Nitric oxide modification of rat brain neurogranin. Identification of the cysteine residues involved in intramolecular disulfide bridge formation using site-directed mutagenesis.

Neurogranin (Ng) is a neuron-specific protein kinase C-selective substrate, which binds calmodulin (CaM) in the dephosphorylated form at low levels of Ca2+. This protein contains redox active Cys residues that are readily oxidized by several nitric oxide (NO) donors and other oxidants to form intramolecular disulfide. Identification of the Cys residues of rat brain Ng, Cys3, Cys4, Cys9, and Cys51, involved in NO-mediated intramolecular disulfide bridge formation was examined by site-directed mutagenesis. Mutation of all four Cys residues or single mutation of Cys51 blocked the oxidant-mediated intramolecular disulfide formation as monitored by the downward mobility shift under nonreducing SDS-polyacrylamide gel electrophoresis. Single mutation of Cys3, Cys4, or Cys9 or double mutation of any pair of these three Cys residues did not block such intramolecular disulfide formation, although the rates of oxidation of these mutant proteins were different. Thus, Cys51 is an essential pairing partner in NO-mediated intramolecular disulfide formation in Ng. Cys3, Cys4, and Cys9 individually could pair with Cys51, and the order of reactivity was Cys9 > Cys4 > Cys3, suggesting that Cys9 and Cys51 form the preferential disulfide bridge. In all cases tested, the intramolecularly disulfide bridged Ng proteins displayed dramatically attenuated CaM-binding affinity and ∼2-3-fold weaker protein kinase C substrate phosphorylation activity. The data indicate that the N-terminal Cys3, Cys4, and Cys9 are in close proximity to the C-terminal Cys51 in solution. The disulfide bridge between the N- and C-terminal domains of Ng renders the central CaM-binding and phosphorylation site domain in a fixed conformation unfavorable for binding to CaM and as a substrate of protein kinase C.

Nitric oxide (NO) 1 has been established as a messenger molecule in physiological processes as diverse as host defense, vascular regulation, and neuronal communication (1)(2)(3)(4). NO is synthesized by a family of nitric oxide synthases (NOS) which utilize arginine as their substrate in the 5-electron oxidation of the guanidino nitrogen. NO is a free radical gas that readily diffuses into cells where it reacts with molecular targets. NO is extremely susceptible to both oxidation and reduction resulting in the formation of NO surrogates such as nitrosonium (NO ϩ ) and nitroxyl anion (NO Ϫ ) (5). In addition, NO free radical reacts readily with other free radical such as superoxide anion (O 2 . ) to form peroxynitrite (ONOO Ϫ ) (6), with O 2 to form NO 2 , and with transition metal ions to form adducts (5). These secondary reaction products and products of NO oxidation and reduction are capable of reaction with metals, thiols, and additional targets to give further products with biological activities (5,7). The signal transduction pathways of NO can be broadly classified as cGMP-dependent and -independent; the former pathway involves NO binding at the heme of soluble guanylyl cyclase leading to stimulation of cGMP formation and the latter pathway involves reactions with other heme and nonheme iron, N-nitrosation of nucleic acids, and modifications of target proteins by S-nitrosation, ADP-ribosylation, and tyrosine nitration (2,3). S-Nitrosation of thiols is a common occurrence in biological systems; RS-NO serves as a bioactive reservoir of NO that targets reactive sulfhydryl groups in Snitrosothiol-thiol exchange reactions and accelerates intramolecular disulfide formation from vicinal thiol groups (8). Reversible formation of disulfide bonds is utilized biologically in enzyme catalysis, transport of reducing equivalents, metabolic regulation, cellular defense, and provision of structural stability (9). Rat brain neurogranin (Ng), also known as RC3 or BICKS, is a 78-amino acid PKC-selective substrate which binds calmodulin (CaM) at low levels of Ca 2ϩ (10 -15). This protein and another PKC substrate neuromodulin (also known as GAP-43, F1, or B-50) contain a conserved 19-amino acid sequence region, where the CaM-binding and phosphorylation site domain is located (11,16,17). Phosphorylation of these two proteins by PKC weakens their binding affinities for CaM. It has been proposed that the PKC-catalyzed phosphorylation of these two proteins frees the CaM for other CaM-dependent enzymes (18,19). The signal transduction pathway involving activation of PKC and phosphorylation of Ng and neuromodulin has been linked to the modulation of ion channel and neurotransmitter release and the induction and maintenance of long term potentiation (20 -25). Recently, neuromodulin has been shown to be a target of nitric oxide, which inhibits thioester-linked longchain fatty acylation possibly through a direct modification of Cys thiols (26). Both Ng and neuromodulin contain two Nterminal Cys residues at position 3 and 4; the former contains two additional Cys residues at position 9 and 51. So far, there is no evidence for the long-chain fatty acylation of any of the Ng Cys residues (27). Treatment of rat brain Ng with NO donors or other oxidants, such as H 2 O 2 or o-iodosobenzoic acid (IBZ), results in intramolecular disulfide(s) formation with resulting attenuation of this protein's binding to CaM and the ability to serve as a substrate of PKC (28). Oxidation of Ng was also seen in the NO donor-treated rat brain synaptosomes when analyzed by immunoblot with an antibody against Ng. Here, we examine the role of the four Cys residues of rat brain Ng, Cys 3 , Cys 4 , Cys 9 , and Cys 51 , in NO modification of its function by site-directed mutagenesis of those residues. Cys 51 , close to the centrally located PKC phosphorylation site and the CaM-binding domain, is a critical pairing partner for intramolecular disulfide bridge formation. Cys 3 , Cys 4 , and Cys 9 individually can form a disulfide bridge with Cys 51 and the order of reactivity of the former three Cys residues with Cys 51 is Cys 9 Ͼ Cys 4 Ͼ Cys 3 , indicating that Cys 9 -Cys 51 is the preferential intramolecular disulfide bridge formed by NO modification.

EXPERIMENTAL PROCEDURES
Materials-Commercial products utilized were: [␥-32 P]ATP (28 Ci/ mmol) from DuPont NEN; ␣-35 S-dATP (1,000 Ci/mmol) from Amersham Corp.; oligodeoxynucleotides from Curachem; PCR reagent system and 100-base pair ladder from Life Technologies, Inc.; Wizard miniprep kits, competent HB101 and JM109 cells from Promega; pRSET expression plasmid and TA cloning kit from Invitrogen; competent BL21(DE3) expression cells from Novagen; all restriction endonucleases, T4 ligase, and SP6 promoter and T7 universal primers from New England Biolabs; Sequenase 2.0 kit from U. S. Biochemical Corp.-Amersham; CaM and CaM-Sepharose from Pharmacia Biotech Inc.; brain phosphatidylserine (PS), 1,2-dioleoylglycerol (DG) from Avanti Polar Lipids; P81 phosphocellulose paper from Whatman; diethylamine nitric oxide (DEANO) from Molecular Probes; 3-morpholinosydnonimine (SIN-1), S-nitrosoglutathione (SNOG), S-nitroso-N-acetylpenicillamine (SNAP) and Reductacryl (0.202 meq/g) from Calbiochem; sodium nitroprusside (SNP) and IBZ from Sigma; Dowex AG 1-X8 from Bio-Rad; ultrafiltration YM3 membrane from Amicon; and C4 HPLC column from Vydac. All other chemicals were reagent grade or of higher purity. The following oligodeoxynucleotide primers were utilized in the PCR reactions: Preparation of WT and Cys Mutant Ng Constructs-The 4 Cys residues of rat brain Ng, Cys 3 , Cys 4 , Cys 9 , and Cys 51 , were mutated at single, double, or tetra positions ( Table I). The strategy used for generating the PCR site-directed mutants consisted of using primers corresponding to the start codon region of Ng with an upstream NdeI site linker and contained single base substitutions for the codon of Cys 3 , Cys 4 , and Cys 9 . The 3Ј primer was complementary to the stop codon region of Ng and contained a downstream BamHI site linker. Fulllength WT or mutant PCR products were then directly T4 ligated to TA cloning vector (pCRII). Subcloning into the pRSET expression vector was carried out by double cutting the TA-cloning plasmid with NdeI and BamHI and the purified insert was T4 ligated with the double cut (NdeI and BamHI) and purified pRSET vector. Alternatively, the full-length WT or mutant PCR products were double cut with NdeI and BamHI, purified by agarose gel electrophoresis, and T4 ligated with double cut (NdeI and BamHI) and similarly purified pRSET expression vector. In the case of C51G-containing mutants (Tetra and C51G) a two-step procedure was used; a 3Ј primer complementary to the Cys 51 region containing a single base substitution (Cys 51 3 Gly) and a downstream EaeI site linker was used with either the 5Ј WT or mutant primers to generate PCR products corresponding to the N-terminal half of the protein (residues 1-51). These C51G containing N-terminal half PCR products were cut with EaeI, gel-purified, and T4 ligated with the EaeI-digested and gel-purified C-terminal half of the WT PCR product. The purified full-length C51G-containing constructs were double cut with NdeI and BamHI, purified, and T4 ligated with double cut (NdeI and BamHI) and gel-purified pRSET expression vector. The following primer pairs were used for the various constructs: WT, P1 and P2; C3S, C3S, and P2; C4G, C4G, and P2; C9S, C9S, and P2; Tetra, C3S/C4G/ C9S, and C51G; C51G, P1, and C51G; C3S/C4S, C3S/C4S, and P2; C3S/C9S, C3S/C9S, and P2; C4S/C9S, C4S/C9S, and P2. PCR reactions were carried out in the presence of 200 M dNTPs, 1 mM MgCl 2 , 10 ng of Ng cDNA template, 2.5 g each of 5Ј and 3Ј primers, 20 mM Tris-Cl (pH 8.4), and 50 mM KCl in a final volume of 100 l. Reactions were started by denaturing at 94°C (3 min) and then adding Taq Polymerase (2.5 units) at 80°C. Thermal cycling was carried out in a Perkin-Elmer 480 instrument with a program consisting of 94°C (1 min), 55°C (2 min), and 72°C (1 min) for 30 cycles with a final extension of 72°C (5 min). The PCR products were extracted successively with 1 volume of phenol/chloroform/isoamyl alcohol (25:24:1) and 1 volume of chloroform, and the products in the aqueous phase were precipitated in the presence of 0.3 M sodium acetate (pH 5.3) with 3 volumes of cold ethanol (4°C). After incubation on dry ice (15 min), the PCR products were pelleted (14,000 ϫ g, 15 min, 4°C), washed with cold 70% ethanol (1 ml), repelleted, and dried under a vacuum on a SpeedVac. Dried pellets were resuspended in 10 mM Tris-Cl (pH 8.0), 1 mM EDTA (TE). The size and yield of the PCR products was checked by analyzing an aliquot on 2% agarose gel in the presence of TAE (40 mM Tris acetate, pH 8.0, 1 mM EDTA) and 0.5 g/ml ethidium bromide. All PCR products and constructs were recovered from agarose gels by electrophoretic transfer to DEAE-cellulose paper strips, eluted with high salt (1 M NaCl), and ethanol precipitation (29). The T4-ligated full-length constructs (1-10 ng DNA) were used to transform competent JM109 cells by the heat shock method (Promega protocol). Colonies were screened for the presence of the insert following digestion with NdeI and BamHI. Plasmid DNA was sequenced to confirm the proper coding sequence and the site-directed mutations. Plasmid DNA was then transformed into BL21(DE3) expression cells by the heat shock method. Expression of WT and mutant Ng proteins was checked by pelleting (14,000 ϫ g, 5 min) a 1-ml aliquot of the induced culture and resuspending the pellet with SDS sample buffer containing ␤-mercaptoethanol (143 mM); analysis was carried out on 10 -20% gradient SDS-polyacrylamide gels, and pUC19-transformed BL21(DE3) cells were used as a negative control for protein expression.

Expression and Purification of WT and Cys Mutant Ng Proteins from BL21(DE3) cells-WT and mutant
Ng proteins were expressed and purified from single colonies of BL21(DE3) that had been grown overnight (500 ml ϫ 6, 37°C). Cell pellets were resuspended in 5 volumes of H buffer (20 mM Tris-Cl, pH 8.0, 1 mM EGTA, 1 mM EDTA, 1 mM DTT, 0.5 mM phenylmethylsulfonyl fluoride). Lysozyme (1 mg/g of cells) was added to the resuspended cells, and lysis of the cell walls was at 20°C for 15 min. The viscous solution was freeze-thawed (dry ice/37°C) for two to three cycles to completely break open the cells and was sonicated (4°C, three times for 1 min each). Cell debris was pelleted (30,000 ϫ g, 30 min), and the supernatant was removed and was made 2.2% HClO 4 by dropwise addition of 60% HClO 4 with stirring. The majority of proteins were precipitated by this treatment and were pelleted (30,000 ϫ g, 30 min). The supernatant was neutralized with dropwise addition of 10 N KOH, and the resulting KClO 4 was allowed to precipitate (4°C, 10 min). The KClO 4 precipitate was removed by centrifugation (30,000 ϫ g, 20 min), and the supernatant was concentrated to 1-2 ml by ultrafiltration using YM3 membrane (Amicon). The concentrate was made 0.1 M in trifluoroacetic acid and applied to a Vydac C4 HPLC column (1 ϫ 25 cm). Elution solvents were: A, 0.1% trifluoroacetic acid; B, acetonitrile containing 0.1% trifluoroacetic acid; the gradient program consisted of 0% B (5 min) followed by 0 -60% B (55 min) with a flow rate of 2 ml/min; detection was at 214 nm, and 1-ml fractions were collected. Ng and mutant proteins eluted at ϳ40% B. Fractions were  4 3 Gly, Cys 9 3 Ser, Cys 51 3 Gly frozen and lyophilized, and the dried fractions were resuspended in 100 l of distilled water. An aliquot of each fraction was analyzed by 10 -20% SDS-PAGE, and those fractions containing Ͼ95% Ng or mutant proteins were pooled. Ng was additionally purified from frozen rat brain as described previously (15). PKC, PKM, and Protein Quantitation Assays-Rat brain PKC and PKM (30) were purified to near homogeneity by the previously described methods. PKC activity was determined using a lipid vesicle assay containing 30 mM Tris-Cl (pH 7.5), 6 mM Mg-acetate, 60 M [␥-32 P]ATP (1,000 -3,000 cpm/pmol), 100 g/ml PS, 20 g/ml DG, variable substrate, 400 M CaCl 2 , and 0.5-1.0 g/ml PKC, unless otherwise noted. PKM activity was determined under similar conditions in the presence of 2 mM EGTA but without PS, DG, and CaCl 2 . Reactions were carried out at 37°C in a final volume of 25 l and 32 P i incorporation was determined using P81 phosphocellulose paper (31) or a Dowex 1/DEAE cellulose mini column (15). Stocks of phospholipid/DG were prepared by drying mixtures in CHCl 3 under a N 2 stream; lipids were resuspended in 20 mM Tris-Cl (pH 7.5), vortexed, and sonicated (2 min, bath sonicator). The concentration of WT and mutant Ng proteins was estimated by either quantitative phosphorylation (37°C, 1 h) or by amino acid analysis.

RESULTS
The four Cys residues, Cys 3 , Cys 4 , Cys 9 , Cys 51 , of rat brain Ng were examined for their role in NO modification of this protein by site-directed mutagenesis. WT, four single, C3S, C4G, C9S, C51G, three double, C3S/C4S, C3S/C9S, C4S/C9S, and a Tetra Cys mutant Ng proteins were constructed by recombinant DNA methods and expressed in and purified from Escherichia coli. The purified WT and eight Cys mutant Ng proteins were over 90% pure as judged by SDS-PAGE, and all were phosphorylated by PKC (Fig. 1). Rat brain Ng, which contains 78 amino acids and has a 7.5-kDa molecular mass based on the amino acid sequence, migrates aberrantly as a M r ϭ 17,000 species upon 10 -20% SDS-PAGE due to its putative rodlike conformation in the presence of SDS. Intramolecular disulfide bridging of Ng resulting from oxidation with H 2 O 2 causes the protein to have a more compact conformation which is reflected in its faster mobility on 10 -20% SDS-PAGE (Fig.  2). WT, C3S, C4G, C9S, C3S/C4S, C3S/C9S, and C4S/C9S Ng proteins were all intramolecularly disulfide bridged by H 2 O 2 treatment. In contrast, the electrophoretic mobility of C51G and the Tetra Cys mutant proteins were not altered by H 2 O 2 treatment (Fig. 2). The Tetra Cys mutant Ng protein, as expected, could not be intramolecularly disulfide-bridged by H 2 O 2 since it lacks all four Cys residues. The only other Cys mutant protein that could not form the oxidized species was C51G, indicating that Cys 51 is a critical pairing partner in intramo-lecular disulfide bridging as demonstrated by the SDS-PAGE mobility shift assay. The ability of single Cys mutants, C3S, C4G, and C9S, and double Cys mutants C3S/C4S, C3S/C9S, and C4S/C9S to form intramolecular disulfide bridge indicates that Cys 3 , Cys 4 , and Cys 9 individually can pair with Cys 51 (Fig. 2).
Various NO donors, DEANO, SNP, SIN-1, SNOG, and SNAP, as well as the general oxidant IBZ were examined for their ability to form intramolecular disulfide bridging in the WT and 8 Cys mutant Ng proteins (Fig. 3). The NO donors, DEANO, SNP, and SNAP, and IBZ were variably effective in causing intramolecular disulfide bridging in WT, single Cys mutants, C3S, C4G, C9S, and double Cys mutants, C3S/C4S, C3S/C9S, and C4S/C9S. In contrast, SIN-1 and SNOG were ineffective (Fig. 3, all panels, lanes 5 and 6). SIN-1 simultaneously generates 1 mol of NO and 1 mol of O 2 . /mol of SIN-1, and reaction of these products forms peroxynitrite (32). Peroxynitrite, a strong oxidant, oxidizes the SH group to sulfenic and sulfonic acids (32, 33) which likely explains why SIN-1 does not induce disulfide bridging in WT or any of the Cys mutant Ng proteins (Fig. 3, all panels, lane 5). SNOG treat-  Table I for the protein names and corresponding mutations. ment, the only other NO agent tested that did not cause disulfide bridging in the WT or Cys mutant Ng proteins, resulted in an upward shift on SDS-PAGE, or a somewhat higher M r (Fig.  3, all panels except E, lane 6). We interpret this result as SNOG forming stable disulfide adducts (1-4 mol of SNOG/mol of protein, depending on the Ng protein) without concomitant intramolecular disulfide bridge formation. SNOG treatment of the Tetra Cys mutant Ng, as expected, did not result in a shift to the higher M r species (Fig. 3E, lane 6), whereas all other Ng proteins tested did. DEANO and SNP were similarly effective in the extent of disulfide bridging induced in WT and Cys mutant Ng proteins, with the exception of Tetra and C51G proteins which were not reactive. In contrast, SNAP was less effective in the extent of disulfide bridging induced for virtually all of the reactive Ng proteins (Fig. 3, all panels, lane 7). As expected, the Tetra Cys mutant Ng lacking all four Cys residues, showed no intramolecular disulfide bridging with all oxidants tested as demonstrated by a lack of a mobility shift on SDS-PAGE (Fig. 3E). Similar to the results with H 2 O 2 , C51G was the only other mutant Ng which could not be intramolecu-larly disulfide bridged by any of the oxidants tested with resulting increase in electrophoretic mobility, indicating that Cys 51 is a critical pairing partner in NO-mediated disulfide bridging in Ng. Also similar to the H 2 O 2 results, the ability of single Cys mutants, C3S, C4G, C9S, and double Cys mutants, C3S/C4S, C3S/C9S, and C4S/C9S to form intramolecular disulfide bridging with NO donors, DEANO, SNP, and SNAP, indicates that Cys 3 , Cys 4 , and Cys 9 individually can pair with Cys 51 in NO-mediated disulfide bridging (Fig. 3).
To determine the role of disulfide bridge formation in Ng's ability to bind CaM in the absence of Ca 2ϩ , WT, and the eight Cys mutant Ng proteins were examined for their ability to bind CaM under reducing and DEANO-oxidized conditions using a CaM-affinity column assay. Reduced or DEANO-oxidized WT or the eight Cys mutant Ng proteins were applied to a CaM-Sepharose column in the presence of EGTA and after washing in the presence of EGTA (fractions 1-13), the proteins were eluted in the presence of Ca 2ϩ (fractions 14 -28) (Fig. 5). Reduced WT Ng bound to CaM-Sepharose in the presence of EGTA and was eluted with Ca 2ϩ , as expected, whereas the majority of DEANO-oxidized WT Ng did not bind and was excluded from the column, indicating that the NO-mediated intramolecularly disulfide bridged WT Ng was unable to bind CaM (Fig. 5, WT panel). A small fraction of DEANO-oxidized WT Ng did bind to the column and was eluted with Ca 2ϩ ; this is attributed to the minor unoxidized fraction of WT Ng. Similar to WT Ng all of the eight Cys mutant Ng proteins in the reduced state totally bound to the CaM-affinity resin in the presence of EGTA and were eluted with Ca 2ϩ , indicating that single, double, or tetra mutation of the various four Cys residues of Ng did not affect the ability of Ng to bind to CaM (Fig.  5). The single Cys mutant Ng proteins, C3S, C4G, C9S, similar to WT Ng, when DEANO-oxidized, the majority of each of these proteins did not bind and were excluded from the column (Fig.  5). Since the Tetra Cys mutant Ng lacks all Cys residues, it is expected that DEANO oxidation will not affect its binding to CaM. This is in fact what was observed. The majority of the DEANO-treated mutant protein bound to the CaM-affinity resin in the presence of EGTA and was eluted with Ca 2ϩ , and a minor fraction was eluted at the tail part of the EGTA washing. The nature of this material remains unknown (Fig. 5, panel Tetra, fractions [7][8][9][10][11][12][13][14][15]. The C51G mutant when DEANOoxidized, similar to the Tetra Cys mutant, totally bound the CaM-affinity resin in the presence of EGTA and was eluted with Ca 2ϩ (Fig. 5, panel C51G). This result again indicates that Cys 51 is a critical pairing partner in DEANO-mediated disulfide bridge formation in Ng. The ability of the three double Cys mutants, C3S/C4S, C3S/C9S, and C4S/C9S, when DEANOoxidized, to bind CaM was also determined. For the DEANOoxidized C3S/C4S mutant (Cys 9 -Cys 51 pairing), C3S/C9S mutant (Cys 4 -Cys 51 pairing), and C4S/C9S mutant (Cys 3 -Cys 51 pairing) the relative CaM-affinity ratios of not bound to bound protein in the presence of EGTA were ϳ90%/10%, ϳ70%/30%, and ϳ50%/50%, respectively (Fig. 5). These results indicate that the order of reactivity for Cys 3 , Cys 4 , and Cys 9 pairing with Cys 51 is Cys 9 Ͼ Cys 4 Ͼ Cys 3 , which is consistent with the kinetic results (Fig. 4). Rat brain Ng when disulfide bridged by oxidants is a poorer substrate for PKC (28). It was of interest to know how the oxidized Cys mutant Ng proteins behave as substrates for PKC. WT Ng when oxidized with H 2 O 2 was a ϳ2-fold poorer substrate for PKC (Fig. 6, WT). The single Cys mutant Ng proteins, C3S, C4G, and C9S, when H 2 O 2 -oxidized were also poorer substrates for PKC, indicating that the disulfide bridging in these proteins occurs as in WT and results in a poorer substrate for PKC. In contrast, the Tetra Cys and C51G mutant Ng proteins when H 2 O 2 -oxidized, displayed normal activity as a substrate for PKC. The Tetra Cys mutant, lacking all four Cys residues, was, as expected, not affected by H 2 O 2 oxidation. The lack of an effect on substrate ability for H 2 O 2 -oxidized C51G indicates that Cys 51 is a critical pairing partner for disulfide bridging; this is consistent with the above data supporting an essential role of Cys 51 in NO-mediated disulfide bridging of Ng and the subsequent effects on CaM binding and PKC substrate ability.

DISCUSSION
The two known functions of rat brain Ng, namely, binding of CaM and as a substrate of PKC, are simultaneously attenuated following oxidation by several NO donors and other oxidants, such as H 2 O 2 and IBZ, to form intramolecular disulfide. The rate of oxidation of Ng by NO is considerably faster than that of serum albumin, glutathione, or DTT (28), an indication that Ng is a highly favorable acceptor of NO. This previous study did not define the pairing among the four Cys residues of Ng located at positions 3, 4, 9, and 51. Both the mutation of all four Cys residues and the single mutation at Cys 51 to other amino acids prevent the intramolecular disulfide cross-linking that converts the M r ϭ 17,000 to the 10,000 species under nonreducing SDS-PAGE. Single mutation at Cys 3 , Cys 4 , or Cys 9 does not prevent the oxidation. Thus, Cys 51 must form disulfide bond with either Cys 3 , Cys 4 , or Cys 9 , that results in bridging the N-and C-terminal ends of the Ng molecule. This conclusion is supported by the observations that double mutants involving any two of the Cys 3 , Cys 4 , and Cys 9 still can be oxidized. Since the intramolecular disulfide cross-linking is the predominant reaction products of the wild type and the various mutants, this indicates that Cys 51 must be in close proximity to Cys 3 , Cys 4 , and Cys 9 so that the high local concentration favors intramolecular disulfide. Comparison of the rates of oxidation of the three double mutants involving Cys 3 , Cys 4 , and Cys 9 indicates that Cys 51 pairs with Cys 9 most favorably among the three. The current study defines only the oxidant-induced intramolecular disulfide bridge formation involving Cys 51 and any one of the three N-terminal Cys residues that induces mobility shift upon SDS-PAGE. It is conceivable that other disulfide bonds among the N-terminal Cys residues are likely to form upon oxidation, especially for the C51G mutant in which Cys 3 or Cys 4 could form disulfide with Cys 9 .
Reversible oxidation/reduction of Ng affects both the CaMbinding affinity and substrate phosphorylation by PKC. The wild type and all the Cys mutants of Ng are phosphorylated by PKC, and they all bind to the CaM-affinity column under reducing condition. Thus, all four Cys residues are not directly involved in either the CaM binding or substrate recognition by PKC. This is perhaps anticipated since the PKC phosphorylation site and the CaM-binding domain is located within a 19-amino acid Ng/neuromodulin homology (11) situated at the middle of the Ng molecule. Synthetic peptides corresponding to this domain bind CaM and serve as an effective substrate of PKC (15). Intramolecular disulfide cross-linking between the N-and C-terminal ends of Ng results in a gross conformational change that renders the central domain unable to bind CaM and also results in a poorer substrate of PKC. It has been suggested that the Ng central domain can be induced to form ␣-helical structure upon binding to CaM (34). Disulfide crosslinking between the two ends apparently disrupts the "inducedfit" between Ng and CaM. Similar structural constraint probably also plays a role in converting Ng to a weaker PKC substrate following oxidation.
The physiological relevance of Ng oxidation by NO in controlling its binding to CaM and as a substrate of PKC has not been established. Toward this goal, we have shown that Ng associated with rat brain synaptosomes can be oxidized by DEANO when analyzed by immunoblot analysis (28). In addition, we noticed that the oxidized form of Ng was present in the rat brain extracts prepared in buffer without reducing agent, and this oxidized form can be reduced by NADPH. 2 It appears that there are enzymes catalyzing the redox reactions of Ng in the brain. In this respect, Ng is similar to thioredoxins and glutaredoxins (35); all three are small molecular mass proteins (7.5-12 kDa) containing redox active thiols. A major function of thioredoxin and glutaredoxin is as a cofactor for the enzyme ribonucleotide reductase, which catalyzes the conversion of ribonucleotides to deoxyribonucleotides (36). In the case of thioredoxin,thereducedcofactorisregeneratedbyaspecificNADPHdependent thioredoxin reductase (37). However, oxidized glutaredoxin is reduced directly by cellular glutathione, and the resulting GSSG, in turn, is reduced by NADPH via the glutathione reductase (38). Ng does not exhibit any sequence homology with these two proteins, including the short sequence regions containing redox active Cys residues for thioredoxin (-Cys-Gly-Pro-Cys-) (39) and glutaredoxin (-Cys-Pro-Tyr-Cys-) (40). The redox-active thiols of Ng are far apart in its primary structure but are in close proximity in solution. Binding of Ng to CaM in the absence of Ca 2ϩ blocks the oxidation of Ng (28), indicating that these active thiols are no longer available for redox reactions. Thus, Ng can be considered as a Ca 2ϩ -sensitive redox protein as a rise in [Ca 2ϩ ] i favors its dissociation from CaM and subsequent oxidation by NO due to activation of Ca 2ϩ /CaM-dependent NO synthase or other oxidants.