from the NMR solution structure) was calculated using a 1.4 Å radius probe. Whereas the solvent accessibility of Cys 51 and Cys 80 side chains were similar and fairly shielded from the solvent, Cys 118 was solvent exposed. The buried nature of residues Cys 80 and Cys 51 provides a structural basis of why a single S-nitrosylation occurs on p21 ras upon exposure to NO.
A mechanistic understanding of how S-nitrosylated Cys 118 leads to enhanced guanine nucleotide exchange will likely be provided by solving its x-ray crystal structure. However, some insight is gained from considering what is currently known about p21 ras structure and function. Cys 118 is located within a highly conserved region (NKXD) of the ras superfamily and indeed all GTP-binding protein sequences. This NKXD motif, FIG. 2. Effect of NO on GTPase activity of p21 ras and p21 ras C118S. GDP-preloaded wild-type (wt) p21 ras or p21 ras C118S was exposed to the indicated concentrations of NO for 10 min prior to assay of [␥-32 ]GTP hydrolysis as described under "Materials and Methods." Basal rates for the two enzymes were 24.6 Ϯ 6 fmol of PO 4 Ϫ released/ min/mg for the wild-type enzyme and 18.4 Ϯ 5 fmol/min/mg for p21 ras C118S. Data represent mean Ϯ S.D. of three experiments. in which Cys 118 is the variable X residue in p21 ras , interacts directly with the guanine nucleotide ring of GTP and GDP and with other nucleotide-binding loops of the protein (29). Independent mutation of residue 116, 117, or 119 leads to an increased dissociation rate of bound nucleotide, resulting in an increased rate of nucleotide exchange (30). Although our p21 ras C118S mutant had basal rates of guanine nucleotide exchange similar to those of the wild-type protein, it is possible that S-nitrosylation results in an alteration in protein-GDP contact, resulting in nucleotide exchange.
The presence of a redox-active residue in such a critical domain suggests that its conservation may reveal enzymes and transductional systems that may be similarly regulated. In the ras superfamily, Ha-, Ki-, and N-ras, rap1A, rap1B, rab1, and rab3 contain a Cys residue in this conserved region. In contrast, ral, tc21, R-ras, rap2, and rho gene products do not. Within the ras subfamily, this Cys residue is conserved from slime mold to man. Such conservation suggests that this molecular redox trigger is an important mechanism by which cells respond to reactive free radicals.
It is likely that this molecular switch is regulated by cellular antioxidant levels. We have shown previously that reducing cellular glutathione levels renders the p21 ras signaling pathway dramatically more sensitive to redox activation (18). According to this model, as glutathione is oxidized, the Cys switch on p21 ras becomes available for modification, and a preprogrammed signaling cascade is initiated. Glutathione or other low molecular weight thiols likely play an important role in S-nitrosylation of p21 ras . NO, under anaerobic conditions, cannot participate in S-nitrosylation. NO must be converted to a redox active form such as nitrosonium ion (NO ϩ ; Ref. 9). This occurs rapidly in the presence of metals or thiols, and thus S-nitrosoglutathione may be a crucial reservoir of redox active NO.
Such a redox-triggered signaling pathway may also provide a mechanistic understanding of some recent observations identifying a requirement for reactive free radicals in mediating platelet-derived growth factor (PDGF) signaling. In those studies (4), PDGF stimulated H 2 O 2 production in vascular smooth muscle cells. When H 2 O 2 production was blocked, PDGF-induced enhancement of MAP kinase activity, chemotaxis, and DNA synthesis was prevented. Many of these PDGF-dependent events require p21 ras , and thus the redox-sensitive target, Cys 118 , is likely to be involved in this reactive free radical-dependent signaling cascade. Modification of Cys by redox mod-ulators other than NO, such as hydroxyl radical or heavy metals, would necessitate a Cys modification that is chemically different from the S-nitrosylation described herein. However, the structural alterations may ultimately be the same.