Requirements for cholera toxin-dependent ADP-ribosylation of the purified regulatory component of adenylate cyclase.

The requirements for cholera toxin-catalyzed ADP ribosylation of the purified regulatory component of adenylate cyclase are described. In addition to the toxin, this reaction is dependent on or is facilitated by NAD, GTP, phospholipid, and a factor found associated with plasma membranes from several sources. Factor activity is heat-labile and protease-sensitive but is unaffected by treatment with N-ethylmaleimide. Gel filtration indicates that the factor behaves as a monodisperse species with a Stokes radius of 3.2 nm. The factor thus appears to be a protein that is distinct from any of the known components of adenylate cyclase. Factor activity was also detected in the cytoplasm of S49 cells. The cytoplasmic factor was smaller (Stokes radius = 2.0 nm) than the membrane-derived factor, and it was inactivated in the presence of sodium cholate. The initial rate of activation of the regulatory component of adenylate cyclase by toxin was found to be linearly related to the amount of factor present in the reaction. This has allowed the quantitation and partial purification (33-fold from detergent extracts) of the factor from turkey erythrocyte membranes.

NAD (which serves as the ADP-ribosyl donor), and GTP (4). GTP may serve several functions, as described by Nakaya et al. (5), including acting as a cofactor in the toxin-dependent activation reaction and stabilizing the ADP-ribosylated product. In addition, assay of GTP-dependent adenylate cyclase activity serves as a sensitive index of the extent of covalent modification of G/F. It is not known if the occupation of a single GTP binding site is sufficient for observation of all of these effects of guanine nucleotides.
This laboratory has purified the regulatory component of adenylate cyclase from membranes of rabbit liver (3,6) and turkey erythrocytes (7). Rabbit liver G/F consists of three peptides with apparent M , = 35,000, 45,000, and 52,000. Turkey erythrocyte G/F lacks the 52,000-dalton peptide. Only the 52,000-and 45-000-dalton peptides are substrates for toxindependent ADP ribosylation (3, 6, 7). The requirements for toxin-dependent ADP ribosylation of pure G/F are described below; these include the presence of a membrane-derived or a soluble protein factor for optimal modification of the regulatory protein.

EXPERIMENTAL PROCEDURES
Materials-Cholera toxin and L-a-dimyristoyl phosphatidylethanolamine were purchased from Calbiochem. L-a-dimyristoyl phosphatidylcholine, L-a-dipalmitoyl phosphatidylethanolamine, phosphatidylserine, phosphatidylcholine, and cholesterol were purchased from Sigma. Lipid suspensions were sonicated in 20 mM sodium Hepes (pH 8) and 1 mM EDTA for 15 to 20 min prior to use. Trypsin, chymotrypsin, N-ethylmaleimide, and ATP were also obtained from Sigma. [u-"P]ATP was synthesized according to the procedure of Johnson and Walseth (8). This radiolabeled nucleotide was used both for adenylate cyclase assays and for the synthesis of ["'PINAD. NAD labeled with 32P in the AMP moiety was synthesized by the method of Cassel and Pfeuffer (9). G/F was purified from rabbit liver and turkey erythrocytes as described by Sternweis et al. (6) and Hanski et al. (7). respectively. All protein determinations were performed by the method of Schaffner and Weissmann (10).
S49 Lymphoma Membranes and Extracts-Clones of S49 lymphoma cells that were utilized and the methods for their culture have been described previously (11). The cyc-variant of the S49 cell, which is essentially devoid of Mg-ATP-dependent adenylate cyclase activity, contains the catalytic subunit of the enzyme but is deficient in the activity of G/F (12). Plasma membranes from these cells were purified according to the method of Ross et al. (11). Cholate extracts of S49 plasma membranes were prepared as described by Sternweis and Gilman (13), except that solubilization was performed at an initial protein concentration of 10 mg/ml in 20 mM Tris-HC1 (pH 7.5), 1 mM EDTA, 100 mM NaC1, 1% sodium cholate.
G/F and Adenylate Cyclase Assay-G/F was assayed by its ability to reconstitute adenylate cyclase activity in cyc-membranes. Samples of G/F (15 pl) were added to 25 p1 (50 p g of protein) of cycmembranes. Reconstitution and assay then proceeded as described by Sternweis and Gilman (13) in the presence of the activators indicated in the text. Conditions were chosen such that the rate of production of cyclic AMP in the assay was constant and linearly related to the concentration of G/F. Turkey Erythrocyte Membranes and Extracts-Fresh turkey blood was collected and stored at 4 "C for up to 3 days in the presence of 4.75 mM EDTA and 0.25 mM ethylene glycol bis(P-aminoethy1 ether)-N,N,N',N'-tetraacetic acid to prevent coagulation. Membranes were purified as described by Hanski et al. (7), and membrane proteins were solubilized with 1.25% sodium cholate in 25 mM Tris-HC1 (pH 8), 1 mM EDTA, 0.25 M sucrose, 1 mM dithiothreitol, 10 pM phenylmethylsulfonyl fluoride, 0.7 M NaCl at an initial protein concentration of 10 mg/ml. Partial Purification of Factor from Turkey Erythrocyte Menbranes-The membrane-bound factor (see below) from turkey eryth-rocytes was purified about 33-fold from cholate extracts of membranes. The chromatographic steps utilized were identical with the first two steps in the purification of turkey erythrocyte G/F (7). Factor activity eluted from the DEAE-Sephacel column in two peaks, the f m t a t about 80 mM NaCl and the second, which co-eluted with G/F, at about 120 mM NaCI. The first peak of factor activity was further fractionated on Ultrogel AcA-34, from which it eluted with a KO of approximately 0.7. The peak fractions were pooled and stored at -80 "C. When this partially purified factor is used in the labeling reaction described below, only G/F and cholera toxin are radioactively labeled. The recovery of factor activity after these two steps was approximately 178, while recovery of protein was 0.5%. Preparation of Factor from the Cytoplasm of cyc-Cells-Cytoplasmic fractions from cyc-cells were prepared by suspension of packed cells in an equal volume of 20 m~ Hepes (pH 8.0), 2 mM MgCh, 1 m~ EDTA at 4 "C. The cells were homogenized with 20 strokes of a Dounce homogenizer, and the homogenate was centrifuged (4 "C) for 60 min at 200,000 X g. The supernatant was then collected and centrifuged again as before.
Assay of Factor by the Activation of G/F-Cholera enterotoxin was activated by incubation of toxin (0.5 mg/ml) in potassium phosphate (25 mM, pH 8) and dithiothreitol (20 mM) at 37 "C for 15 min. Unless otherwise specified, incubation of purified G/F with activated toxin was then performed in siliconized tubes by the addition of 5 pl of purified G/F (150-900 ng, in a solution containing not more than 0.1% Lubrol PX) to 5 pI of factor (in not more than 1.25% cholate) and 45 pl of a mixture of other reagents. The reaction was then begun by the addition of 5 pl of activated toxin. The final reaction volume was 60 pl and contained 250 m~ potassium phosphate (pH 7.5), 10 m~ thymidine, 30 p~ GTP, 1 m~ MgCL, 0.1 m~ EDTA, 3 mM dimyristoyl phosphatidylcholine, 100 p~ NAD. Incubation was at 30 "C for the indicated times. The reaction was stopped by a 20-fold dilution into ice-cold 20 m~ Tris-HC1 (pH 7.5), 1 mM EDTA, 0.18 Lubrol PX. Reconstituted adenylate cyclase activity was determined as described (13) using GTP (100 p~) as the stimulator. When the concentrations of detergents were held constant, the rate of activation of G/F by toxin was proportional to the amount of factor added (see below).
Radioactive Labeling of G/F-ADP ribosylation of G/F was monitored by autoradiography of labeled protein after incubation with ["*PINAD. The reaction of NAD with G/F was performed as described above, except the concentration of NAD was reduced to 5-20 p~ (3,000-30,000 cpm/pmol). Samples for autoradiography were diluted 5-fold with Laemmli's sample buffer (14) and were then boiled for 5 min before being applied to 0.75-mm slab gels of 11% acrylamide. One-dimensional sodium dodecyl sulfate-polyacrylamide gel electrophoresis was performed as described by Laemmli (14). The gels were stained with Coomassie brilliant blue R-250, destained, and dried on filter paper. Autoradiography was performed as described previously (15).

RESULTS AND DISCUSSION
The ability of activated cholera toxin to catalyze the ADP ribosylation of purified rabbit liver G/F was found to be very poor. The reaction occurred extremely slowly, as judged by either the incorporation of radioactivity from ["*PINAD into G/F or the enhancement of the GTP-dependent reconstitutive activity of G/F. However, if purified G/F is first reconstituted into cyc-membranes, which lack G/F, and subsequently treated with activated cholera toxin and NAD, covalent modification is markedly enhanced. Initial experiments also indicated that a factor present in detergent extracts of plasma membranes would facilitate the toxin-catalyzed modification of G/F. It thus became desirable to define further the requirements for ADP ribosylation of pure G/F by cholera toxin and the nature of the component(s) supplied by cyc-membranes, NAD-When partially purified factor from turkey erythrocyte membranes was used in the complete reaction mixture, only G/F and the A, subunit of cholera toxin were significantly labeled with "P. The requirement for the factor can be seen in the experiment shown in Fig. 1, in which purified rabbit liver G/F served as the substrate for the toxin. Similar results were obtained when purified turkey erythrocyte G/F was used (Fig. 1). When either the factor or dimyristoyl phosphatidylcholine was omitted, the incorporation of 3rP into G/F was almost completely abolished. Cholera toxin is an absolute requirement for the incorporation of :v2P into any macromolecule under these conditions. Omission of GTP from the reaction mixture also resulted in decreased incorporation of label into G/F. A similar pattern of requirements is observed if covalent modification of G/F is monitored by assay of its ability to reconstitute GTP-dependent adenylate cyclase activity in cyc-membranes (Fig. 1).
The activation of G/F as a function of time is shown in Fig.  2A. It is clear that the activation is absolutely dependent on cholera toxin and markedly accelerated by the factor. The initial rate of activation of G/F in the presence of the factor is constant for about 30 min. When initial rates are plotted as a function of the concentration of the factor in the activation reaction, a linear relationship was obtained (Fig. 2B). This can thus be used as a quantitative assay for the factor.
Membrane Sources of the Factor-The experiments shown above were performed with factor from turkey erythrocyte plasma membranes.
Cyc-membranes are also a useful source of factor that is free of G/F activity (see below). Attempts to demonstrate factor activity in crude cholate extracts of rabbit liver membranes were unsatisfactory. This was likely the result of very high levels of NADase activity in such preparations (data not shown).
However, rabbit liver factor activity eluted after G/F activity on the AcA-34 gel filtration column (the second step in the purification of G/F from rabbit liver) in the same way that turkey erythrocyte and partially purified factor from turkey erythrocytes (2 pg/ml) as described under "Experimental Procedures." The complete reaction mixture, as well as the same reactions minus factor, minus DMPC, minus GTP, and minus cholera toxin are shown for each source of G/F. Reactions were stopped with Laemmli's sample buffer. Slab gel electrophoresis and autoradiography were performed as described under "Experimental Procedures." The autoradiogram of the 45,000-dalton region is shown. The numbers in each lane represent the GTP-stimulated adenylate cyclase activity (pmol/l5 min/ng of G/F) that can be reconstituted with G/F that has been treated similarly. The reactions were identical with those used for labeling of G/F, with the exception of omission of labeled NAD and elevation of the concentration of NAD to 100 p~. RBC, red blood cells. G/F and factor are separated. It is at this stage of purification that the G/F from both rabbit liver and turkey erythrocytes lose the ability to serve as substrates for cholera toxin without the addition of factor. Thus, a membrane-bound factor required for cholera toxin-dependent ADP ribosylation of purified G/F is present in each of the three membrane preparations studied.

ADP-ribosylution of G / F
Characterization of the Factor-The factor from cyc-appears to be firmly associated with membranes, since complete solubilization requires at least 0.8% cholate. (Solubilization of factor or G/F from turkey erythrocyte membranes requires similar concentrations of cholate and high concentrations of salt, e.g. 0.7 M NaC1.) No activity is extracted from cycmembranes with either 0.5 M NaCl or 3 m~ EDTA in the absence of detergent. Initial characterization was performed using factor from a 1% cholate extract of purified cyc-membranes. The activity in this extract is heat labile and has a tl,Z of about 3 min at 50 "C (not shown). However, there was no loss of activity when extracts were incubated a t 30 "C for 1 h.
Factor activity is destroyed by treatment with chymotrypsin or trypsin (the former was consistently more effective), and it thus appears to be a protein or have a protein component necessary for activity (Table I). The factor does not contain a crucial sulfhydryl group, based on the results of treatment with N-ethylmaleimide (Table I). Lipids alone (soybean lipids, dimyristoyl phosphatidylcholine, or phospholipids extracted from rabbit liver plasma membranes) failed to replace the requirement for factor in the labeling reaction (not shown).
Because cyc-and other membranes contain the catalytic unit of adenylate cyclase, we hypothesized that the ADP ribosylation of G/F by cholera toxin may require an interaction of G/F with C. This hypothesis is not supported by the data obtained. Complete thermal inactivation of C activity in cyc-membranes does not affect their ability to supply factor activity. In addition, C activity is destroyed by N-ethylmaleimide, while factor activity is unaltered by similar treatment.
Gel filtration of the factor from wild type membranes in the presence of 1% sodium cholate indicates that the activity behaves as a monodisperse species (Fig.   3A). The Stokes radius of the factor in cholate is approximately 3.2 nm, equivalent to a globular protein with M , 50,000. Results of gel fdtration chromatography and determination of the sensitivity of the factor to heat and N-ethylmaleimide thus suggest that the membrane-derived factor is distinct from the known components of adenylate cyclase.
Cytoplasmic Factors that Enhance ADP Ribosylation-A factor(s) that supports the activation of G/F by cholera toxin and NAD was also detected in the 200,000 X g supernatant fraction of cyc-or wild type S49 cell homogenates. Based on TABLE I Sensitivity of factor to proteases or N-ethylmaleimide Extracts of cyc-membranes (2 mg of protein/ml) were treated as indicated and the appropriate reagent was then added to stop the treatment. In the control samples, the protease inhibitor or dithiothreitol was mixed with the protease or N-ethylmaleimide prior to exposure to the factor. The extracts were then used as a source of factor for activation of G/F by cholera toxin as described under "Experimental Procedures." G/F activity was assessed by reconstitution with cyc-membranes, and adenylate cyclase activity was assayed in the presence of 100 p~ GTP or 10 mM NaF. Protease (1 mg/ml) treatments were at 30 "C for 40 min. Treatment with trypsin and chymotrypsin was stopped by the addition of 3 mg/ml soybean trypsin inhibitor or 0.2 mM phenylmethylsulfonyl fluoride, respectively. N-Ethylmaleimide (10 mM) treatment was at 0 "C for 40 min and was stopped by the addition of 11 mM dithiothreitol. gel filtration in detergent-free solution, this activity behaves as a single species with an apparent M , = 16,000 (Fig. 3 B ) .
The activity of cytoplasmic factor was destroyed by sodium cholate (1%).
It seems possible or probable that this activity is the same as that detected by Enomoto and Gill (17), who described a cytoplasmic factor from pigeon erythrocytes that lowered the concentration of cholera toxin required to activate adenylate cyclase. The factor from cytoplasm of pigeon erythrocytes also appeared to have M , = 15,000 to 20,000 as determined by gel exclusion chromatography (17). Enomoto and Gill detected factor activity only in the presence of limiting concentrations of toxin; this may be due to the presence of membrane-bound factor in the pigeon erythrocyte preparations that were studied.
It is possible that the smaller, cytoplasmic factor represents a fragment of the membrane-bound species. Alternatively, the membrane-bound and cytoplasmic factors could represent distinct proteins with similar activities. Purification of membrane-bound and cytoplasmic factors and subsequent comparison will be required to establish their relationship.
Effect of Lzpzds-The presence of detergent (Lubrol PX and cholate are added with the G/F and factor, respectively) in the activation reaction was found to inhibit the toxindependent ADP ribosylation. Pure lipids were added in an attempt to lower the concentration of detergent in solution. Although factor activity cannot be supplied solely by the addition of lipid to purified G/F, the addition of certain phospholipids or cholesterol to reaction mixtures that contain the factor increases the rate of the reaction (Fig. 1). This effect of added lipid can be met equally well by dimyristoyl phosphatidylcholine, dimyristoyl phosphatidylethanolamine, dipalmitoyl phosphatidylethanolamine, or cholesterol. Dioctanoyl lecithin was inhibitory at all concentrations tested. It is not known whether the lipid acts by providing a hydrophobic environment, by sequestering detergent, or by some other mechanism. The finding that dioctanoyl lecithin inhibits the reaction may be explained by its detergent properties.
Effect of Guanine Nucleotides-Omission of G T P from the G/F labeling reaction results in decreased incorporation of "P into G/F (Fig. 1). The magnitude of the decrease is somewhat variable (usually 50430%) and much less than that seen when either the factor or the lipid was omitted. It is not known why a guanine nucleotide enhances covalent modification of purified G/F by cholera toxin, but GTP does appear to be the preferred nucleotide (ATP is without effect). We agree with Nakaya et al. ( 5 ) that there are three distinct functional effects of GTP. The nucleotide facilitates covalent modification of G/F, stabilizes the ADP-ribosylated regulatory protein, and activates adenylate cyclase (particularly when G/F has been ADP-ribosylated). Guanine nucleotides have been reported to increase the cholera toxin-dependent labeling of all membrane proteins to the same extent (18,19). This suggests a site of action for guanine nucleotides other than on each individual substrate for cholera toxin. If this is the case, the toxin itself or the factor described above would seem to be likely candidates.