Purification and Properties of the Inhibitory Guanine Nucleotide Regulatory Unit of Brain Adenylate Cyclase*

Hormonal inhibition of adenylate cyclase is mediated by a guanine nucleotide regulatory protein (Ni) which is different from the one which mediates hormonal stimulation. There is substantial evidence that the active component of Ni (termed ai) can be ADP-ribosy- lated by a toxin from Bordetella pertussis. We have found that in bovine cerebral cortex there are three proteins of similar molecular weight (39,000-41,000) which are modified by pertussis toxin. We have purified these proteins and have resolved the 41,000-dal- ton protein from the 40,000/39,000-dalton doublet. All three forms of pertussis toxin substrate can be isolated in free form or together with a 36,000 j3 component. We have also purified this j3 component. ADP-ribosy- lation of the three pertussis toxin substrates is greatly enhanced by the addition of the purified component. This makes possible an assay of j3 subunit activity based on its interaction with mi. The three forms of pertussis toxin substrate which we have purified differ in two functions: susceptibility to ADP-ribosylation and GTPase activity. The 41,000-dalton protein is more readily ADP-ribosylated by per- tussis toxin than the smaller forms. The 39,000-dalton protein has GTPase activity with a low K , (0.3 WM) for GTP. The GTPase activity can be doubled by phospho- lipids. The GTPase activity of the 41,000-dalton protein is almost

Hormonal inhibition of adenylate cyclase is mediated by a guanine nucleotide regulatory protein (Ni) which is different from the one which mediates hormonal stimulation. There is substantial evidence that the active component of Ni (termed ai) can be ADP-ribosylated by a toxin from Bordetella pertussis. We have found that in bovine cerebral cortex there are three proteins of similar molecular weight (39,000-41,000) which are modified by pertussis toxin. We have purified these proteins and have resolved the 41,000-dalton protein from the 40,000/39,000-dalton doublet. All three forms of pertussis toxin substrate can be isolated in free form or together with a 36,000 j3 component. We have also purified this j3 component. ADP-ribosylation of the three pertussis toxin substrates is greatly enhanced by the addition of the purified component. This makes possible an assay of j3 subunit activity based on its interaction with mi.
The three forms of pertussis toxin substrate which we have purified differ in two functions: susceptibility to ADP-ribosylation and GTPase activity. The 41,000dalton protein is more readily ADP-ribosylated by pertussis toxin than the smaller forms. The 39,000-dalton protein has GTPase activity with a low K , (0.3 WM) for GTP. The GTPase activity can be doubled by phospholipids. The GTPase activity of the 41,000-dalton protein is almost undetectable. It is not yet known what the relationship of the forms is to each other. The smaller forms may be derived from the larger by proteolysis or it may be intrinsically different. It remains to be shown whether one of the forms represents a different type of regulatory protein which transmits a hormonal signal to effectors other than adenylate cyclase.
Hormonal inhibition of adenylate cyclase is mediated by a guanine nucleotide binding protein (Ni') which is different from the one which mediates hormonal stimulation (N.) (re-* This work was supported by National Institutes of Health Grant AM19277 and by Grant BC380A from the American Cancer Society.
A preliminary report of some of the studies described here was presented at the 5th International Conference on Cyclic Nucleotides and Protein Phosphorylation, June 27-July 1, 1983, Milan, Italy and at the 13th Meeting of the Society for Neurosciences, November 6- 11,1983, Boston, MA. 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: Ni, the guanine nucleotide regulatory unit which mediates hormonal inhibition of adenylate cyclase. It is composed of ai and 6; Ne, the guanine nucleotide regulatory unit which mediates hormonal stimulation of adenylate cyclase. It is composed of as and p; as, the polypeptide component of N. which can be ADP-ribosylated by cholera toxin and which activates the catalytic viewed in Ref. 1). Like N,, the inhibitory component (Ni) is made up of at least two proteins: an ai subunit that can be ADP-ribosylated by a toxin from Bordetellu pertussis and a p component that seems to be similar or identical from both N,.* In addition, there may be a small molecular weight y subunit of unknown function associated with /3 (5).
The conclusion that the protein ADP-ribosylated by pertussis toxin is indeed the ai component of adenylate cyclase rests on substantial evidence from studies with intact cells, with cell membranes, and with purified adenylate cyclase components. In intact cells, exposure to pertussis toxin attenuates or abolishes the ability of some hormones to decrease intracellular cyclic AMP levels and blocks their physiological effects (6,7). Treatment of cells or membranes with pertussis toxin diminishes or blocks the ability of inhibitory hormones to attenuate adenylate cyclase activity (8-11) or to increase GTPase activity (11). Pertussis toxin treatment can also reduce receptor affinity for an inhibitory agonist (12). These actions are correlated with ADP-ribosylation of 41-kDa plasma membrane protein (9-12). The 41-kDa pertussis toxin substrate, in association with a 35,000-dalton /3 subunit, has recently been purified from rabbit liver (2) and human erythrocytes (3,13).
In addition to covalently modifying the inhibitory guanine nucleotide regulatory protein of the adenylate cyclase system, pertussis toxin can also ADP-ribosylate a closely related protein from bovine retinal rods, the Ta component of transducin (14, 15). This 39-kDa protein is structurally similar to the ai component (4 In the studies reported here, we have used ADP-ribosylsation by pertussis toxin as a means of identifying the a, component. We have found that in bovine cerebral cortex, there are three substrates for ADP-ribosylation by pertussis toxin, with molecular weights of 39,000, 40,000, and 41,000. The forms differ not only in size but in function. For simplicity, we have referred to all three as ai throughout the paper. However, this is a tentative assignment since some of these proteins may also couple hormone receptors to effectors other than adenylate cyclase (see "Discussion"). The approximately 40-kDa proteins have been purified both in association with the fl component and alone. We have also purified the /3 component free of ai. Northup et al. (19) have recently deunit. Heavier (aeH) and lighter ( a d forms of as are found in some cells; (3, a polypeptide component of both Ni and N. with a molecular weight of 35,000-36,000; ai, the polypeptide component of Ni which can be ADP-ribosylated by pertussis toxin; Gpp(NH)p, guanosine 5'-(@,a-imino)triphosphate; SDS-PAGE, sodium dodecyl sulfate-polyacrylamide gel electrophoresis. scribed an assay for activity of the p subunit based on its association with as, the guanine nucleotide binding subunit of N,. We now show that the activity of p can also be measured based on i t s interaction w i t h ai.

MATERIALS AND METHODS
ADP-ribosylntion of Purified Ni-Ni was identified by [32P]ADPribosylation with pertussis toxin under the following conditions: 5 p~ NAD containing 0.3-0.5 pCi of (32P]NAD, 2.5 mM ATP, 2 mM GTP, 10 mM isoniazide, 10 mM thymidine, 60 ng of pertussis toxin, 0.2-0.5 pg of substrate protein in a total volume of 25-30 pl. Some earlier experiments also included 0.2 mM Gpp(NH)p. After 45 min at 37 "C, the reaction was stopped by the addition of 2% SDS in Laemmli sample buffer (20) and boiling for 1 min. For reconstitution of a i and 8, the two components were incubated together for 10 min a t 23 "C before the assay reagents were added. The entire reaction mixture was applied to 9-11% SDS-polyacrylamide gels as described by Laemmli (20). The gels were stained with silver nitrate (21) or Coomassie Blue, dried and used to expose Kodak X AR50 film for 1-5 days a t -70 "C. Alternatively, the region containing the radioactive protein was cut out and counted in Ultrafluor (National Diagnostics) in a Beckman LS250 Liquid Scintillation Counter.
Protein Determination-Protein was determined by the method of Lowry et al. (22) as modified by Bailey (23).
Assay of GTPase and ATPase Actioity-GTPase activity was measured using [ -p3'P]GTP (Amersham Corp.) as a substrate as described by Neer et al. (24) except that no GTP-regenerating system was included. The labeled nucleotide was lyophilized and taken up in a small volume of 0.05 M Tris.CI, pH 7.6, prior to use. The specific activity ranged from 100 to 2500 cpm/pmol depending on the requirements of the assay. The blank was 3-7% of the total radioactivity in the assay.
Preparation of N,-The procedure was a modification of the method of Northup et al. (25) and Sternweis et al. (26). Fresh bovine cerebral cortex (500 g) was homogenized in 3 liters of buffer of the following composition: 50 mM Tris.Cl, pH 8.1, 5% sucrose (w/v), 6 mM MgC12, 1 mM EDTA, 1 pg/ml soybean and lima bean trypsin inhibitors, 3 mM benzamidine.Cl, 1 mM dithiothreitol. The homogenate was centrifuged for 20 min a t 10,000 X g a t 4 "C. The pellet, which measured approximately 450 ml, was taken up in 2.5 liters of homogenizing buffer and recentrifuged for 20 min a t 10,000 X g. The pellet from this centrifugation, which measured approximately 350 ml, was brought up to 500 ml of homogenizing buffer and frozen a t -70 "C. Before solubilization, it was thawed, brought to 1 liter with homogenizing buffer and recentrifuged. The final pellet of 250 ml was brought up to 1 liter with homogenizing buffer. To this was added 1 liter of homogenizing buffer without sucrose containing 2% cholate to give a final cholate concentration of 1%. The homogenate was allowed to solubilize at 4 "C for 90 min and was then centrifuged at 32,000 X g for 70 min. The total volume of supernatant was 1700 ml.
The cholate supernatant was made 10 p~ in AICIJ and 10 mM in NaF and applied to a 1500-ml column of DEAE-Sephacel (Pharmacia) equilibrated with 50 mM Tris.CI, pH 8,75 mM sucrose, 6 mM MgC12, 1 mM dithiothreitol, 1 mM EDTA, 3 mM benzamidine.Cl, 1 mg/liter soy bean and lima bean trypsin inhibitor (Sigma), 10 p M AICI,, 10 mM NaF, and 0.9% cholate (Buffer A). The column, termed DEAE-Sephacel I in Table I, was eluted with a 1800-ml linear NaCl gradient from 0 to 0.25 M NaCl in Buffer A. The column was assayed for N. activity by reconstitution with resolved catalytic unit from bovine caudate nucleus as described previously (27). Most of the approximately 40 and 36 kDa proteins were included in the peak of N. activity. This was pooled, concentrated in an Amicon concentrator with a XM-50 or YM-10 membrane to 1/4 its volume, and applied to a 1700-ml column of Sepharose 6B equilibrated with Buffer A. N. activity, as well as the approximately 40 and the 36-kDa proteins, eluted from the column with a distribution coefficient of 0.5-0.6.
The peak of N. activity from the Sepharose 6B column was pooled and applied to a second column of DEAE-Sephacel (DEAE-Sephacel 11, Table I) equilibrated with Buffer A. The volume of this column was 350 ml and it was eluted with a 1400-ml linear gradient of NaCl (0-0.25 M) in Buffer A. N. activity eluted from this column of two peaks as shown in Fig. lA. Ni eluted between the two N, peaks. The distribution of polypeptides in the column effluent is shown in Fig.   1B. The subsequent steps of the purification are described under "Results."

RESULTS
Purification of cui-A flow chart for the purification of the cui and / 3 components is s h o w n i n Table I. T w o routes led to the preparation of pure components. The first used fractions separated from N. activity by the DEAE-Sephacel column s h o w n i n Fig. 1, A and B. Fractions 136-148 were pooled and applied to a column of heptylamine-Sepharose synthesized by the method of Shaltiel (28). The 50-ml heptylamine-Sepharose column was equilibrated with Buffer A which contained 100 mM NaCl and 0.4% Tris. cholate; lima bean and soybean trypsin inhibitors were omitted from buffers for this and  Table I) was the starting material for this column. A 150-ml sample containing 360 mg of protein was applied to a 350-ml column and eluted as described under "Materials and Methods." The fraction size averaged 6 ml. Samples of 0.5 pl were assayed for N. activity by reconstitution with resolved catalytic unit from bovine caudate nucleus as described by Bender and Neer (27). The reconstitution assay contained 20 p~ Gpp(NH)p and 50 p~ forskolin (0). Protein in 5O-pl samples (0) was measured by the method of Lowry (22,23). E , samples of 40 pl from the designated column fractions were applied to an 11% acrylamide Laemmli gel (20). The gel was stained with Coomassie Blue.  subsequent columns. After the sample was applied, the column was washed with 50 ml of Buffer A and then 50 ml of Buffer A containing 0.5 M NaCl and 0.3% cholate. The column was eluted with a 250-ml linear gradient beginning with 50 mM Tris, 150 mM NaC1, and 0.3% cholate and ending with 50 mM Tris, 25 mM NaC1, and 1.3% cholate. The elution profile was analyzed by SDS-PAGE and is shown in Fig. 2. A protein doublet with molecular weights of 39,000 and 40,000 eluted in two peaks from the hydrophobic column. About half did not bind and eluted in the flow-through volume (peak 1). These flow-through fractions also contained a large amount of an as yet unidentified group of approximately 30-kDa proteins which can be seen as contaminants in Fig. 1B. The remainder of the 39/40-kDa protein bound weakly to the resin and was eluted in the buffer wash (peak 2A/2B). Peak 2A contained some 36-kDa protein as well as the 39/40-kDa doublet. Another peak of the 36-kDa protein, well resolved from all the others, was eluted from the column with the cholate gradient (peak 3). In contrast to the 39/40-kDa protein doublet, which eluted from the heptylamine-Sepharose in the void volume or with the 0.5 M NaCl wash, the 36-kDa protein seemed to bind hydrophobically.
The 39/40-kDa protein eluting in heptylamine-Sepharose peak 1 could be separated from a set of 30-kDa proteins by DEAE-Sephacel chromatography in 0.6% Lubrol 12A9 as shown in Fig. 3 to give rise to mi-I (Table I). The proteins were concentrated by the DEAE-Sephacel chromatography SO that on SDS-PAGE we could now see a 41-kDa protein which eluted a little ahead of the main 39/40-kDa doublet. The functional properties of these proteins will be discussed below.
Purified mi containing predominantly the 39-kDa protein could also be obtained by heptylamine-Sepharose chromatography of fractions adjacent to the single peak of N, activity found in the first Sepharose 6B column (see "Materials and Methods" and Table I). When these fractions were applied to heptylamine-Sepharose under the same conditions as Fig. 2, the 30-kDa contaminant eluted in the flow-through volume. The 39-kDa protein eluted at the position of peak 2B in Fig.  2. Most of the 36-kDa component bound firmly to the hydrophobic resin and was eluted in the cholate gradient. The 39-kDa protein could be separated from the remaining contaminants by gel filtration over Ultrogel AcA-44 (LKB) equilibrated with 0.05 M Tris. C1, pH 7.6, 6 mM MgC12, 1 mM EDTA, 1 mM dithiothreitol, and no detergent. This preparation, called ai-I1 (Table I), contained only a trace of the 40-kDa and no 41-kDa protein. An overloaded SDS-PAGE pattern of a peak fraction is shown in Fig. 4.
Even without detergent, the 39-kDa protein eluted from the Ultrogel AcA-44 gel filtration column with the same distribution coefficient as hemoglobin, indicating that the 39-kDa protein does not aggregate in the absence of detergent. This suggests that the protein is not strongly hydrophobic and is consistent with the observation that it can be eluted from heptylamine-Sepharose with buffer or 0.5 M NaCl.
We have also purified the 39-41-kDa proteins in association with the 36-kDa ( 3 component (Ni-I in Table I). Fractions 125-135 from the DEAE-Sephacel column shown in Fig. 1, A and B, were diluted to decrease the salt concentration and applied to a 130-ml DEAE-Sephacel column equilibrated with 0.05 M Tris. c1, 75 mM sucrose, 6 mM MgC12, 1 mM dithiothreitol, 3 mM benzamidine, 0.6% Lubrol 12A9 without AlClB or NaF. The column was eluted with an NaCl gradient from 0 to 0.25 M NaCl in the equilibrating buffer. Although the 36-kDa component was present, the cluster of approximately 40-kDa proteins distributed similarly to the column profile shown in Fig. 3. Lanes 2 and 3 of Fig. 5 show the protein distribution from the leading edge of the peak and from the back edge. The 41-kDa protein elutes ahead of the 39/40-kDa proteins. Fig. 5 shows a comparison of the mobility on a single SDS-PAGE gel of ai purified without-the /3 unit on heptylamine-I R ' 2 A 28 Purification of the @ Subunit-Purified 36-kDa p component could be obtained from two sources: it could be separated from (vi by heptylamine-Sepharose chromatography as shown in Fig. 2, and it could be further purified by gel filtration over an Ultrogel AcA 44 column in 50 mM Tris.C1, 6 mM MgClP, 75 mM sucrose, 3 mM benzamidine-CL, 1 mM dithiothreitol, and 0.4% cholate. This is preparation @-I1 on the chart in Table I. An overloaded SDS-PAGE gel of such a preparation is shown in Fig. 3. A small amount of a 35-kDa protein is also found in all the preparations. Fig. 4 (lanes 6 and 7) shows that the pure brain @ component has the same electrophoretic mobility as the TP component of transducin.

-
The 36-kDa protein could also be purified from peak 3 in  Fig. 2 was pooled, diluted 3-fold into 50 mM Tris.Cl, pH 7.9, 6 mM MgC12, 75 mM sucrose, 10 P M AICln, 10 mM NaF, 1 mM EDTA, 0.6% Lubrol 12A9 and applied to a 50-ml DEAE-Sephacel column equilibrated with the same buffer. The column was washed with 50 ml equilibrating buffer then eluted with linear gradient from 0 to 0.25 M NaCl in the equilibrating buffer. No protein eluted before the start of the NaCl gradient which began a t fraction 116. The same amount of each fraction was applied to the silver-stained SDS-PAGE gels shown in the fop and bottom figure but the lower figure was more lightly stained to increase the resolution of the protein bands and shows the peak region in more detail.
(preparation p-I in Table I). The ( 3 subunit from liver also does not bind to DEAE-Sephacel in Lubrol 12A9 (26). The subunit obtained from both sources is active in the functional assay that will be described below.
Properties of Ni-Pertussis toxin predominantly ADP-ribosylates a 41-kDa protein in crude preparations of membranes from many cell types (9-12). Fig. 6A shows that only one radioactive band is found when a relatively crude preparation of solubilized brain proteins is incubated with ['"PI NAD and pertussis toxin. However, this band is rather broad, suggesting that there may be heterogeneity in the brain substrates for pertussis toxin. This heterogeneity is confirmed by the experiment shown in Fig. 6B. Preparations enriched in either the 41-or the 39-kDa proteins were ADP-ribosylated with pertussis toxin as described under "Materials and Methods." The fractions used in the experiment are the same ones shown in lanes 2 and 3 of Fig. 5. The larger and smaller proteins can all be ADP-ribosylated by pertussis toxin but the larger seems to be the better substrate. Although the two assays contained equal amounts of 41-and 39-kDa substrate, and of p subunit, there was 2-5 times as much radioactivity  Table I; lane 7, the transducin /3 component from bovine retinal rods purified as described in Ref. 17. in lune 2 contained no detectable 41-kDa protein, yet its presence in trace amounts was revealed by ADP-ribosylation. Fig. 6A shows that we were able to ADP-ribosylate the ai component in rather crude preparations. However, we found that the pure preparations of 39-41-kDa proteins labeled poorly if a t all. The reason for this difficulty became apparent from the results of the experiment shown in Fig. 7

. Heterogeneity of pertussis toxin substrates in bovine brain ADP-ribosylation of a relatively crude fraction (left) and ADP-ribosylation of resolved forms of a i (right).
Left, a relatively crude fraction containing a, was taken from the DEAE-Sephacel I1 step in the purification procedure (see Fig. 1) and treated with ["'PINAD and pertussis toxin as described under "Materials and Methods" (lane 2). A control fraction was treated in the same way except that pertussis toxin was omitted (lane I ). The silverstained SDS-PAGE pattern is shown on the left (lane 3 ) and a radioautograph of the gel is shown on the right. Kodak X AR5 film was exposed for 2 days with two Cronex enhancing screens. Right, the fractions from preparation Ni-I (Table I)  proteins: one which is a heterodimer and which can be ADPribosylated, one which is not associated with the fi subunit and cannot be ADP-ribosylated. Addition of excess pure fi component allowed ADP-ribosylation of ai in all the fractions. This shows that the centrifugation had not separated a pool of 39/40-kDa proteins which could not be ADP-ribosylated. Heat-inactivated / 3 component did not enhance ADP-ribosylation. The amount of ["'PIADP-ribose incorporated into the 39-kDa protein depends on the amount of fi added (Fig. 8). In the experiment shown, 0.2 mol of ADP-ribose are incorporated per mol of 39-kDa protein at saturation. We used tri-I1 (Table I)  interaction with ai. This would be complementary to assays described by Northup et al. which measure the function of the /I unit based on its interaction with aB (19). The finding that ai needs /3 for ADP-ribosylation by pertussis toxin is consistent with the observation of Codina et al. (3) that after ADPribosylation, the ai and / 3 subunits of Ni from human red cells are associated into a heterodimer.
GTPuse Activity of Ni-The purified 39-kDa protein and the 39/40-kDa doublet have GTPase activity with a high affinity for GTP. Fig. 9 shows the GTP concentration dependence of the ai GTPase activity. The apparent K , for G T P is 0.3 f 0.1 ~L M G T P ( n = 4). The GTPase activity distributes with the 39-kDa protein or the 39/40-kDa doublet through a variety of separation procedures. An example is shown in Fig. 10, which shows the correlation of GTPase activity with the amount of 39/40-kDa protein on DEAE-Sephacel column. In other experiments, we found that GTPase activity also coincided with the peak of 39/40-kDa protein on sucrose density gradient centrifugation in Lubrol 12A9-containing buffers, and with the peak of 39-kDa protein on Ultrogel AcA 44 gel filtration in buffers with no detergent (not shown). This consistent distribution suggests that the  Table I) was incubated with increasing amounts of purified p as indicated in the figure. The reaction mixture contained 0.3 pg of a, or 0.2 p~ ai. The ADP-ribosylation reaction was carried out as described under "Materials and Methods." For each point, a control incubation containing 0 but no ai was also carried out. All the reactions were analyzed by SDS-PAGE. The gels were stained with silver nitrate to locate the protein bands and the 39-kDa protein was excised and counted by liquid scintillation spectroscopy in 10 ml of Ultrafluor (National Diagnostics). The equivalent region was cut out of the control lanes containing fi but not a, and the counts subtracted from the test samples containing ai. Incorporation of radioactivity into any ai which may have contaminated p was less than 20 cpm over background.
GTPase activity is indeed a property of these proteins and not of a contaminant. Pure /3 component has no detectable GTPase activity.
The ability to separate the 41-and 39-kDa forms of Ni allowed us to compare their GTPase activity. We used the fractions shown in lanes 2 and 3 of Fig. 4 for this experiment. Surprisingly, the GTPase activity of the 41-kDa protein was much lower than that of 39-kDa one. The specific activity of the former was 0.2 f 0.1 nmol of Pi/(mg X rnin), n = 4 while that of the latter was 1.4 nmol of Pi/(mg X rnin), n = 4. In both cases we tested the activity at three concentrations of protein. The activity was linear with protein and with time. The specific activity of the 39-kDa protein was similar whether it was measured in the presence of the /3 subunit (as in the experiment described above) or without the / 3 unit. The pure 39-kDa protein (ai-II, Table  I; lane 2, Fig. 3) had a specific GTPase activity of 2.4 f 0.4 nmol of P,/(mg x min), n = 5 .
Although the GTPase activity of the ai protein is easily measurable, its turnover number is quite low: 0.08-0.20 mol of Pi product/(mol of enzyme x min) in four different preparations. The reason for the low turnover number is not clear. The GTPase activity associated with ai can be increased 2fold by the addition of mixed soybean phospholipids (Sigma, P3644) which contain predominantly phosphatidylcholine (Fig. 11). This suggests that the activity might be higher if measured in a membrane environment.  Fig. 2 were assayed for GTPase activity by the procedure described under "Materials and Methods" (0). The concentration of GTP was 0.5 p~. The position of the approximately 40-kDa polypeptides was determined hy SDS-PAGE followed by densitometry of the silverstained, dried gel. The area under the peaks of the densitometer scan is reported in arbitrary units (0).

DISCUSSION
The methods described here lead to the purification of three substrates for ADP-ribosylation by pertussis toxin from bovine brain as well as purification of a 36-kDa p component.
The latter can interact both with the stimulatory guanine nucleotide regulatory protein, as, ' and, as we now show, with the inhibitory unit ai. The ( 3 unit enhances ADP-ribosylation of all three pertussis toxin substrates. The capacity to interact  (Table I) was measured in the presence of the indicated concentration of mixed soybean phospholipids. The phospholipids were homogenized in 0.05 M Tris. C1, pH 7.6, 10 mM MgC12, 75 mM sucrose, 1 mM dithiothreitol with a Dounce homogenizer to produce a smooth emulsion containing 4 mg/ml of phospholipids. There was no effect of the phospholipids on the assay blank at any concentration used.
with both as and c y i components is consistent with the role proposed for the / 3 subunit by Sternweis et al. (26) and Northup et al. (19). The observation that the amount of ["PI ADP-ribose incorporated into ai depends on the concentration of 6 subunit, means that the ADP-ribosylation can only be used to quantitate ai under conditions where the amount of /3 is known to be saturating. The substrate for ADP-ribosylation by pertussis toxin differs in brain from that reported in other tissues. Instead of being a single polypeptide, there appear to be three proteins of very similar molecular weight which can be modified by the toxin. These proteins can be separated by chromatography over DEAE-Sephacel. The protein present in greatest quantity is a 39-kDa one which can be obtained either in free form or associated with a 36-kDa /3 component. This protein is less effective as a substrate for pertussis toxin than the 41-kDa protein which is also found in association with a 36-kDa subunit. In addition to these two, we can detect a 40-kDa ADP-ribosylated protein which we have not yet obtained free of the 39-kDa polypeptide. The relationship among these proteins is not yet known. It is tempting to speculate that the proteins may be derived from each other by proteolysis but this must await analysis of the separated subtypes. The molecular weight of the larger ADP-ribosylated protein is the same as that reported for purified a i from rabbit liver (2). The smaller molecular weight is the same as that reported for N, purified from human red blood cells (3). It also corresponds to that of the Ta subunit of transducin from bovine retinal rods, a guanine nucleotide binding protein which can also be ADP-ribosylated by pertussis toxin (15), and the two molecules migrate identically in our SDS-PAGE gels. It has previously been shown that the 41-kDa ai subunit from rabbit liver and the Ta component of transducin share some peptides, suggesting that they have a structural similarity (4). Extrapolation from that data suggests that the three approximately 40-kDa brain proteins may be related.
Similar heterogeneity occurs in as from brain and other tissues. In many tissues cholera toxin identifies only a 42-45-kDa form of the as component of N,. In brain (29), S49 mg/rnl Phospholipid lymphoma cells (30), 3T3-Ll (31, 32), and other cell types (33, 34), two forms of N, can be identified, a heavier form, as", with a molecular mass of 47-52 kDa and a lighter form, asL, with a molecular mass of 42-45 kDa. Hudson and Johnson (30) showed that the heavier and lighter forms of a, from S49 lymphoma cells have common peptides. This observation has been confirmed in fat cells (33) and liver (34).
The larger and smaller forms of ai from cerebral cortex differ in their GTPase activity. The larger form has barely detectable GTPase activity. In contrast, the 39-kDa protein has a clearly measurable GTPase activity. The affinity for G T P is high with a K, of 0.3 p~. The K,,, is similar to that of the epinephrine-stimulated GTPase of human platelets (35). Although the enzymatic activity is easily measured, the turnover number of the ai-associated GTPase is low (0.08-0.2 mol of Pi formed per mol of enzyme X min). The activity is increased 2-fold by the addition of phospholipids. Unlike transducin, it is increased only slightly if at all by the addition of purified p subunit (data not shown). The purified N, protein has no detectable GTPase activity unless it is reconstituted in lipid vesicles with a hormone receptor (34). The specificity of the hormone response argues that the GTPase is indeed an intrinsic activity of N, and not of a contaminant. However, even after hormonal stimulation, the turnover number is only 1 mol of Pi formed per mol of enzyme/min. Brandt et al. (36) calculate that this value is similar to that observed in intact turkey erythrocyte membranes. The activity of the GTPase of N,, Ni, and transducin may be limited by factors such as the rate of substrate binding and the slow rates measured may reflect the regulatory function of GTP turnover.
The heterogeneity of size, the differences in susceptibility to ADP-ribosylation by pertussis toxin and in GTPase activity which we find in the a, proteins from brain may be due to a primary difference in protein structure or to secondary changes such as partial proteolysis or post-translational modifications. Whatever the molecular cause, the differences may reflect a physiologically important heterogeneity of function.
A number of examples are known of hormones which in some cells inhibit adenylate cyclase but in others have effects independent of cyclic AMP and do not inhibit adenylate cyclase. Nevertheless, in both, binding of the hormone is modulated by guanine nucleotides. For example, in the rat mesenteric artery and rabbit myocardium, binding of angiotensin I1 is regulated by cations and guanine nucleotides but angiotensin I1 does not inhibit prostaglandin E,-stimulated cyclic AMP accumulation (37,38). In liver, on the other hand, angiotensin I1 has complex effects some of which are linked to adenylate cyclase inhibition and some which are not (39). Similar kinds of observations have been made for the aladrenergic receptor (40)(41)(42). These observations suggest that guanine nucleotide binding proteins may exist which mediate receptor activation of enzyme systems other than adenylate cyclase or which allow receptors to couple to membrane ion channels. It will be important to determine whether the 39-, 40-, and 41-kDa proteins we have isolated are different types of regulatory proteins which transmit the hormonal signal to different effectors. The answer to this fundamental question must await reconstitution of hormonal inhibition of adenylate cyclase in a defined system as well as demonstration of a role for these proteins in regulation of cyclic AMP-independent hormonal actions.