Activation by thiol of the latent NAD glycohydrolase and ADP-ribosyltransferase activities of Bordetella pertussis toxin (islet-activating protein).

Pertussis toxin (islet-activating protein) activates adenylate cyclase in susceptible cells by ADP-ribosylating an inhibitory component of the cyclase system. This toxin, assayed in a cell-free system in the presence of high concentrations of thiol, catalyzed the hydrolysis of NAD to ADP-ribose and nicotinamide. This NAD glycohydrolase activity co-chromatographed on Sephacryl G-200 in 6.5 M urea, pH 3.2, 0.1 M glycine with the ADP-ribosyltransferase activity of the toxin, as monitored by the transfer of [32P]ADP-ribose from [32P]NAD to a 41,000-Da protein in NG108-15 neuroblastoma X glioma hybrid cells. In the absence of thiol, the native holotoxin was enzymatically inactive. Following addition of 250 mM dithiothreitol to the assay, maximal enzymatic activity was evident after a delay of approximately 1 h; with 20 mM thiol, the delay was longer. The Km for NAD with the fully activated enzyme was 25 microM; the Km did not appear to vary with the extent of activation. Thiol was necessary in a cell-free system to demonstrate NAD glycohydrolase activity. When extensively washed membranes were used as a source of 41,000-Da substrate, thiol was necessary to observe ADP-ribosylation in some cases (human erythrocytes) and significantly stimulated activity in others (NG108-15 cells). In contrast to the bacterial toxins choleragen and Escherichia coli heat-labile enterotoxin that ADP-ribosylate stimulatory components of the cyclase system, pertussis toxin did not transfer ADP-ribose to low molecular weight guanidino compounds, such as arginine or agmatine.

x glioma hybrid cells. In the absence of thiol, the native holotoxin was enzymatically inactive. Following addition of 250 mM dithiothreitol to the assay, maximal enzymatic activity was evident after a delay of -1 h; with 20 mM thiol, the delay was longer. The K,,, for NAD with the fully activated enzyme was 25 PM; the K,,, did not appear to vary with the extent of activation.
Thiol was necessary in a cell-free system to demonstrate NAD glycohydrolase activity. When extensively washed membranes were used as a source of 41,000-Da substrate, thiol was necessary to observe ADP-ribosylation in some cases (human erythrocytes) and significantly stimulated activity in others . In contrast to the bacterial toxins choleragen and Escherichia coli heat-labile enterotoxin that ADPribosylate stimulatory components of the cyclase system, pertussis toxin did not transfer ADP-ribose to low molecular weight guanidino compounds, such as arginine or agmatine. The adenylate cyclase system is critical to the control of certain metabolic pathways by hormones and bacterial toxins. Hormones acting through specific cell surface receptors appear to have either stimulatory or inhibitory effects on cyclase (l-3). With the stimulatory receptors, it appears that activation of the catalytic unit depends on a coupling protein that requires GTP for activity and is known as G, or G/F factor (l-2). With the inhibitory hormones, a similar case has been made for the presence of a GTP-dependent inhibitory coupling factor (Gi) (3). At least three bacterial toxins appear to exert their effects on cells through activation of the adenylate cyclase system (4)(5)(6)(7). Choleragen (cholera toxin) and Esche-*This work was supported by the Rockefeller Foundation, National Institutes of Health Grant ROl-A118000, and University of Virginia Diabetes Center Grant AM22125. 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 USC. Section 1734 solely to indicate this fact. richia coli heat-labile enterotoxin, the agents responsible in part for cholera and "traveler's diarrhea," respectively, activate cyclase by catalyzing the transfer of the ADP-ribose moiety of NAD to the G, protein (4,(8)(9)(10). ADP-ribosylation of G, appears to promote the association of the protein with GTP and maintenance of an active G,. GTP complex (11,12). A toxin from Bordetella pertussis appears to increase the responsiveness of cyclase to stimulatory hormones and decreases the effect of inhibitory agonists (6,7). The effects of this toxin result from ADP-ribosylation of a 41,000-Da membrane protein (13). Based on the effects of the modification of cyclase and the effects of GTP on action of inhibitory hormones, this protein has been termed Gi, the inhibitory GTP-binding protein.
The ADP-ribosylation of Gi appears to be catalyzed by both holotoxin and a specific subunit, termed Sl (14).
In the present report, we note that the enzymatic activity of the toxin is latent; activation of the toxin was achieved in the presence of high concentrations of thiol. Some membrane preparations were capable of toxin activation independent of exogenous thiol, whereas others were not. Pertussis toxin showed significant differences in enzymatic properties from choleragen and E. coli heat-labile enterotoxin.

EXPERIMENTAL PROCEDURES
Methods NAD Glycohydrolase and ADP-ribosyltransferae Assays-The NAD glycohydrolase activity of pertussis toxin was determined in a standard assay mixture that contained 50 mM potassium phosphate, pH 7.5,,000 cpm), 1 mg/ml of ovalbumin, and the indicated additions in a total volume of 0.3 ml.  centrifuged (500 X g, 5 min); membranes were pelleted from the supernatant by centrifugation at 18,000 X g for 5 min, washed twice with 50 volumes of 50 mM potassium phosphate buffer, pH 7.5, and suspended in the same buffer (-2 mg of protein/ml). Human erythrocytes were lysed, washed twice in potassium phosphate buffer, pH 7.5, and suspended in the same buffer (-2 mg/ml).
Electrophoresis in 12% polyacrylamide gels containing sodium dodecyl sulfate and autoradiography were performed as described (20).
High Pressure Liquid Chromatography-High pressure liquid chromatography was performed on a DuPont 8800 high pressure liquid chromatograph equipped with an LKB multirack fraction collector using a DuPont Zorbax-SAX column (25 X 4.6 cm) with a mobile phase of 50 mM potassium phosphate buffer, pH 4.5, at a flow rate of 1.5 ml/min at 45 "C. Fractions of 0.2 min were collected for radioassay.
Protein Assay-Protein was determined by the method of Lowry et al. concentrations and could be reduced by incubations with thiol prior to assay (Fig. 3). In addition to dithiothreitol, other thiols were also capable of activating the toxin (Table I); dithiothreitol and 8-mercaptoethanol appeared to be more effective than glutathione or cysteine.
The K,,, for NAD in the thiol-dependent reaction was " 2 0 p~ (Fig. 4); the K,,, was similar with toxin preparations that varied over a 10-fold range in the extent of activation (data  pg/ml) was incubated for 8 min at 37 "C in buffer containing 50 mM glycine, pH 8.0, 0.25 mg/ml of ovalbumin, with or without 20 mM dithiothreitol. 25 pl of one of the toxin mixtures or an identical mixture not containing toxin were then added to samples of erythrocyte membranes (50 pg of protein) or NG108-15 membranes (50 pg of protein) to give a final volume of 100 pl which also contained 20 p~ [32P]NAD (-20 pCi/ml), 20 mM thymidine, 0.5 mM ATP, 0.5 mM GTP in 50 mM potassium phosphate, pH 7.5. Samples were incubated at 37 "C for 30 min. Trichloroacetic acid precipitation, electrophoresis, and autoradiography were performed as described in the legend to Fig. 5

FIG. 5.
Co-chromatography of the ADP-rihyltransferase and NAD glycohydrolase activities of pertussis toxin on gel permeation columns. Pertussis toxin was dialyzed extensively against 6.5 M urea, pH 3.2, 0.1 M glycine, as described previously (15, 22); this procedure was adapted from that used to resolve the enzymatically active A subunit of choleragen from the B or binding subunit. The sample (34 pg) was then chromatographed on a Sephacryl G-200 column (1.2 X 94.7 cm); 1.0-ml fractions were collected. A, identification of column fractions containing ADP-ribosyltransferase activity. Samples of NG108-15 membranes (150 pg of protein) were incubated with 10 pl of column fraction in 50 mM potassium phosphate buffer, pH 7.5, containing 20 p~ [32P]NAD (-20 pCi/ml), 0.5 mM ATP, 0.5 mM GTP, 6.5 mM dithiothreitol, and 20 mM thymidine (total volume = 100 pl). Thymidine was included to inhibit endogenous mono-ADP-ribosyltransferases (23) and poly(ADP-ribose) synthetases (24,25). Samples were incubated at 37 "C for 30 min. After addition of 2 ml of 10% trichloroacetic acid and centrifugation at 18,000 X g for 5 min, the precipitated material was solubilized in 1% sodium dodecyl sulfate and subjected to electrophoresis and autoradiography. Autoradiogram shows proteins labeled during incubation of membranes with [32P]NAD and samples of column fractions 65-72. The arrow points to the 41,000-Da substrate for pertussis toxin. B, determination of the NAD glycohydrolase activity in fractions from a Sephacryl G-200 column. Samples (30 pl) were assayed for 18 h at 30 "C in a reaction mixture containing 250 mM dithiothreitol. The presence of the column buffer resulted in a significant inhibition of enzymatic activity (-45%). The yield of NAD glycohydrolase activity was 68.8%. NAD glycohydrolase activity of various column fractions is plotted (0). Densitometric scans of autoradiograms of the 41,000-Da protein labeled by various column fractions were made and intensity of these bands is also plotted (0).

Activation of Pertussis
Toxin by Thiol not shown). NAD glycohydrolases are fairly ubiquitous; to establish that the thiol-dependent hydrolytic reaction noted in pertussis toxin preparations was catalyzed by the toxin, the preparation was dialyzed against 6.5 M urea in 0.1 M glycine, pH 3.2, and chromatographed on a Sephacryl G-200 column. The ADP-ribosyltransferase activity of the toxin was monitored by the transfer of radioactivity from ["PINAD to a 41,000-Da membrane protein in NG108-15 cells (Fig. 5 A ) , and co-chromatographed with NAD glycohydrolase activity (Fig. 5B).
Although thiol was clearly necessary to demonstrate the NAD glycohydrolase activity of pertussis toxin, it did not appear to be necessary in all systems to demonstrate ["PI ADP-ribosylation of membrane proteins (Figs. 6, A and E). A dithiothreitol requirement was observed for erythrocyte membranes (Fig. 6A); it was not observed for membranes from NG108-15 cells, although the thiol enhanced [32P]ADP-ribosyltransferase (Fig. 6B).

DISCUSSION
The enzymatic activity of bacterial toxins dependent on ADP-ribosylation for their action (e.g. diphtheria toxin, Pseudomonas exotoxin A, choleragen, E. coli heat-labile enterotoxin) appears to be latent; upon mild denaturation, proteolytic digestion, or reduction of the native protein, there is a large increase in enzymatic activity (15,(26)(27)(28)(29).
It appears that activation of choleragen by thiol results from reduction of a single disulfide bond linking two peptides in the A subunit (30). With pertussis toxin, there is no evidence that a similar critical bond linking two proteins exists (14). Pertussis toxin is an oligomeric protein; one subunit, termed SI, appears to possess ADP-ribosyltransferase activity (14). Conceivably, however, thiol may act to release the SI subunit from the holotoxin complex or to cleave a disulfide in SI.
The K,,, for NAD with thiol-activated pertussis toxin was -25 PM; this figure is considerably lower than that observed with choleragen and E. coli heat-labile enterotoxin, where the K,,, values were - 4-8 mM (15, 27). It is similar, however, to the K,,, values obtained with diphtheria toxin and Pseudomonas exotoxin A (31, 32). Katada and Ui (33), investigating the ADP-ribosylation reaction catalyzed by pertussis toxin, report a low K , in the micromolar range; the apparent K , observed in the ADP-ribosylation reaction, however, may be dependent on the effect of the ADP-ribose acceptor Gi on the toxin's affinity for NAD. Thus, one might not expect a priori that the apparent K , values in the ADP-ribosyltransferase and NAD glycohydrolase reactions would be similar. In addition, the NAD concentration is affected by endogenous NAD glycohydrolases present in membrane preparations. Although thiol was clearly necessary to demonstrate NAD glycohydrolase activity, a requirement for thiol was not demonstrated in studies by Katada and Ui (33). Indeed, the present studies demonstrate that thiol is necessary only in some membrane systems. These findings suggest that membranes contain a factor(s) capable of activating pertussis toxin. For choleragen, it has been shown that disulfide exchange catalyzed by an oxidoreductase resulted in significant activation of the toxin at low thiol concentrations (34) haps a similar factor may potentiate activation of pertussis toxin by endogenous thiols.