Nitric Oxide-independent, Thiol-associated ADP-ribosylation Inactivates Aldehyde Dehydrogenase*

the activity of glyceraldehyde-3-phosphate dehydrogenase NAD-de- pendent automodification of a cysteine NAD-utilizing dehydro- genase that has a catalytic cysteine, aldehyde dehydrogenase (ALDH), inhibited by nitric oxide. glyceraldehyde-3-phosphate dehydrogenase, in a oxide-independent ADP-ribose, ALDH”, in enzyme activity to less than 10% of control. of modification of the ALDH-active sitive and mercuric hydroxylamine, the ADP-ribosylcysteine linkage synthesized enzymati-cally by pertussis demonstrate a novel means of inactivation of an NAD-dependent enzyme, namely the affinity-based

* The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Recipient of a National Research Council, National Institutes of Health Research Associateship during this work. To whom correspondence should be addressed: National Institutes of Health, Bldg. known to be present in erythrocyte cytosol at 0.45 FM (11). ADP-ribose is degraded by both specific (12,13) and nonspecific (14) pyrophosphatases in animal cells.
The experiments reported here demonstrate a different mechanism of NAD-associated modification of a dehydrogenase cysteine residue. Aldehyde dehydrogenase (ALDH; EC 1.2.1.5): another NAD-dependent dehydrogenase with an active-site cysteine (23-25), was inactivated by stoichiometric, thiol-associated modification by ADP-ribose, independent of the presence of nitric oxide.

Preparation of Radiolabeled ADP-ribo~e-[adenylate-~~P]ADP-ri-
bose was prepared by hydrolysis of [adenylate-32P]P-NAD with an NADase purified from the particulate fraction of rat brain (to >30 pmol (min. mg)") as described elsewhere (26). 100 mM MOPS, pH 7, 100 mM KCl, 5 mM EDTA, 10 mM DTT, 1 ADP-ribosylation of ALDH-ALDH was dissolved at 5 mg. ml" in mM PMSF, 20% (v/v) glycerol (ALDH buffer). PMSF was present to inhibit a protease activity present in some lots of ALDH. ALDH was ADP-ribosylated in reactions containing 6.5 p~ ALDH tetramer (1.5 mg.ml-') and 1 mM [32P]ADP-ribose (approximately 0.1 pCi.nmo1-I) in ALDH buffer (total volume = 0.1 ml). Following incubation at 30 "C for the indicated times, samples (0.01 ml) were transferred to tubes containing 0.24 ml of PD-10 buffer (100 mM sodium acetate, pH 6, 1 mM EDTA, 1 mM PMSF, 0.1% Triton X-100) and applied to PD-10 columns equilibrated with 35 ml of PD-10 buffer. PD-10 buffer, and then ALDH was collected in 0.8 ml. This fraction represented the early portion of the void peak, containing -70% of the applied protein with a background of less than 20 cpm of free, unincorporated [32P]ADP-ribose. The PD-10 fraction was used for each of the assays: ALDH activity as described below, protein by the BCA method (Pierce Chemical Co.), and incorporation of ADP-ribose by scintillation counting. During incubations at 30 "C without ADPribose ALDH was stable for 48 h.
ALDH Assay-ALDH activity was measured by quantifying the absorbance at 340 nm of NADH, a reaction product, in assays containing 100 mM Tris-HC1, pH 8.0, 100 mM KC1, 10 mM DTT, 5 m M EDTA, 1 mM PMSF, 2 mM acetaldehyde, and 0.5 mM P-NAD (total volume = 1 ml), and using an extinction coefficient for NADH of 6.2 mM" (27). Reactions were initiated with ALDH (0.1 or 0.2 pg) and incubated at 25 "C for 30 or 60 min, conditions under which the reaction rate was constant with time, and utilization of NAD was below 20%. Effects of DTT and SNP were assayed using enzyme dialyzed against ALDH buffer without DTT; appropriate blanks for each reaction were included to correct for the absorbance of nitroso-DTT at 340 nm.

RESULTS AND DISCUSSION
Incubation of a brain homogenate with ["'PINAD or [3'P] ADP-ribose resulted in labeling of several proteins, including some that were sensitive to HgC12, suggesting a thiol-associated linkage (28)(29)(30). A -53-kDa brain protein that was labeled with ADP-ribose in a H$+-sensitive linkage was purified by successive column chromatography steps, subjected to micro-sequencing, and identified as the mitochondrial isozyme of ALDH.
The sequence obtained was of a tryptic peptide, which perfectly matched amino acids 476-489 of the deduced amino acid sequence from the bovine isozyme cDNA (31). The sequence was: Glu-Leu-Gly-Glu-Tyr-Gly-Leu-Gln-Ala-Tyr-Thr-Glu-Val-Lys. Previous studies revealed that several mitochondrial proteins react with free ADP-ribose, including a 36-kDa inner membrane protein, and a 52-kDa matrix protein (5,32), raising the possibility that the ADPribosylated matrix protein described previously is ALDH.
The activity and NAD-dependent modification of a different dehydrogenase, glyceraldehyde-3-phosphate'dehydrogenase, were affected by NO (15)(16)(17)(18)(19)22), so the effects of NO on the activity and modification of ALDH were assessed. ALDH activity was stimulated by DTT, with maximal activity at 1-10 mM DTT (Fig. 1A). DTT is required for SNP to release its nitric oxide (21). In the presence of 10 mM DTT, SNP inhibited ALDH activity (Fig. 1B). The effects of SNP and another nitric oxide-donor, S-nitroso-dithiothreitol (prepared as in Ref. 33) on the activity of ALDH and glyceraldehyde-3phosphate dehydrogenase are shown in Table I. The activity of both enzymes was inhibited by the two nitric oxide donors, and in all cases excess DTT reduced the extent of inhibition. These data are consistent with a reversible nitrosylation of the active site thiol, which inactivates the enzyme (18,21).
The interaction of ADP-ribose with ALDH was characterized via NAD kinetics with ADP-ribose as inhibitor. ADPribose from 0.2 to 1 mM inhibited ALDH activity competitively with respect to NAD in 20 min reactions, with no  with excess D T T ALDH was assayed as described under "Experimental Procedures," and GAPDH was assayed as previously described (38), measuring NADH production by A340, and correcting for background A3a of nitrosothiols. S-Nitroso-DTT was prepared as described elsewhere (33). Control activity (100%) was 5.2 pmol (min.mg)" for GAPDH  Ki value of 0.46 mM for ADP-ribose (Fig. 3B). Either higher ADP-ribose concentrations (as in the experiment with 2 mM ADP-ribose shown in Fig. 3A) or longer incubations (data not shown) resulted in mixed-type inhibition with a pronounced V,,, effect, possibly because of the covalent modification of the enzyme with ADP-ribose. Covalent modification of ALDH by 1 mM [32P]ADP-ribose increased to a stoichiometry of >2 mol ADP-ribose.mo1 ALDH" in 12 h and then continued to increase at a slower rate through 48 h (Fig. 4). Over the same time course, the activity of ALDH incubated with ADP-ribose decreased to -10% of control. Control ALDH incubated at 30 "C without ADP-ribose retained full activity for 48 h. In the presence of  Fig. 5. The -90% inhibition observed at -2 mol ADP-ribose.mo1 ALDH" is consistent with the property of "half-of-the-sites reactivity" in the ALDH homotetramer.

-
Consistent with ADP-ribosylation occurring at the active site of ALDH, both ADP-ribose-mediated modification (Fig.  6 A ) and inhibition (Fig. 6 B ) of the enzyme were decreased in the presence of substrates. NAD (10 mM) decreased ADPribosylation and also slowed inhibition of ALDH to a halftime of 7.6 h, compared to a half-time of 2.7 h with ADPribose alone. Acetaldehyde (20 mM) was less effective at protecting ALDH but consistently decreased ADP-ribosylation and slowed the inhibition (half-time = 3.2 h).
The modification of ALDH ADP-ribosylated to 1.8 mol ADP-ribose -mol ALDH" was stable to acid and neutral hydroxylamine but sensitive to base and mercuric ion (Table  11). This pattern of sensitivities, especially the sensitivity to the mild treatment with HgC12, is strongly indicative of an ADP-ribosylcysteine linkage (28-30).
ADP-ribosylation and inhibition of ALDH were markedly pH-dependent. Modification increased with increasing pH from pH 5.5 to 7.5 and then was constant up to pH 9. ADPribosylated ALDH was >90% sensitive to HgC12 over the entire pH range, indicating the presence of mainly ADPribosylcysteine linkages. Incubation with 1 mM ADP-ribose for 2 h inactivated ALDH by 55-60% at pH 7 and above (Fig.  7 ) . Inactivation increased from pH 5 to 7 , with an apparent pK. value of -6.
Several dehydrogenases and other enzymes, both with and without active-site cysteine residues, were incubated with ADP-ribose and the extent of modification measured (Table  111). ALDH was the most reactive acceptor protein of the group, and with a linkage >90% sensitive to HgC12. Other acceptors were modified to 10% or less of the extent of ALDH, and with variable HgCl, sensitivity (0-60%), suggesting that amino acids other than cysteine were modified.
The present work demonstrates a specific active-site thiolassociated modification of ALDH by a novel affinity reagent, ADP-ribose. The modification of an essential cysteine in ALDH appears to be a special case of an enzyme with a reactive cysteine in a coenzyme site that has an affinity for ADP-ribose. Whether ALDH reacts with ADP-ribose in the mitochondria is unknown. Considering that the reaction of 1 mM ADP-ribose with ALDH was rather slow, on the order of hours, and not knowing the concentration of ADP-ribose in the mitochondrial matrix, it is difficult t o guess to what extent Stability of the ALDH-ADP-ribose linkage ALDH was incubated with 1 mM [32P]ADP-ribose for 16 h, and then separated from free ADP-ribose, as described under "Experimental Procedures." Samples (0.01 ml) from the PD-10 void fraction were combined with an equal volume of buffer containing 1% SDS, 1 mM PMSF, and the following additions: control, 10 mM EDTA HCI, 0.2 M HC1; NaOH, 0.2 M NaOH, 10 mM EDTA; HgC12,lO mM HgCL; NH,OH, 2 M NH20H, 0.1 M Tris, 10 mM EDTA, adjusted to pH 7.0 with NH,OH. After incubation at 30 "C for 2 h, mixtures were applied to PD-10 columns equilibrated with PD-10 buffer. Columns were washed successively with 0.6 ml and 1.75 ml of PD-10 buffer, and then the void fraction ( V , = 2.75-4 ml) was collected for liquid scintillation counting. Data shown are means of duplicates from one experiment representative of three. this enzyme may be ADP-ribosylated in uiuo. If the modification were stable chemically, and not removed by the action of an ADP-ribosylcysteine hydrolase, then the modification of ALDH could accumulate over time, dependent on turnover of the enzyme. ADP-ribose has been shown to react with free cysteine to form a thiazolidine linkage in which both the amino and sulfhydryl groups of cysteine are linked to ADP-ribose (26). Presumably, the ADP-ribose linkage with ALDH is not of the thiazolidine type, because the amino group of the ALDH cysteine would be included in an amide linkage, and therefore not available for thiazolidine formation. Rather, the ADPribosylcysteine linkage in ALDH resembles that synthesized by pertussis toxin from NAD in the guanine nucleotidebinding regulatory proteins (28,29). The chemical reactivity of the ADP-ribose bond in ALDH is also consistent with a thioglycoside linkage; the linkage was stable in hydroxylamine, but sensitive to H$+, whereas the thiazolidine-type ADP-ribosylcysteine linkage is sensitive to both hydroxylamine and H$+ (26).  = 0.1 ml). Samples (0.025 ml) of reaction mixtures were boiled for 5 min in 0.25 ml of 1% SDS, 0.1 M sodium acetate, 5 mM EDTA, pH 6, and applied to PD-10 columns equilibrated with the same buffer. The void fraction (2.75-3.55 ml) was collected for protein and radioactivity measurements. Acceptor proteins were: alcohol dehydrogenase and aldehyde dehydrogenase from S. cereuisiae; lactate-, glycerol 3-phosphate-, and glyceraldehyde 3phosphate dehydrogenases from rabbit muscle; papain; cathepsin C from bovine spleen; L-1-tosylamido-2-phenylethyl chloromethyl ketone-treated trypsin from bovine pancreas (Worthington); and V-8 protease (ICN). Unless specified otherwise, all proteins were from Boehringer Mannheim. The data shown are averages of duplicates from one experiment representative of three. A different modification is the NAD-dependent covalent modification of glyceraldehyde-3-phosphate dehydrogenase that is stimulated by NO (15)(16)(17)(18)(19)(20)(21)(22). The modification is not auto-ADP-ribosylation, but is in fact a novel linkage of intact NAD to a glyceraldehyde-3-phosphate dehydrogenase thiol group (22). The NO-stimulated modification of glyceraldehyde-3-phosphate dehydrogenase by NAD was too limited to account for the much greater extent of enzyme inhibition (15)(16)(17)(18)22). Both ALDH and glyceraldehyde-3-phosphate dehydrogenase were inhibited by nitric oxide itself, probably because of nitrosylation of the catalytic cysteines (18,21). ALDH was also inhibited by the NO-independent modification with ADP-ribose, which was at stoichiometric levels and also probably at the active site.
Together, these nonenzymatic reactions of NAD or its metabolite ADP-ribose with thiols add a previously unrecognized level of complexity to the investigation of endogenous cysteine-specific ADP-ribosylation in mammalian cells. The one reported eukaryotic NAD:cysteine ADP-ribosyltransferase was purified on the basis of its ability to ADP-ribosylate cysteine methyl ester, without positive identification of a thioglycoside linkage in the product (34). As shown previously, the product of this enzyme could result from generation of ADP-ribose with a subsequent nonenzymatic condensation of ADP-ribose with cysteine methyl ester to yield a thiazolidine (26). Despite the uncertainty about endogenous NAD:cysteine ADP-ribosyltransferases, there is, nevertheless, direct evidence for the presence of ADP-ribose attached to cysteine residues in the plasma membrane fraction of rat liver (35). Circumstantial evidence consistent with the existence of ADP-ribosylcysteine linkages in animal cells comes from the identification of thioglycosidases in animals (36), including an enzyme that removes ADP-ribose from a cysteine in the a-subunit of the heterotrimeric guanine nucleotide-binding protein, Gi ("inhibitory" G-protein) (37).
The data reported here demonstrating high-stoichiometry nonenzymatic ADP-ribosylation of cysteine in ALDH, and the earlier results of nonenzymatic ADP-ribosylation of free cysteine (26), point out the precautions that must be taken in attempting to study endogenous ADP-ribosylation of cysteine residues.