Pertussis Toxin-catalyzed ADP-ribosylation of Transducin

Pertussis toxin catalyzes the transfer of ADP-ribose from NAD to the guanine nucleotide-binding regula- tory proteins Gi, Go, and transducin. Based on a partial amino acid sequence for a tryptic peptide of ADP- ribosylated transducin, asparagine had been charac-terized as the site of pertussis toxin-catalyzed ADP- ribosylation. Subsequently, cDNA data for the a subunit of transducin indicated that the putative aspara- gine residue was, in fact, not present in the protein. To determine the amino acid that served as the ADP- ribose acceptor, radiolabel from [ader~ine-U-~~C]NAD was incorporated, in the presence of pertussis toxin, into the a subunit of transducin (0.3 mol/mol). An ADP-ribosylated, tryptic peptide was purified and fully se- quenced 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 cod-ing the carboxyl-terminal nonapeptide, Glu 342-Phe 350, which includes by cDNA sequence cysteine at position 347. Neither

* The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
'The abbreviations used are: Gi, the guanyl nucleotide-binding protein coupling inhibitory receptors to adenylate cyclase; G. , the guanyl nucleotide-binding protein coupling stimulatory receptors to adenylate cyclase; Go, a guanyl nucleotide-binding protein of unknown function; Gh, a subunit of Gi; Go,, a subunit of Go. protein coupling inhibitory receptors to adenylate cyclase, and Go (8), a GTP-binding protein of unknown function.
The substrate amino acid ADP-ribosylated varies from toxin to toxin. Diphtheria toxin transfers ADP-ribose to a modified histidine residue which appears to be restricted to elongation factor 2 (18). Cholera toxin ADP-ribosylates an arginine residue (14,19). It was reported that pertussis toxin modifies an asparagine (13), but this conclusion has been questioned in the light of newer data on the cDNA sequence of 01 transducin (10,20). Because understanding the molecular mechanism of toxin action has been instrumental in t h e identification among higher organisms of native mono-ADPribosyltransferases similar to cholera toxin (21-24) and diphtheria toxin (25), we sought to ascertain the ADP-ribose acceptor site in the pertussis toxin-catalyzed ADP-ribosylation of transducin.
Purification of Transducin-Rod outer segments were prepared from fresh bovine eyes according to an abbreviated Fung and Stryer procedure (26) in which the 40,000 X g crude rod outer segment pellet was washed twice and stored frozen at -70 "C. All operations were carried on in room light. Phenylmethanesulfonyl fluoride was omitted from buffers. Transducin was purified from rod outer segments according to Kuhn (27) with 100 p~ GTP. Protein was assayed according to Lowry et al. (28) with bovine serum albumin as standard.
Automated Edman Degradation-Automated Edman degradation was performed in an Applied Biosystem gas-phase sequencer 470A. Degradation was carried out in the presence of Polybrene (34). Phenylthiohydantoins were identified by high pressure liquid chromatography using a Zorbax-octadecylsilane column and an acetonitrile gradient in 0.025 M sodium acetate, pH 5.3 (39, or in some cases hy back hydrolysis (36).
High Performance Liquid Chromatography-Fifty-microliter samples were injected onto a DuPont SAX column (0.46 X 25 cm) in a mobile phase of 50 mM sodium phosphate, pH 4.4, at a flow rate of 1 ml/min. Absorbance was monitored at 254 nm and l-ml fractions were collected for liquid scintillation counting.  I  Purification of the ADP-ribosylated tryptic peptide from transducin  Transducin (21 nmol), ADP-ribosylated as described, was dried under vacuum and resuspended in 1 ml of 100 mM NH4HCOs and digested for 18 h at 37 "C with 50 pg of trypsin added at zero time and again after 6 h. DEAE chromatography (0.76 X 15 cm (LKB), 50 mM Tris acetate, pH 7.5 (25 "C), 0-600 mM NazS04/30 min, 2 ml/ min) of the vacuum-dried digest applied in Tris-acetate yielded a single .peak of radioactivity eluting at 200 mM Na2S04. Radioactive fractions were pooled, diluted in 100 mM glycine, pH 9/100 mM NaCl/ 10 mM MgC1, and applied to 2 ml of phenylboronate polyacrylamide (AG 601, Bio-Rad). After 20 bed volumes of wash, the radioactivity was eluted with 50 mM HCOONH4, pH 4.7. The eluate was pooled, vacuum-dried, and resuspended in water for reverse-phase chromatography on an Ultrapore RPSC (Altex) C3 column in a gradient of 5-35% acetonitrile/0.05% trifluoroacetic acid/30 min, 2 ml/min. The major peak of radioactivity (91%) was used for amino acid and sequence analysis.

RESULTS
Radiolabel from [~denine-U-~~C]NAD incubated with purified transducin in the presence of pertussis toxin was incorporated into the a subunit of transducin as determined by autoradiography (data not shown). Maximal labeling (which was achieved in 2 h) was 0.3 mol/mol of transducin. Trypsin digestion and subsequent purification, as described in Table   I, yielded a peptide the amino acid content of which is summarized in Table 11. The amino acid sequence of this peptide as shown in Table I11 is Glu-Asn-Leu-Lys-Asp-X-Gly-Leu-Phe. Analysis of phenylthiohydantoins by high pressure liquid chromatography and by amino acid analysis following performic acid oxidation and back hydrolysis indicated a blank in position 6. If phenylthiohydantoin-cysteine had been present, cysteic acid would have been detected by the amino acid analysis. Seventy-six per cent of the radioactivity was eluted from the sequencing resin with 5% acetic acid; 24% of the recovered radioactivity was evenly distributed through 10 cycles. A sample of the resin eluate was analyzed by anionexchange chromatography as shown in Fig. 1. Fifty-seven per cent of the radioactivity eluted like 5'-AMP (RT = 6 min) and 43% (RT = 9 min) eluted between NAD (RT = 7.1 min) and ADP-ribose (RT = 9.8 min). The remainder of the sample was   subjected to performic acid oxidation followed by back hydrolysis. Amino acid analysis of the hydrolysate revealed cysteic acid and contaminants.

DISCUSSION
As described previously, pertussis toxin catalyzed transfer of ADP-ribose from NAD into the a subunit of holotransducin (13). Amino acid analysis and complete Edman degradation of a purified tryptic peptide yielded the sequence Glu-Asn-Leu-Lsy-Asp-X-Gly-Leu-Phe which corresponds to the car- cent of the recovered peptide-associated radioactivity was eluted from the sequencing resin along with cysteine, half as free 5'-AMP and half as a species with a mobility between NAD and ADP-ribose on anion-exchange chromatography. We conclude that the latter peak is ADP-ribosylcysteine which because of its strong negative charge was retained on the sequencing resin and thus not detected in cycle six.
These data contradict an earlier report of pertussis toxincatalyzed ADP-ribosylation of asparagine in transducin (13). The amino acid compositions and sequence analyses for the tryptic peptide (Glu-Asn-Leu-Lys-Asn-Gly-Leu-Phe) in Ref.
13 precluded the possibility of an ADP-ribosylcysteine linkage. The amino acid sequence reported here is consistent with the recently published cDNA sequence for a transducin (20, 37-39). In retrospect, since ADP-ribosylation of a single critical cysteine is sufficient to disrupt inhibitory coupling, the effect of N-ethylmaleimide, which disrupts inhibitory receptor-Gi-adenylate cyclase but not stimulatory receptor-G.-adenylate cyclase coupling, might be more specific than originally supposed (40,41).
Judged by the paucity of reports in the literature, glycosylcysteine linkages occur infrequently in nature. A galactosylcysteine-containing peptide has been isolated from human urine and a glucosylcysteine-containing peptide from human erythrocyte membranes (42,43). Nonenzymatic condensation of cysteine with arabinose or xylose, as well as a number of hexoses, has also been reported (44). The nature of the pertussis toxin-catalyzed linkage, presumably a thioacetal between the cysteine S and C-l of ribose, will be of further interest.
Among the ADP-ribosyltransferases whose sites of action have been ascertained, pertussis toxin, transferring to a cysteine, is unique. Glutamate is the site of nuclear poly-ADPribosylation (45-47). Cholera toxin, Escherichia coli heatlabile enterotoxin, and the mono-ADP-ribosyltransferases native to turkey erythrocytes and chicken liver nuclei modify a substrate arginine (14, 19, 21-23); diphtheria toxin and apparently, Pseudomonas exotoxin A and an enzyme native to bovine liver transfer ADP-ribose to diphthamide, a modified histidine apparently restricted to elongation factor 2 (17, 18, 25). The limited occurrence of diphthamide may explain, in part, the specificity of diphtheria toxin for elongation factor. The determinants of specificity for cholera toxin, E. coli heatlabile enterotoxin, and the vertebrate transferases are unclear since all of these will transfer ADP-ribose to arginine and other guanidino compounds in solution as well as to their native substrates (19, 21-24). The degree of specificity of pertussis toxin for the ADP-ribose acceptor site remains to be seen. Given the structural similarities among the pertussis toxin substrates, Gi,, Go,, and a transducin, and substantial sequence homologies between a transducin and Go,, including the carboxyl terminus of both, where pertussis toxin-catalyzed ADP-ribosylation occurs (10,13), cysteine likely serves as the ADP-ribose acceptor in the pertussis toxin-catalyzed modification of all three proteins. Use of this sequence, or less depending upon active site requirements, may be helpful in a search for NAD:cysteine ADP-ribosyltransferases native to higher organisms as has been done with NAD:arginine ADP-ribosyltransferases of turkey erythrocytes, chicken liver nuclei, and rabbit skeletal muscle (21-24).