ADP-ribosylation of a Mr 21,000 membrane protein by type D botulinum toxin.

When crude membrane fraction from bovine adrenal gland was incubated with type D botulinum toxin in the presence of NAD, a membrane protein with a molecular weight of 21,000 was specifically ADP-ribosylated. This ADP-ribosylation occurred dependent on the dose of the toxin and was abolished by prior boiling ADP-ribose transfer to the membrane protein was significantly suppressed when agmatine and L-arginine methyl ester were included in the reaction mixture. Dithiothreitol stimulated this ADP-ribosylation about 3-fold. Incubation of membrane fractions from mouse brain and pancreas with this toxin also resulted in ADP-ribosylation of a protein of the same molecular weight. These results suggested that type D botulinum toxin catalyzed transfer of an ADP-ribose moiety of NAD to the specific membrane protein common to secretory cells.

Botulinum toxins are potent neurotoxins which act at presynaptic terminals of cholinergic as well as other neurons and block the release of neurotransmitters (1). Although the exact mechanism of this inhibition has not yet been clarified, a three-step model has been proposed to account for their actions (2). This model consists of the binding step in which the toxins bind to receptors on the plasma membrane of target cells, the internalization step in which they enter the cells possibly via receptor-mediated endocytosis, and the poisoning step in which they damage release function of the cells. This model has been critically examined in recent years. Among the three steps, the former two steps have been proved in several tissues using various types of botulinum toxins (3-61, but the mechanism of intracellular poisoning remains to be elucidated. A hypothesis that botulinum neurotoxin is an enzyme has been proposed based on several findings (2). They include the remarkable potency and long duration of action which are difficult to explain by a nonenzymatic mechanism. The hypothesis is also based on many similarities of botulinum toxin in structure and origin to diphtheria toxin which * This work was supported in part by Grants-in-Aid for Scientific Research 60570130, 61570138, and 61770187 from the Ministry of Education, Science and Culture of Japan and by Special Project Research Grant 60214020 and grants from the Japanese Foundation on Metabolism and Diseases and the Tokyo Biochemical Foundation. 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.
B To whom correspondence should be addressed.
is an enzyme with ADP-ribosyltransferase activity (2, 7). In this study we have examined enzyme activity of type D botulinum toxin on the homogenate of bovine adrenal gland, since adrenal medullary cells are sensitive to this type of botulinum toxin (8). We provide evidence that type D toxin is really an enzyme with ADP-ribosyltransferase activity.

EXPERIMENTAL PROCEDURES
Materials-Type D botulinum toxin (2.0 X 10' mice LDSo1/mg of protein) was obtained from Wako Chemicals, Osaka, Japan. The toxin was purified according to the method of Miyazaki et al. (9) and the preparation was apparently homogeneous on native polyacrylamide gel electrophoresis. On SDS'-polyacrylamide gel electrophoresis the preparation showed four protein bands at molecular weights of 110,000, 100,000, 48,000, and 16,000. The 7 S neurotoxin component was isolated from this progenitor toxin by DEAE-Sephadex chromatography as described (9, 10). Cholera toxin was from the Chemo-Sero Therapeutic Research Institute, Kumamoto, Japan. After incubation for 45 min at 30 "C, 20 pl of 100% (w/v) trichloroacetic acid was added. Proteins were precipitated by centrifugation, solubilized in 2% SDS, and subjected to SDS-polyacrylamide gel electrophoresis. SDS-polyacrylamide gel electrophoresis was carried out according to the method of Laemmli (12). Autoradiography of the dried gel was carried out as previously described (13). Quantification of the incorporated radioactivity was carried out by excising radioactive bands from the gel and measuring 32P contents in a Triton/toluene scintillator.
Release of [14C]nicotinamide from [~arbonyl-'~C]NAD was assayed as described by Moss and Vaughan (14). Briefly, the assay mixture (200 pl) contained 100 mM Tris-HC1 (pH 7.6), 10 mM dithiothreitol, 19 p~ [carbonyl-"C]NAD (100 cpm/pmol), 75 mM either agmatine sulfate or L-arginine methyl ester, and 15 pg either type D toxin or cholera toxin. After the reaction was carried out at 30 "C for 90 min, the mixture was chilled and applied to the Dowex 1 column (formate form, 1 ml of bed volume). Released ["CJnicotinamide was eluted The abbreviations used are: LD6o, median lethal dose; SDS, sodium dodecyl sulfate. with 2.7 ml of 20 mM formic acid, and the radioactivity in the 2-ml eluate was determined.
Product Analyses-Trypsin digestion of the radioactive product was carried out as follows. Acid-insoluble precipitates of radioactive products were rinsed with diethyl ether and dissolved in 100 pl of 0.1 M NaOH. The pH of the solution was adjusted to pH 8.0 by adding 50 p l of 0.3 M Tris-HC1 (pH 7.0). Trypsin (10 pg) was added to the solution, and the mixture was incubated a t 37 "C for 2 h. After incubation, 15 pl of 100% trichloroacetic acid was added, and the mixture was centrifuged to separate acid-soluble and acid-insoluble radioactivities.
Digestion of products with phosphodiesterase and paper and Dowex 1 column chromatographies of the digested products were carried out as described (13,15). The gel corresponding to a radioactive band a t M, 21,000 was excised out, and the radioactive product was extracted from the gel by the method of Ferro and Olivera (16). Phosphodiesterase digestion was carried out by incubating the product with 2 units of phosphodiesterase in 25 mM Tris-HCI (pH 9.0), 10 mM MgCIZ, and 1 mM AMP at 37 "C for 6 h. A digested sample containing 3,000 cpm of "P was applied to a Dowex 1 column (formate form, 0.7 X 10 cm) together with unlabeled AMP (0.12 pmol) and ADP-ribose (0.2 pmol). The column was eluted with a linear gradient of 0 to 6 M formic acid (total volume, 100 ml). Fractions (1 ml) were taken and the radioactivity and absorbance a t 260 nm were determined.

RESULTS
Crude membrane fraction from bovine adrenal gland was incubated with type D botulinum toxin in the presence of [CY-"'PINAD. After incubation, radioactivity incorporated into the acid-insoluble fraction was analyzed by SDS-polyacrylamide gel electrophoresis and autoradiography. As shown in Fig. L4, an intense radioactive band was observed a t a M, of 21,000 in the presence of the toxin in the reaction mixture (lunes 1 and 2). When the labeled membrane was treated with trypsin and then centrifuged, more than 99% of the radioactivity was recovered in the supernatant, confirming that the radioactivity was incorporated into protein. The addition of 10 mM dithiothreitol significantly enhanced this labeling ( l a n e 3 ) , whereas agmatine or L-arginine methyl ester, known inhibitors of NAD:arginine ADP-ribosyltransferase (17,18), significantly attenuated the labeling (lanes 4 and 5 ) . The botulinum toxin also catalyzed labeling of a membrane protein of the same molecular weight in tissues other than bovine adrenal gland. As shown in Fig. lB, (Fig. 2). T o identify the radioactive product, the "2P-labeled product was extracted from the gel and treated with snake venom phosphodiesterase. When the digested products were analyzed by Dowex 1 column chromatography, more than 95% of the radioactivity was recovered as ["'P]5'-AMP. A similar finding was also obtained by paper chromatography; about 93% of the total radioactivity was comigrated with authentic 5'-AMP at the RF of 0.54 (data not shown). Thus, these results taken together strongly suggested that type D botulinum toxin catalyzed mono ADPribosylation of the specific membrane protein of M , 21,000.
In order to confirm that the enzyme activity was associated with the toxic component of the progenitor toxin, we carried out DEAE-Sephadex column chromatography of the toxin to isolate the 7 S toxic component (Fig. 3). As described previously (9), the toxin was eluted in two protein peaks on this chromatography; the first and second peaks were the toxic and nontoxic components, respectively. On SDS-polyacryl- of the type D progenitor toxin. The toxin (600 pg) in 10 mM potassium phosphate (pH 7.8) was applied to a column of DEAE-Sephadex (0.9 X 1.1 cm) equilibrated with the same buffer. Elution was performed with the linear gradient from 0 to 0.3 M NaCl in the buffer in a total volume of 15 ml. Fractions (0.5 ml) were collected, and protein contents (0) and ADP-ribosyltransferase activity on bovine adrenal membrane in the presence of 10 mM dithiothreitol (0) were assayed. amide gel electrophoresis the toxic component showed two protein bands a t molecular weights of 100,000 and 48,000 which corresponded to a heavy and light chain of the toxic component, respectively, as described (9, 10). The ADP-ribosyltransferase activity was eluted only with the first toxic component.
Effects of various compounds which affect ADP-ribosylating activity of other bacterial toxins such as cholera toxin and pertussis toxin (17-23) were examined on this reaction. As shown in Table I, dithiothreitol stimulated the activity of the toxin about 3-fold. Small stimulation was also observed with GTP and ATP (about 1.5-fold). Among various amino acid analogues, L-arginine, L-arginine methyl ester, and agmatine  could inhibit the enzyme activity. Inhibition by agmatine was observed in a concentration-dependent manner over 2 mM and was complete at 75 mM. Since these compounds work as alternative acceptors of the ADP-ribose moiety in the cholera toxin-catalyzed reaction (14), we measured the releases of [14C]nicotinamide from [curbor~yl-'~C]NAD in the presence of 75 mM agmatine or arginine methyl ester by type D toxin and compared them with those by cholera toxin. The radioactivity released in the absence of toxins was 1,893 cpm. Cholera toxin enhanced the release to 2,300 cpm in the absence of compounds and to 18,601 and 17,687 cpm in the presence of agmatine and arginine methyl ester, respectively. In contrast, no significant enhancement of the release was observed with the type D toxin. The released radioactivities were 1,895 cpm with the toxin only and 1,899 and 2,108 cpm in the presence of agmatine and arginine methyl ester, respectively. Enhanced release was not observed a t lower concentrations (5, 10, 20, and 50 mM) of agmatine or arginine methyl ester, either.

DISCUSSION
Although accumulating evidence has suggested an enzyme nature of botulinum neurotoxins (1,2), there has been no report to identify any enzyme activity in this species of toxin. In this communication, we have demonstrated for the first time that type D botulinum neurotoxin has an enzyme activity and ADP-ribosylates a M , 21,000 membrane protein in bovine adrenal gland. This modification reaction was stimulated by dithiothreitol and inhibited by agmatine and arginine methyl ester. Since botulinum toxins consist of light and heavy chains bound by an interchain disulfide bond and toxic activity resides solely in the light chain (1,2), the dithiothreitol effect is probably due to cleavage of this disulfide bond and release of active light chain molecules. Agmatine and arginine methyl ester are known as inhibitors of an NAD:arginine ADPribosyltransferase such as cholera toxin (17,18); however, unlike cholera toxin, the type D toxin did not show significant release of [14C]nicotinamide from [c~rbonyl-'~C]NAD on incubation with these compounds. This result suggests that both agmatine and arginine methyl ester work as pure inhibitors and not as alternative acceptors of the enzyme. Thus, type D botulinum toxin is somewhat different in catalytic properties from cholera toxin (14, 19).
We found that the ADP-ribosyltransferase activity was associated with the toxic component of the progenitor toxin. It is, therefore, most likely that the ADP-ribosylation occurs in situ in poisoned cells and alters cell function. Protein substrate for ADP-ribosylation by type D toxin was a M , 21,000 membrane protein. There has been no report on ADPribosylation of this protein by other bacterial toxins and mammalian ADP-ribosyltransferases. We have also shown that this protein is present not only in adrenal gland but also in other tissues such as brain and pancreas. Recently Knight et al. (8), using bovine adrenal medullary cells, showed that type D botulinum toxin inhibited the release of catecholamine downstream from Ca'+ entry into cells. These results strongly indicate that the M, 21,000 protein identified here is directly involved in the exocytosis process in general. Since almost all substrate proteins for bacterial toxin-catalyzed ADP-ribosv-lation thus far characterized are GTP-binding proteins, it is probable that this M, 21,000 protein is also a GTP-binding protein. Recently, a GTP-binding protein with a molecular weight of 21,000 (G,) was purified from human placental membranes (24). Kahn and Gilman (25) also reported recently that the protein cofactor necessary for ADP-ribosylation of G, is a GTP-binding protein of similar molecular weight (25). Whether the substrate protein for type D toxin is identical with such a protein should be rigorously examined, as should whether or not ADP-ribosyltransferase is a common property of all botulinum neurotoxins. We could not demonstrate such activity in type A botulinum toxin under the same conditions as for type D toxin.' Whether this is due to the lack of such activity in type A toxin or to suboptimal experimental conditions for type A toxin should be clarified in the future.