Posttranslational Modification of Glyceraldehyde-3-phosphate Dehydrogenase by S -Nitrosylation and Subsequent NADH Attachment*

Nitric oxide (NO)-related activity has been associated with an NAD (cid:49) -dependent modification of the glycolytic enzyme, glyceraldehyde-3-phosphate dehydrogenase (GAPDH). However, the mechanism by which NO effects covalent attachment of nucleotide and its role in regu- lation of enzyme activity are controversial. Recent studies have shown that S -nitrosylation of GAPDH (Cys 149 ) initiates subsequent modification by the pyridinium co- factor. Here we show that NADH rather than NAD (cid:49) is the preferred substrate. Transnitrosation from active site S -nitrosothiol to the reduced nicotinamide ring system appears to facilitate protein thiolate attack on the enzyme-bound cofactor. This results in attachment of the intact NADH molecule. Moreover, we find that S - nitrosylation of GAPDH is responsible for reversible enzyme inhibition, whereas attachment of NADH accounts for irreversible enzyme inactivation. S -Nitrosylation may serve to protect GAPDH from oxidant inactivation in settings of cytokine overproduction and to regulate glycolysis. NADH attachment is more likely to be a pathophysiological event associated with inhibition of gluconeogenesis. The versatility of NO as a biological messenger reflects its participation in rich additive and redox chemistry.

NO-related signal transduction can be broadly classified as cGMP-dependent or mediated through redox signaling events (1). The latter is, perhaps, best exemplified in the regulation of protein function by S-nitrosylation (16). In the case of enzymes that contain critical thiols at their active site, covalent attachment of the NO group leads uniformly to functional attenuation. Examples of enzymes in this category include cathepsin B, aldolase, ␥-glutamylcysteinyl synthetase, aldehyde dehydrogenase, and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) (see Ref. 1 and references therein). Studies on the potential regulation of GAPDH have received particular attention in view of evidence that NO-related activity (nitric oxide synthase activity or NO donors) stimulates an NAD ϩ -dependent posttranslational modification of active site thiol in association with the loss of protein function (3,17,18). The demonstration of such a modification in cells has led to the proposal that NO induces an ADP-ribosylation reaction reminiscent of that catalyzed by bacterial and mammalian enzymes (19 -21). In this reaction, the ADP-ribose moiety of NAD ϩ is transferred to acceptor amino acids with the release of nicotinamide. Studies by Pancholi and Fischetti (22) may provide the strongest evidence in favor of a true thioglycosidic linkage induced by NO. More recently, however, this mechanism has been challenged by the demonstration that both the ribose and the nicotinamide moieties of NAD ϩ are incorporated by GAPDH (23). This activity implies linkage of the intact molecule to the active site of the enzyme. Furthermore, inhibition of enzyme activity has seemed to correlate better with the extent of S-nitrosylation than the attachment of 32 P-nucleotide, the latter representing only a small fraction of the total protein (23).
We recently probed the mechanism of GAPDH modification using several NO donors. Our studies revealed that NO ϩ transfer to active site thiol is requisite for subsequent modification by [ 32 P]NAD ϩ (3). These data, however, raise a fundamental paradox, as the pathway by which S-nitrosylation facilitates covalent modification by NAD ϩ is not readily apparent. We reasoned, therefore, that NADH rather than NAD ϩ is involved in this reaction, since reduction of nicotinamide would make it susceptible to activation via nitrosative attack. Here we show that 1) S-nitrosylation promotes covalent attachment of reduced nicotinamide; 2) covalent modification by [ 32 P]NADH occurs (largely) via a thionicotinic linkage; 3) S-nitrosylation of GAPDH accounts for reversible enzyme inhibition, and 4) covalent modification by NADH is responsible for irreversible protein inactivation.

EXPERIMENTAL PROCEDURES
Materials-[ 32 P]NAD ϩ (800 Ci/mmol) was purchased from DuPont NEN. SIN-1 was provided by Cassella AG (Frankfurt, Germany). BF 4 NO was purchased from Aldrich, and BF 4 NO 2 was obtained from Fluka. Rabbit muscle GAPDH (80 units/mg) and trypsin (high sequencing grade) were obtained from Boehringer Mannheim. Pertussis toxin A protomer and isocitrate dehydrogenase (NAD ϩ -specific, 31 units/mg) were obtained from Calbiochem. Other chemicals including nicotinamide 1,N 6 -ethenoadenine dinucleotide were of the highest grade of purity made available by Sigma.
HgCl 2 Cleavage of Pertussis Toxin/[ 32 P]NAD ϩ -treated Proteins and of SIN-1/[ 32 P]NADH-treated GAPDH-Pertussis toxin-induced ADPribosylation was achieved by incubating human platelet microsomes (80 g/assay) with 1 mM ATP, 0.1 mM GTP, 10 mM thymidine, 10 M cold NAD ϩ , [ 32 P]NAD ϩ (0.5 Ci/assay), and pertussis toxin A protomer (0.5 g/assay, activated by treatment with 10 mM DTT for 45 min at 37°C). Preparation of platelet microsomes was performed according to published procedures (24). Proteins were precipitated with 200 l of 20% ice-cold trichloroacetic acid and left on ice for 30 min. Following centrifugation (10,000 ϫ g, 15 min) protein pellets were washed twice with ice-cold water-saturated ether. Samples of either pertussis toxin/ [ 32 P]NAD ϩ platelet membranes or SIN-1/[ 32 P]NADH-treated GAPDH were resuspended in 100 l of 100 mM Hepes buffer (pH 7.5) containing 0.5-5 mM HgCl 2 . Cleavage experiments were carried out for 90 min at 37°C and followed by protein precipitation (500 l of 20% trichloroacetic acid). Samples were then processed for detection of protein-bound radioactivity as described above. Controls were treated in the same way, but HgCl 2 was replaced by NaCl.
Tryptic Digestion of NAD ϩ -modified GAPDH-GAPDH (200 g) was modified as described above using 20 mM DTT, 5 mM SIN-1, and 200 M NAD ϩ /[ 32 P]NAD ϩ (500,000 cpm/assay). Reaction mixtures (0.5 ml) were incubated for 60 min at 37°C. Protein was precipitated with 500 l 50% ice-cold trichloroacetic acid, and incubates were left at 4°C for 45 min. After centrifugation, pellets were washed twice with ice-cold watersaturated ether. Samples were then resuspended in 1 ml of 50 mM NH 4 HCO 3 buffer (pH 7.4) and incubated with 20 g of trypsin at 37°C for 4 h. Peptide fragments were separated by reversed phase high pressure liquid chromatography analysis (Nucleosil C 18 ) using gradient elution: 0 -10 min, 100% A; 10 -60 min, 0 -100% B at a flow rate of 1 ml/min. Buffer A contained H 2 O and 0.04% trifluoroacetic acid, while B consisted of 40% H 2 O, 60% acetonitrile, and 0.03% trifluoroacetic acid. Fragments were detected by serial UV absorbance and on-line radioactivity measurements. Radioactive fractions were concentrated, and radioactivity was allowed to decline. The fractions were then subjected to commercial peptide sequencing analysis. Based on theoretical predictions, tryptic digestion of rabbit muscle GAPDH should give three fragments containing cysteine residues. The following peptide contains the active site thiol Cys 149 : IVSNASC 149 TTNC 153 LAPLAK.
GAPDH Activity-GAPDH (1 g/assay), 1 mM DTT, 10 M NAD ϩ , or 10 M NADH, and up to 200 M of a NO donor were incubated in 50 mM triethylammonium buffer (pH 7.5) in a total volume of 50 l at 37°C. Incubation times are indicated below. Samples were diluted in 950 l of 50 mM triethylammonium buffer (pH 7.5) containing 50 mM arsenate, 2.4 mM glutathione, and 100 g/ml glyceraldehyde 3-phosphate, at 37°C. The enzymatic reduction of NAD ϩ to NADH was initiated by the addition of 250 M NAD ϩ . GAPDH activity was monitored by recording the fluorescence emission above 430 nm after excitation at 313 and 366 nm, respectively. Samples without NO donors served as controls.
ADP-ribosylation and GAPDH Modification Using Nicotinamide 1,N 6 -Ethenoadenine Dinucleotide-Nicotinamide 1,N 6 -ethenoadenine dinucleotide (⑀-NAD ϩ ), a fluorescent cofactor analog, can substitute for NAD ϩ in dehydrogenase reactions, including those catalyzed by GAPDH (25). Cleavage of the ␤-nicotinamide linkage results in an increase in fluorescence (25). When ⑀-NAD ϩ was used in pertussis toxin-induced ADP-ribosylation reactions, we observed an increase in fluorescence intensity (compared with controls, in the absence of toxin). ⑀-NAD ϩ (10 M) was used during GAPDH (10 g) modification by 200 M SIN-1 and 2.5 mM DTT. After 40 min we determined GAPDH activity and changes in fluorescence relative to controls lacking the NO donor.
Stoichiometry of GAPDH Modification Achieved with NADH-GAPDH (1 g/assay) was incubated in 100 mM Hepes buffer (pH 7.5) containing 2.5 mM DTT, 200 M SIN-1, [ 32 P]NADH (200,000 cpm/ assay), and 10 M NADH at 37°C for 20 min. Samples (whole assay) were transferred to microconcentrators (Amicon, Microcon, for volumes below 500 l) and centrifuged at 10,000 ϫ g for 10 min. Proteins were washed three times by the addition of 200 l of 100 mM Hepes buffer (pH 7.5) and centrifuged, according to published protocol (23). Samples were then counted in the liquid scintillation counter. A standard curve derived from [ 32 P]NADH/cold NADH ratios served as the basis for stoichiometry calculations.

Covalent Modification of GAPDH: NADH versus NAD ϩ -In
the presence of [ 32 P]NAD ϩ , SNP stimulated thiol-dependent covalent radiolabeling of GAPDH, confirming previous reports. However, the incorporation of radioactivity was more pronounced when using the reduced pyridinium nucleotide, i.e.
[ 32 P]NADH in place of [ 32 P]NAD ϩ (Fig. 1). Although some nonspecific NO-independent labeling occurred with [ 32 P]NADH, the site of protein attachment was found to be distinct from that supported by S-nitrosylation and could be readily controlled for (see below).
Importantly, NO donors of several different molecular classes were found to induce [ 32 P]NADH-dependent GAPDH labeling. However, the time course of enzyme modification varied among compounds (Fig. 2). For example, SNP-induced labeling was detected at 2.5 min and reached saturation by 5 min, whereas SIN-1 modification occurred at a much slower rate. Radiolabeling was noted after 10 min and required 40 min to achieve levels comparable with those with SNP. Nonspecific labeling was observed with NADH but required reducing conditions, i.e. DTT, and much longer reaction times (i.e. labeling was not detectable during the first 20 -30 min).
The pH optimum for nucleotide incorporation was 7.5 for NADH and above 8.5 for NAD ϩ , in agreement with involvement of enzyme (active site) thiolate (Fig. 3). In aggregate, these data are compatible with reports that S-nitrosylation of GAPDH is rate-limiting, since SNP is a better nitrosating agent than SIN-1. We speculate that the higher pH optimum for NAD ϩ may reflect its more efficient reduction by reduced thiol at alkine pH (26). Nonspecific labeling by reduced nucleotide is also more prevalent under alkine conditions.
The amount of radioactivity incorporated into GAPDH (based on equivalent amounts of cold and radioactive labeled nucleotide) was much higher with NADH than NAD ϩ . Under optimal labeling conditions ( Following protein precipitation, the remaining radioactivity was measured as described under "Experimental Procedures." Data are representative of three experiments.

FIG. 3. pH-dependent modification of GAPDH by NAD ؉ and NADH.
Modification of GAPDH in the presence of 10 M [ 32 P]NAD ϩ (120,000 cpm/assay) and 10 M [ 32 P]NADH (120,000 cpm/assay) was determined under the following assay conditions: 100 mM Mes, pH 6.5; 100 mM Hepes, pH 7.5; 100 mM Tris, pH 8.5. SNP and DTT were incubated in each reaction mixture for 20 min. For experimental details see Fig. 1. One of three representative experiments is shown.
14 C]NAD ϩ was modest in comparison with that induced by [nicotinamide-14 C]NADH. These data emphasize involvement of NADH and suggest that its linkage to protein occurs via the nicotinamide ring.
Experiments performed with ⑀-NAD ϩ further support the notion that binding involves the intact pyridine cofactor. Upon cleavage of the ␤-nicotinamide bond in ⑀-NAD ϩ by pertussis toxin, the fluorescence of the molecule increased, much as described previously for NADase (25). In comparison, modification of GAPDH using ⑀-NAD ϩ did not result in a change in fluorescence (Fig. 5), even though the substitution for NAD ϩ led to a comparable degree of GAPDH inhibition. For example, modification of GAPDH (10 g) by 200 M SIN-1 (2.5 mM DTT) in the presence of either 10 M NAD ϩ or 10 M ⑀-NAD ϩ resulted in a 55 Ϯ 0.5% versus 54 Ϯ 3.5% decrease in enzyme activity relative to controls after 40 min, respectively.
We further reasoned that nitrosation of reduced nicotinamide was required for ring activation. This would then facilitate protein thiolate attack on the nucleotide by increasing its electrophilicity. To test the validity of this mechanism, we first examined the effects of nitrosonium-tetrafluoroborate (BF 4 NO), a strong NO ϩ donor (Fig. 6). In the presence of DTT, BF 4 NO potently induced GAPDH labeling by [ 32 P]NADH. Moreover, the NO 2 ϩ -donor nitronium-tetrafluoroborate (BF 4 NO 2 ), which on theoretical grounds should be equally capable of nicotinamide activation (3), resulted in comparable degrees of GAPDH modification (Fig. 6). With both agents, maximal labeling was achieved at concentrations of 50 M under reducing conditions.
Cleavage experiments with HgCl 2 were then performed in order to confirm that the NADH linkage involved protein thiol groups (i.e. Cys 149 of GAPDH). Specifically, the enzyme preparation was treated with HgCl 2 after covalent modification had been induced with SIN-1. HgCl 2 (5 mM) was found to displace the greater part of the [ 32 P]NADH radiolabel (Fig. 7). Importantly, radioactivity incorporated by incubating GAPDH with 10 M NADH and 2.5 mM DTT, i.e. nonspecific labeling, could be readily discriminated from active site modifications due to its resistance toward Hg 2ϩ treatment. 5 mM HgCl 2 or 10 mM DTT was unable to remove any such incorporated radioactivity. NO-independent radioactivity, however, was partially cleaved (around 40% decrease) by treatment with 5 mM NH 2 OH for 90 min. For comparison, the effects of HgCl 2 were examined on human platelet membranes that had been ADP-ribosylated by treatment with pertussis toxin. As previously shown, HgCl 2 (0.5 mM) removed all the radiolabel. Furthermore, over 90% of the radioactivity remained bound to GAPDH following treatment with hydroxylamine.
To further identify the cysteine residue involved by NO, a tryptic digestion of radiolabeled GAPDH was performed, attachment followed by sequence analysis of the single fragment that contained radioactivity. Amino acid sequencing identified the peptide IVSNAS, after which analysis resisted further cycling. The sequence matches identically with the predicted tryptic digestion fragment containing the active site Cys 149 residue of rabbit muscle GAPDH. Thus, cleavage experiments combined with tryptic digestion strongly suggest that the modification of GAPDH occurs at Cys 149 .
Inhibition of Enzyme Activity-In order to explore the relationship among covalent modification of GAPDH, the mechanism of nucleotide attachment, and enzyme activity, we com-FIG. 8. Action of BF 4 NO and NADH on GAPDH enzyme activity. In the upper panel GAPDH was incubated with 100 M nitrosonium tetrafluoroborate (BF 4 NO) for 5 min, followed by the addition of 10 M NADH or 10 mM DTT. Incubations were then continued for the times indicated. Samples treated with NADH/DTT (but not BF 4 NO) served as controls. In the lower panel GAPDH was preincubated with 100 M BF 4 NO, followed by the addition of 10 M NADH (150,000 cpm/assay). Incorporation of radioactivity was carried out as described above. Data are representative of four similar experiments. pared the effects of GAPDH inhibition supported by NADH, NAD ϩ , ␤-nicotinamide mononucleotide, nicotinamide, and nicotinamide N-propanesulfonic acid (Table I).
In general, GAPDH inhibition appeared to correlate well with the extent of nucleotide incorporation. In particular, enzyme inhibition induced by SIN-1 was greatest with NADH. NAD ϩ was significantly less effective and equal in activity to ␤-nicotinamide mononucleotide. Nicotinamide and nicotinamide N-propanesulfonic acid also exhibited inhibitory effects, although they were the least active compounds. The time-and concentration-dependent effects of NADH are detailed in Table  II. Importantly, radiolabel incorporation induced by S-nitrosylation of Cys 149 correlated well with loss of enzyme activity, whereas NO-independent modification of GAPDH by NADH did not (data not shown). Inhibition of GAPDH clearly increased with time and with higher concentrations of the nucleotide.
Experiments were then performed to delineate the role of S-nitrosylation vis à vis covalent NADH attachment in the inhibition of GAPDH (Fig. 8). GAPDH was incubated with BF 4 NO for 5 min followed by the addition of either DTT or NADH. GAPDH catalysis (Fig. 8, upper panel) was then monitored over the ensuing 30 min, and enzyme activity was correlated with the amount of incorporated radioactivity (Fig. 8,  lower panel).
Following the addition of BF 4 NO, we observed an initial drop in enzyme activity by 40% compared with untreated controls. Over the next 5 min, enzyme activity partially recovered spontaneously. Thereafter, either 10 mM DTT or 10 M NADH were added. With the addition of DTT, GAPDH activity was restored to control values (90 Ϯ 2.5%). In contrast, NADH caused a time-dependent irreversible inhibition of the enzyme. By 35 min, only 30% of the initial enzyme activity remained (Fig. 8,  upper panel). Using [ 32 P]NADH, we confirmed that the extent of radiolabeling incorporated by GAPDH paralleled the degree of inhibition following the addition of nitrosating agent (Fig. 8, lower panel). We conclude 1) that S-nitrosylation is responsible for early, reversible enzyme inhibition and 2) that S-nitrosylation promotes subsequent irreversible attachment of NADH to active site thiol. DISCUSSION Nitric oxide has been associated with a mono-ADP-ribosylation-like reaction in which the pyridinium nucleotide undergoes covalent attachment to the active site thiol of GAPDH. However, the mechanism of this reaction is poorly understood, and its contribution to changes in protein function is contro-versial. We recently demonstrated that GAPDH modification is stimulated by S-nitrosylation, i.e. NO ϩ transfer chemistry rather than reaction of nitric oxide (3). In particular, S-nitrosylation of the enzyme active site thiol (Cys 149 ) was found to initiate subsequent covalent modification by NAD ϩ . In order to rationalize this finding, we reasoned that NAD ϩ must first be reduced under in vitro assay conditions (1,3). Transnitrosation from active site RSNO to the nicotinamide ring could then facilitate protein thiolate attack on the nucleotide (1). Here we show that NO-related activity does indeed depend on the presence of reduced nicotinamide and that NADH rather than NAD ϩ is the preferred reaction substrate. Specifically, labeling with NADH is much more efficient than with NAD ϩ , occurs more rapidly, and correlates better with changes in enzyme activity.
Mechanisms of Covalent Modification of GAPDH: S-Nitrosylation versus Nucleotide Attachment-Our results suggest that both S-nitrosylation and covalent attachment of nucleotide are relevant mechanisms of GAPDH modification. The close proximity of the cofactor binding site of GAPDH to active site thiol, evidenced by reports of their participation in a charge transfer complex (27,28), would be predicted to facilitate nitrosative attack at the C-5 position of the nicotinamide ring (29). Resultant ring activation (increased electrophilicity) would then lead to protein thiolate attack at the C-6 position (Fig. 9, upper pathway).
This proposed mechanism is supported by three findings: 1) that several reduced nicotinamide derivatives can substitute for NADH; 2) that treatment with HgCl 2 liberates the reduced nucleotide from active site thiol; and 3) that binding and inhibition by ⑀-NAD ϩ occurs without cleavage of the ␤-glycosidic bond. The involvement of reduced nicotinamide might also explain reports of a NO-associated ADP-ribosylation reaction involving a true thioglycosidic linkage. In this scenario, activation of nucleotide (via transnitrosation), engenders thiolate attack at ribose C-1Ј by making nicotinamide a better leaving group. This reaction (Fig. 9, lower pathway) may be more viable in other proteins where structural constraints place the RSNO in closer proximity to the sugar moiety than the nicotinamide ring.
GAPDH: Enzyme Inhibition and Covalent Modification-Our studies appear to resolve the controversy over the possible contribution of S-nitrosylation versus covalent attachment of nucleotide to enzyme inhibition. Both are likely to contribute, albeit under very different conditions. Reversible enzyme inhibition is mediated by S-nitrosylation. This posttranslational FIG. 9. Suggested reaction sequence of NO-induced NADH attachment to GAPDH. Modification of GAPDH follows the upper pathway leading to a thionicotinic linkage. For details see "Discussion." modification may be involved in the regulation of glycolysis seen in cellular systems activated with inflammatory cytokines (30). S-Nitrosylation might also provide a means to protect such thiol-containing enzymes from oxidative inactivation. In contrast, irreversible enzyme inhibition (seen in some cellular systems (31)) is likely to be explained by covalent attachment of NADH. This modification is more likely to be a pathophysiological event associated with inhibition of gluconeogenesis.