Heart contains two substrates (Mr = 40,000 and 41,000) for pertussis toxin-catalyzed ADP-ribosylation that co-purify with Ns

Two peptides (Mr = 40,000 and 41,000) in membranes of rabbit heart are radiolabeled when the membranes are incubated in the presence of activated pertussis toxin and [32P]NAD+. The 41,000-Mr peptide appears to be the alpha subunit of the inhibitory regulatory protein of adenylate cyclase, Ni. The 40,000-Mr substrate for pertussis toxin in the heart was investigated. Purification of the stimulatory regulatory protein of adenylate cyclase, Ns, results in the co-purification of the alpha subunits of both Ns and Ni, the putative beta- (Mr = 35,000) and gamma- (Mr approximately equal to 15,000) subunits of Ns and Ni, and the additional 40,000-Mr peptide that is ADP-ribosylated by pertussis toxin. This 40,000-Mr substrate for pertussis toxin action appears to be a major N-protein of mammalian heart.     

Cholera toxin can catalyze ADP-ribosylation of cytoskeletal proteins

Cholera toxin catalyzes transfer of radiolabel from [32P]NAD+ to several peptides in particulate preparations of human foreskin fibroblasts. Resolution of these peptides by two-dimensional gel electrophoresis allowed identification of two peptides of Mr = 42,000 and 52,000 as peptide subunits of a regulatory component of adenylate cyclase. The radiolabeling of another group of peptides (Mr = 50,000 to 65,000) suggested that cholera toxin could catalyze ADP-ribosylation of cytoskeletal proteins. This suggestion was confirmed by showing that incubation with cholera toxin and [32P]NAD+ caused radiolabeling of purified microtubule and intermediate filament proteins.     

Peptide mapping of fat cell membrane substrates for cholera toxin-catalyzed ADP-ribosylation

Incubating rat fat cell membranes with [³²P]NAD⁺ and cholera toxin results in ADP-ribosylation of three distinct components with approximate molecular weights of 42 000, 46 000 and 48 000. Partial proteolytic peptide maps of the M_r = 46 000 and 48 0000 toxin-specific substrates generated by elastase, α-chymotypsin, or Staphylococcus aureus V-8 protease were nearly identical, while those of the M_r = 42 000 target lacked several peptides common to both of the larger molecular weight targets. In addition, peptide maps generated from the M_r = 42 000 target displayed a number of peptides which were absent from the maps generated from either the M_r = 46 000 or 48 000 targets. These data suggest that the M_r = 46 000 and 48 000 substrates are closely related proteins, however the relationship between the M_r = 42 000 toxin-specific substrate and the larger peptides remains to be established. The relative patterns of fat cell membrane labelling by cholera toxin in the presence of [³²P]NAD⁺     

Pertussis toxin-mediated ADP-ribosylation of rabbit luteal Gi uncouples enkephalin inhibition of adenylyl cyclase

1. Some of the actions of pertussis toxin on the rabbit luteal adenylyl cyclase system were analyzed. 2. Incubation of luteal membranes with pertussis toxin and [32P]NAD resulted in the [32P]ADP-ribosylation of a 40,000 Da protein that is distinct from the proteins ADP-ribosylated by cholera toxin. 3. Pertussis toxin specific [32P]ADP-ribosylation was time-dependent and dependent upon the concentration of pertussis toxin present during the incubation. 4. Pertussis toxin mediated [32P]ADP-ribosylation was enhanced by ATP, ADP, adenylyl imidodiphosphate, GTP, guanosine-5'-O-(2-thiodiphosphate), guanosine-5'-O-(3-thiotriphosphate), and NaF but not AMP or guanylyl imidodiphosphate [GMP-P(NH)P]. 5. Treatment of luteal membranes with NAD and pertussis toxin prevents GTP and enkephalin but not GMP-P(NH)P mediated inhibition of forskolin stimulated adenylyl cyclase, demonstrating the existence of a functional Gi in the rabbit corpus luteum.     

A-protein catalyzes the ADP-ribosylation of G-protein from cow rod outer segments

A 20-kilodalton adenosine nucleotide-binding protein (A-protein) extracted from rod outer segments is shown to catalyze the cholera toxin-mediated ADP-ribosylation of GTP-binding protein (G-protein) from the outer segment. Radiolabel from [adenylate-32P] NAD+ was associated specifically with both the alpha-subunit of G-protein and with A-protein in the presence of activated cholera toxin. In the absence of added A-protein, G-protein appears to undergo ADP-ribosylation at a slower rate. In the absence of G-protein, A-protein was found to be labeled following incubation with [adenylate-32P]NAD+ and cholera toxin. In the presence of G-protein, a light-dependent component of A-protein labeling was observed. A-protein is a labile component of rod outer segments and has an affinity for ADP. The findings suggest that A-protein may act as an ADP-ribosyltransferase in the cholera toxin-mediated ADP-ribosylation of G-protein.     

ADP-ribosylation of microtubule proteins as catalyzed by cholera toxin

Incubation of purified rat brain tubulin with cholera toxin and radiolabeled [32P] or [8-3H]NAD results in the labeling of both alpha and beta subunits as revealed on sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE). Treatment of these protein bands with snake venom phosphodiesterase resulted in quantitative release of labeled 5'-AMP, respectively labeled with the corresponding isotope. Two-dimensional separation by isoelectric focusing and SDS-PAGE of labeled and native tubulin revealed that labeling occurs at least in four different isotubulins. The isoelectric point of the labeled isotubulins was slightly lower than that of native purified tubulin. This shift in mobility is probably due to additional negative charges involved with the incorporation of ADP-ribosyl residues into the tubulin subunits. SDS-PAGE of peptides derived from [32P]ADP-ribosylated alpha and beta tubulin subunits by Staphylococcus aureus protease cleavage showed a peptide pattern identical with that of native tubulin. Microtubule-associated proteins (MAP1 and MAP2) of high molecular weight were also shown to undergo ADP-ribosylation. Incubation of permeated rat neuroblastoma cells in the presence of [32P]NAD and cholera toxin results in the labeling of only a few cell proteins of which tubulin is one of the major substrates.     

ADP-ribosylation of transducin by pertussis toxin

Transducin, the guanyl nucleotide-binding regulatory protein of retinal rod outer segments that couples the photon receptor, rhodopsin, with the light-activated cGMP phosphodiesterase, can be resolved into two functional components, T alpha and T beta gamma. T alpha (39 kDa), which is [32P]ADP-ribosylated by pertussis toxin and [32P]NAD in rod outer segments and in purified transducin, was also labeled by the toxin after separation from T beta gamma (36 kDa and approximately 10 kDa); neither component of T beta gamma was a pertussis toxin substrate. Labeling of T alpha was enhanced by T beta gamma and was maximal at approximately 1:1 molar ratio of T alpha : T beta gamma. Limited proteolysis by trypsin of T alpha in the presence of guanyl-5'-yl imidodiphosphate (Gpp(NH)p) resulted in the sequential appearance of proteins of 38 and 32 kDa. The amino terminus of both 38- and 32-kDa proteins was leucine, whereas that of T alpha could not be identified and was assumed to be blocked. The 32-kDa peptide was not a pertussis toxin substrate. Labeling of the 38-kDa protein was poor and was not enhanced by T beta gamma. Trypsin treatment of [32P]ADP-ribosyl-T alpha produced a labeled 37-38-kDa doublet followed by appearance of radioactivity at the dye front. It appears, therefore, that, although the 38-kDa protein was poor toxin substrate, it contained the ADP-ribosylation site. Without rhodopsin, labeling of T alpha (in the presence of T beta gamma) was unaffected by Gpp(NH)p, guanosine 5'-O-(thiotriphosphate) (GTP gamma S), GTP, GDP, and guanosine 5'-O-(thiodiphosphate) (GDP beta S) but was increased by ATP. When photolyzed rhodopsin and T beta gamma were present, Gpp(NH)p and GTP gamma S decreased [32P]ADP-ribosylation by pertussis toxin. Thus, pertussis toxin-catalyzed [32P]ADP-ribosylation of T alpha was affected by nucleotides, rhodopsin and light in addition to T beta gamma. The amino terminus of T alpha, while it does not contain the pertussis toxin ADP-ribosylation site, appeared critical to its reactivity.     

Regulation of acetyl-CoA carboxylase by ADP-ribosylation

Because of certain similarities between acetyl-CoA carboxylase (ACC) and tubulin, and the recent demonstration of the ADP-ribosylation of tubulin by cholera toxin, we have investigated a potential role for ADP-ribosylation in the regulation of ACC activity. Incubation of purified rat liver ACC with cholera toxin in the presence of millimolar concentrations of [adenylate-32P]NAD results in a time-dependent incorporation of ADP-ribose into ACC of greater than 2 mol/mol of enzyme subunit, accompanied by a marked inactivation of enzyme activity. This effect is not mimicked by pertussis toxin, ADP-ribose, or ribose 5-phosphate. Incubation of labeled ACC with snake venom phosphodiesterase and alkaline hydrolysis release 32P-products tentatively identified by high-performance liquid chromatography as 5'-[32P]AMP and [32P]ADP-ribose, respectively. These data are consistent with a mono-ADP-ribosylation of ACC catalyzed by cholera toxin. Phosphodiesterase treatment of inactivated ACC partially restores enzyme activity. The effects of ADP-ribosylation of ACC are expressed both as a decrease in the enzyme Vmax and as an increase in the apparent Ka for citrate. These results suggest that ACC might be a substrate for endogenous ADP-ribosyltransferases and that this covalent modification could be an important regulatory mechanism for the modulation of fatty acid synthesis in vivo.     

Stimulation by cholera toxin of ADP-ribosylation of membrane proteins, adenylate cyclase and insulin release in pancreatic islets

In rat pancreatic islet membranes exposed to [alpha-32P]NAD, cholera toxin stimulated the labelling of three peptides with Mr close to 22 000, 42 000 and 48 000, respectively. In the islets, the toxin-stimulated ADP-ribosylation of the heavy form of the Ns alpha-subunit predominated over that of the light form, in mirror image of the situation found in the exocrine pancreas. When intact islets were preincubated with cholera toxin, the adenylate cyclase activity of a subcellular particulate fraction was increased. The responsiveness of adenylate cyclase to GTP was also augmented, but that to NaF was decreased. In intact islets, the production of cyclic AMP and the glucose-stimulated release of insulin were also enhanced after pretreatment with cholera toxin. These findings reveal the presence in pancreatic islets of the guanyl nucleotide regulatory protein of adenylate cyclase, with an unusual predominance of the heavy form of the Ns alpha-subunit.     

A novel approach to detect toxin-catalyzed ADP-ribosylation in intact cells: its use to study the action of Pasteurella multocida toxin

Certain microbial toxins are ADP-ribosyltransferases, acting on specific substrate proteins. Although these toxins have been of great utility in studies of cellular regulatory processes, a simple procedure to directly study toxin-catalyzed ADP-ribosylation in intact cells has not been described. Our approach was to use [2-3H]adenine to metabolically label the cellular NAD+ pool. Labeled proteins were then denatured with SDS, resolved by PAGE, and detected by flurography. In this manner, we show that pertussis toxin, after a dose-dependent lag period, [3H]-labeled a 40-kD protein intact cells. Furthermore, incubation of the gel with trichloroacetic acid at 95 degrees C before fluorography caused the release of label from bands other than the pertussis toxin substrate, thus, allowing its selective visualization. The modification of the 40-kD protein was ascribed to ADP-ribosylation of a cysteine residue on the basis of inhibition of labeling by nicotinamide and the release of [3H]ADP-ribose from the labeled protein by mercuric acetate. Cholera toxin catalyzed the [3H]-labeling of a 46-kD protein in the [2-3H]adenine-labeled cells. Pretreatment of the cells with pertussis toxin before the labeling of NAD+ with [2-3H]adenine blocked [2-3H]ADP-ribosylation catalyzed by pertussis toxin, but not that by cholera toxin. Thus, labeling with [2-3H]adenine permits the study of toxin-catalyzed ADP-ribosylation in intact cells. Pasteurella multocida toxin has recently been described as a novel and potent mitogen for Swiss 3T3 cell and acts to stimulate the phospholipase C-mediated hydrolysis of polyphosphoinositides. The basis of the action of the toxin is not known. Using the methodology described here, P. multocida toxin was not found to act by ADP-ribosylation.     

ADP-ribosylation of thylakoid membrane polypeptides by cholera toxin is correlated with inhibition of thylakoid GTPase activity and protein phosphorylation

Incubation of pea thylakoid membranes with [32P]-NAD+ in the presence of cholera toxin resulted in the [32P]-ADP-ribosylation of a 60 kDa thylakoid membrane polypeptide. When ATP was included in the incubation mixture, a 29 kDa polypeptide was also labelled. In the absence of electron transfer cofactors or inhibitors, the extent of labelling depended on whether the membranes were preincubated in the light or dark and also on the developmental stage of the leaves used for thylakoid isolation. Irrespective of the latter, the strongest labelling was observed when DCMU was present in the light. After pretreatment of the thylakoid membranes with cholera toxin plus NAD+ under the same conditions, light-stimulated GTPase activity and protein phosphorylation were inhibited. The extent of inhibition for both processes appeared to be correlated with the amount of [32P]-ADP-ribosylation found when [32P]-NAD+ was included in the pretreatment mixture. The data presented are fully consistent with the 60 and 29 kDa polypeptides functioning as thylakoid membrane associated guanine nucleotide binding regulatory proteins.     

Amino acid-specific ADP-ribosylation. Sensitivity to hydroxylamine of [cysteine(ADP-ribose)]protein and [arginine(ADP-ribose)]protein linkages

Hydroxylamine stability has been used to classify (ADP-ribose)protein bonds into sensitive and resistant linkages, with the former representing (ADP-ribose)glutamate, and the latter, (ADP-ribose)arginine. Recently, it was shown that cysteine also serves as an ADP-ribose acceptor. The hydroxylamine stability of [cysteine([32P]ADP-ribose)]protein and [arginine([32P] ADP-ribose)]protein bonds was compared. In transducin, pertussis toxin catalyzes the ADP-ribosylation of a cysteine residue, whereas choleragen (cholera toxin) modifies an arginine moiety. The (ADP-ribose)cysteine bond formed by pertussis toxin was more stable to hydroxylamine than was the (ADP-ribose)arginine bond formed by choleragen. The (ADP-ribose)cysteine bond apparently represents a third class of ADP-ribose bonds. Pertussis toxin ADP-ribosylates the inhibitory guanyl nucleotide-binding regulatory protein (Gi) of adenylate cyclase, whereas choleragen modifies the stimulatory guanyl nucleotide-binding regulatory protein (Gs). These (ADP-ribose)protein linkages are identical in stability to those formed in transducin by the two toxins, consistent with the probability that cysteine and arginine are modified in Gi and Gs, respectively. Bonds exhibiting differences in hydroxylamine-stability were found in membranes from various non-intoxicated mammalian cells following incubation with [32P]NAD, which may reflect the presence of endogenous NAD:protein-ADP-ribosyl-transferases.     

NADP efficiently inhibits endogenous but not pertussis toxin-catalyzed covalent modification of membrane proteins incubated with NAD

Incubation of membranes of human erythrocytes and platelets but not of human neutrophils with [32P]NAD leads to covalent modification of various membrane proteins and of added albumin. In membranes of all three cell types, pertussis toxin (PT), in the presence of NAD, specifically labelled a 40 kDa peptide, i.e. the alpha-subunit of a guanine nucleotide-binding protein. This effect of PT was slightly reduced by NADP, whereas modification of other membrane proteins and of albumin was largely suppressed, independent of whether PT was present or not. Labelling of cytosolic proteins in the presence of NAD was marginal; only in neutrophil cytosol, PT modified a 40 kDa peptide. Membranes of erythrocytes and platelets exhibited NAD-degrading activity, which was inhibited by NADP. The data suggest a high substrate specificity of PT for NAD. Inhibition of endogenous enzymes by NADP may prove useful for the evaluation of PT substrates.     

Identification of the NADH-binding subunit of NADH-ubiquinone oxidoreductase of Paracoccus denitrificans

The NADH dehydrogenase complex isolated from Paracoccus denitrificans is composed of approximately 10 unlike polypeptides and contains noncovalently bound FMN, non-heme iron, and acid-labile sulfide [Yagi, T. (1986) Arch. Biochem. Biophys. 250, 302-311]. When the Paracoccus NADH dehydrogenase complex was irradiated by UV light in the presence of [adenylate-32P]NAD, radioactivity was incorporated exclusively into one of three polypeptides of Mr approximately 50,000. Similar results were obtained when [adenylate-32P]NADH was used. The labeling of the Mr 50,000 polypeptide was diminished when UV irradiation of the enzyme with [adenylate-32P]NAD was performed in the presence of NADH, but not in the presence of NADP(H). The labeled polypeptide was isolated by preparative sodium dodecyl sulfate gel electrophoresis and was shown to cross-react with antiserum to the NADH-binding subunit (Mr = 51,000) of bovine NADH-ubiquinone oxidoreductase. Its amino acid composition was also very similar to that of the bovine NADH-binding subunit. These chemical and immunological results indicate that the Mr 50,000 polypeptide is an NADH-binding subunit of the Paracoccus NADH dehydrogenase complex.     

Pertussis toxin-catalyzed ADP-ribosylation of transducin. Cysteine 347 is the ADP-ribose acceptor site

Pertussis toxin catalyzes the transfer of ADP-ribose from NAD to the guanine nucleotide-binding regulatory proteins Gi, Go, and transducin. Based on a partial amino acid sequence for a tryptic peptide of ADP-ribosylated transducin, asparagine had been characterized as the site of pertussis toxin-catalyzed ADP-ribosylation. Subsequently, cDNA data for the alpha subunit of transducin indicated that the putative asparagine residue was, in fact, not present in the protein. To determine the amino acid that served as the ADP-ribose acceptor, radiolabel from [adenine-U-14C]NAD was incorporated, in the presence of pertussis toxin, into the alpha subunit of transducin (0.3 mol/mol). An ADP-ribosylated, tryptic peptide was purified and fully sequenced by automated Edman degradation. The amino acid sequence, Glu-Asn 343-Leu-Lys-Asp 346-X-Gly 348-Leu-Phe, corresponds to the cDNA sequence coding the carboxyl-terminal nonapeptide, Glu 342-Phe 350, which includes by cDNA sequence cysteine at position 347. Neither Asn 343 nor Asp 346 appeared to be modified; residue 347 adhered to the sequencing resin. Cysteine, the missing residue, was eluted from the sequencing resin with acetic acid along with 76% of the peptide-associated radioactivity, half of which, presumably ADP-ribosylcysteine, eluted from an anion exchange column between NAD and ADP-ribose; the other half had a retention time corresponding to 5'-AMP. We conclude that Cys 347 and not Asn 343 or Asp 346 is the site of pertusis toxin-catalyzed ADP-ribosylation in transducin.     

Neuropeptide y inhibits pertussis toxin-catalyzed ADP-ribosylation in bovine adrenal chromaffin cell membranes

Evidence is presented that the neuropeptide Y receptor is directly coupled to an inhibitory G protein existing in cultured bovine adrenal chromaffin cell membranes. Pertussis toxin catalyzes the [³²P]ADP-ribosylation of a 41 kDa plasma membrane protein. 5′-Guanylylimidodiphosphate inhibited the [³²P]ADP labelling of this protein in a dose-dependent manner whereas GTP had no effect. Preincubation of the plasma membranes with high concentrations of neuropeptide Y followed by a brief exposure to a low concentration of 5′-guanylylimidodiphosphate significant;y inhibited ADP-ribosylation beyond that observed with 5′-guanylimidodiphosphate alone. These result suggest that the neuropeptide Y receptor in bovine adrenal chromaffin cells is directly coupled to a 41 kDa PTX substrate (presumably the α subunit of an inhibitory G protein).     

Cholera toxin affects nuclear ADP-ribosylation in GH1 cells

Incubation of GH1 cells with cholera toxin for 24 h inhibits [32P]ADP-ribose incorporation into histones and non-histone nuclear proteins by more than 50%. The toxin produces a generalized decrease of incorporation into all protein acceptors and into the poly(ADP-ribosyl)ated components excised from chromatin after micrococcal nuclease digestion. The cellular levels of NAD were also decreased (40 to 80%) after treatment with cholera toxin. The inhibition of poly(ADP-ribosyl)ation is preceded by an increase of [32P]ADP-ribose incorporation, since incubation with the toxin for 3 h caused an increase instead of a decrease of incorporation. Incubation with dibutyryl cyclic AMP for 24 h also inhibited nuclear poly(ADP-ribosyl)ation, thus showing that the effect of cholera toxin might be mediated by cyclic AMP.     

Amino acid-specific ADP-ribosylation

[adenine-U-14C]ADP-ribose-agmatine and [adenine-U-14C ))ADP-ribose-histone were synthesized by an NAD:arginine ADP-ribosyltransferase from [14C]NAD and agmatine and histone, respectively. The pseudo-first order rate constants for breakdown of the two components either in 0.4 N NaOH or in 0.4 M neutral hydroxylamine were identical. Hydroxylamine treatment of [14C]ADP-ribose-agmatine or [32P]ADP-ribose-histone yielded a single radioactive product which was separated by high pressure liquid chromatography and identified as ADP-ribose-hydroxamate by the formation of a ferric chloride complex. Hydrolysis of ADP-ribose-hydroxamate with snake venom phosphodiesterase resulted in the formation of 5'-AMP, consistent with the presence of a pyrophosphate bond. Incubation of ADP-ribose-[14C]agmatine, synthesized by the ADP-ribosyltransferase from NAD and [14C]agmatine, with 0.4 M neutral hydroxylamine resulted in the release of [14C]agmatine rather than phosphoribosyl[14C]agmatine. In addition, neither NAD nor ADP-ribose reacts with hydroxylamine; i.e. there was no evidence of nucleophilic attack by hydroxylamine at the pyrophosphate bond. The ADP-ribosyl-protein linkage formed by the NAD:arginine ADP-ribosyltransferase is considerably more stable to hydroxylamine than is the ADP-ribose-glutamate bond. The presence of ADP-ribose-arginine and ADP-ribose-glutamate synthesized by the ADP-ribosyltransferase and poly(ADP-ribose) synthetase, respectively, may be the chemical basis for the "hydroxylamine-stable" and "hydroxylamine-labile" bonds described by Hilz (Hilz, H. (1981) Hoppe-Seyler's Z. Physiol. Chem. 362, 1415-1425).     

NADP improves the efficiency of cholera toxin catalyzed ADP-ribosylation in liver and heart membranes

Guanine nucleotide binding proteins (G-proteins) can be identified by their ability to be ADP-ribosylated using [32P]NAD as the substrate and bacterial toxins as catalysts. This labelling, when performed in liver and sarcolemma membrane preparations, can be complicated by competing enzymes which degrade NAD, making it unavailable to participate in the desired reaction. The addition of NADP in reaction mixtures markedly slows the degradation of NAD, but does not compete with NAD in cholera toxin labelling of stimulatory G-protein. The efficiency of cholera toxin labelling is improved to the extent that saturation curves may be constructed, allowing the quantitation of ADP-ribosylation sites in membranes.     

Identification of the NADH-binding subunit of energy-transducing NADH-quinone oxidoreductase (NDH-1) of thermus thermophilus HB-8

The energy-transducing NADH--quinone oxidoreductase (NDH-1) isolated from Thermus thermophilus HB-8 is composed of approximately 10 unlike polypeptides and contains noncovalently bound FMN and at least three iron-sulfur clusters [Yagi, T., Hon-nami, K., and Ohnishi, T. (1988) Biochemistry 27, 2008-2013]. When NDH-1 of T. thermophilus HB-8 was irradiated by short UV light in the presence of [adenylate-32P]NADH or [adenylate-32P]NAD, radioactivity was incorporated into a single polypeptide of Mr 47,000. The labeling of the Mr 47,000 polypeptide was diminished when UV irradiation of the enzyme complex with [adenylate-32P]NAD was carried out in the presence of NADH or deamino-NADH which act as substrates for the NDH-1, but not in the presence of NADP(H) or AMP which act neither as substrates nor as competitive inhibitors. These results strongly suggest that the Mr 47,000 polypeptide is an NADH-binding subunit of the NDH-1 of T. thermophilus HB-8.