The inhibitory guanine nucleotide-binding protein (Ni) purified from bovine brain is a high affinity GTPase.

Using modifications of the methods of Bokoch et al. (Bokoch, G.M., Katada, T., Northup, J. K., Ui, M., and Gilman, A. G. (1984) J. Biol. Chem. 259, 3560-3567) and Codina et al. (Codina, J., Hildebrandt, J. D., Sekura, R. D., Birnbaumer, M., Bryan, J., Manclark, C. R., Iyengar, R., and Birnbaumer, L. (1984) J. Biol. Chem. 259, 5871-5886), we have purified a pertussis toxin substrate with the expected characteristics of the inhibitory guanine nucleotide-binding protein (Ni) essentially to homogeneity. The purified protein consists of 3 subunits of Mr 40,000, 35,000, and less than 10,000. The Mr 40,000 band is found, upon close examination, to consist of a poorly resolved doublet. Starting with the membranes from 1,320 g of bovine forebrain we purified the protein some 100-fold with approximately 20% yield to obtain 13 mg of a greater than 95% pure protein. Chromatography on octyl-Sepharose provided efficient separation of Ni from Ns (the stimulatory guanine nucleotide-binding protein). Analytical ultracentrifugation indicates an Mr of 82,000 and a sedimentation coefficient S20,w of 5.1. The protein is able to restore opiate-mediated inhibition of adenylate cyclase to membranes prepared from NG 108-15 cells which had been treated with pertussis toxin. Bovine brain Ni has the enzymatic properties of a low Km GTPase with a turnover number of 0.3 and affinities for nucleotides in the order GppNHp greater than or equal to GTP greater than or equal to GDP much greater than ATP, CTP, UTP, and GMP. Na+ specifically stimulates the GTPase and low concentrations of Mg2+ (less than 50 microM) are inhibitory. Some Mg2+ is apparently necessary because EDTA, but not EGTA, abolishes the GTPase activity.

Guanine nucleotides play an obligatory role in the modulation of the activity of adenylate cyclase by both stimulatory and inhibitory agents (1). Separate regulatory GTP-binding proteins controlling stimulation and inhibition of adenylate cyclase were proposed (1). That involved in stimulation (N, or G/F) has been extensively characterized (2-5), largely because of the availability of a mutant S49 murine lymphoma cell line (6) functionally lacking this activity (7). The protein involved in the coupling of inhibitory receptors to adenylate cyclase (Ni) proved more refractory to study until the discovery of a toxin isolated from Bordetellu pertussis which was able to selectively transfer the ADP ribosyl moiety from NAD to this substrate (8,9). Ni has recently been purified from rabbit liver (10) and from human erythrocytes (5) and appears to consist of a heterotrimer with an a subunit of M, near 40,000, a / 3 subunit of M, 35,000 and a M, <10,000 y subunit.
* 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.
In this report we describe the purification of Ni from bovine brain, a particularly rich source. We examine some physical properties of this protein and its enzymatic activity as a high affinity GTPase.

EXPERIMENTAL PROCEDURES' RESULTS
Reconstitution-As described in the Miniprint Supplement, we purified bovine brain Ni by enrichment of the pertussis toxin substrate. To confirm that this protein really is Ni, it was important to show that it has the distinctive functional property of Ni; namely, that it can couple occupancy of inhibitory receptor by an agonist to attenuation of adenylate cyclase activity. The adenylate cyclase of membranes prepared from pertussis toxin-treated NG 108-15 cells is relatively insensitive to the attenuating effects of opiates and other inhibitory hormones, because of ADP-ribosylation of Ni (€49). Incubation of such membranes with purified Ni restores opiate inhibition of adenylate cyclase to levels near those of untreated membranes (Fig. 1). The amounts of Ni required for restoration of inhibitory receptor function are small, halfmaximal effects being achieved with approximately 10 nM.
Thus, the affinity with which Ni interacts with opiate receptors in membranes is at least lo* M.
ADP-ribosylution-Purified Ni can be ADP-ribosylated by pertussis toxin with the time course shown in Fig. 2. With the reasonable assumption that only one ADP ribosyl moiety is incorporated per Ni molecule (27,28), the preparation appears to be homogeneous. Interestingly, ADP-ribosylation of Ni by pertussis toxin shows a characteristic delay under those conditions. The lag is not observed if ADP-ribosylation is carried out at low ionic strength (data not shown).
Ni as a GTPase-Opiates and other inhibitory hormones attenuate adenylate cyclase activity by stimulating an associated, low K,,, GTPase (13). It has often been assumed that this GTPase is a property of Ni (31). Purified bovine brain Ni catalyzes the hydrolysis of GTP with high affinity (Fig. 3). The maximal velocity of the reaction, approximately 4 nmol of GTP hydrolyzed per min/mg of protein is 40 times that observed with membranes of NG 108-15 cells or of rat brain. An Eadie-Hofstee plot of these data is nonlinear (see inset to  exerting GTPase activity. It is not yet possible to distinguish between these alternatives with assurance, although assays in the presence of 50 mM Na' result in data which give linear Eadie-Hofstee plots (data not shown). Most of the GTPase activity can be described by a model assuming a K , for GTP near 100 nM, a number indistinguishable from the K,,, deduced earlier for the membrane-bound enzyme (13). GTP hydrolysis by Ni is linear with protein concentration over a wide range (data not shown). That the enzyme is truly a GTPase is shown by the competition data summarized in Fig. 4 Ni, 0.14 pg/assay, freed of detergent and ethylene glycol on a PD 10 column as described under "Experimental Procedures" was assayed for its ability to hydrolyze various concentrations of GTP. The assays also included 20 mM Tris-chloride, pH 7.5, 5 mM MgCL, and 63,000 cpm of [y3'P]GTP. Data points are the mean of quadruplicate assays. Inset, Eadie-Hofstee transformation of the data. other nucleotides. Ni, 0.14 wg/assay, freed from detergent and ethylene glycol on a PD 10 column as described under "Experimental Procedures" was assayed for GTPase activity in the presence and absence of competing nucleotides. Assays contained 1 p~ [-y-3'P] GTP (90,000 cpm) and 10 mM Tris-chloride, pH 7.5, plus the following compounds at the indicated concentrations: GppNHp, A-A GDP, u, GMP, c 3 , ATP, W , UTP, [ 1 " 0 ; CTP, U . CTP was as its sodium salt and GppNHp as a tetralithium salt. All others were Tris salts. The nucleotides were brought to pH 7.5 before use. Data points are the mean k S.E. of quadruplicate determinations.
nucleoside triphosphates tested (ATP, UTP, and CTP) inhibit only a t much higher concentrations. GMP is not inhibitory at the concentrations tested. A distinguishing feature of the low K, GTPase in membranes is a specific requirement for Na' for inhibitory receptor coupling (32, 33). In membranes of NG 108-15 cells, Na' in the presence of M e lowers GTPase activity, and inhibitory hormones such as opiates reverse this inhibition (33). Purified Ni behaves differently. Na' stimulates GTPase activity by approximately 50% in the absence but not the presence of Mg2' (Fig. 5). The effect of Na+ cannot be attributed to ionic

TABLE I
The effects of EDTA and EGTA on the GTPase actiuity of Ni Ni, 0.14 Gg/assay, freed of detergent and ethylene glycol on a PD 10 column as described under "Experimental Procedures" was assayed for GTPase activity in the presence of either EDTA (1 mM) or EGTA (1 mM). Assays also contained 20 mM Tris, pH 7.5, and 1 p~ [-y-3'P] GTP (80,000 cpm). Data points are the mean & S.E. of quadruplicate assays. Time course of GTP hydrolysis by Nr. Ni, 0.14 pg/ assay, prepared from stock on a PD 10 column as described under "Experimental Procedures" was assayed for GTPase activity at either pH 6.3 (x-X) or pH 7.5 (0L"o). Assays also contained either 20 mM Tris-3,3-dimethylglutarate, pH 6.3, at 37 "C ( U ) or 20 mM Tris.Cl (W), pH 7.5, and 1 p~ [-p3'P]GTP (70,000 cpm).

Pi released
Data points are the mean .C S.E. of triplicate determinations. strength alone because it is saturable (Fig. 5) and because Kf is much less effective (data not shown). Thus, the GTPase activity of purified Ni is specifically affected by Na+, but in a direction opposite to that occurring in membranes. Mg2+ inhibits the GTPase activity of Ni at very low concentrations, half-maximal effects being seen at <50 p~ (Fig. 6). Surprisingly, therefore, EDTA, but not EGTA' reduces the GTPase activity of Ni to negligible levels (Table I). Because the only ion known to be chelated well by EDTA and not EGTA is M e , it would appear that Ni GTPase requires some M e for activity but that excess is inhibitory.
The pH optimum of Ni GTPase is near 6.3. The peak is broad and experiments at higher pH values show a slow decrease in activity as the pH is raised to above 9 (data not shown). In order to allow comparison with earlier results of experiments with intact membranes, most of the measurements made in the present studies were at the slightly suboptimal pH of 7.5. The time courses of Ni-catalyzed GTP hydrolysis at pH 6.3 and 7.5 (Fig. 7 ) confirm that the rates are similar at these two pH values. Interestingly, the data obtained at pH 6.3 suggest a burst of activity at early times which could reflect a rapid phosphorylation of Ni as an early step in catalysis.
Several agents were tested for potential affects on GTPase activity. Hormones known to function via Ni-mediated inhibition of adenylate cyclase such as carbamylcholine, norepinephrine, and the enkephalin analogue [~-Ala',Met~]enkephalinamide have no effect on the GTPase actvity of Ni at concentrations of at least 10 pM. Similarly adenosine, NAD, and forskolin at 10 p~ concentrations are without effect as is platelet-activating factor (1 pM). Fluoride (10 mM) has, at most, a small inhibitory effect on GTPase, both with purified Ni and with intact membranes (13). Ca2+ (1 p~) inhibited the GTPase activity by about 20% and calmodulin reversed the effect, presumably by binding calcium.

DISCUSSION
The protein purified from bovine brain membranes in this work is Ni because it is a substrate for ADP-ribosylation, The abbreviations used are: EGTA, ethylene glycol bis(&aminoethyl ether)-N,N,N',N'-tetraacetic acid; SDS, sodium dodecyl sulfate; GppNHp, guanosine 5'-(/3,y-imid0)triphosphate. catalyzed by pertussis toxin, and able to restore opiate-me&ated inhibition of adenylate cyclase to membranes of NG 108-15 cells which had been pretreated, in uiuo, with pertussis toxin. The purification scheme used is a modified, and slightly simplified version of those used for the purification of Ni from rabbit liver (10) and human erythrocyte membranes (5). A major difference is the use of octyl-Sepharose which affords complete separation of Ni from N.. Bovine brain is a particularly rich source of Ni, as might be expected because of the extraordinarily high activity of adenylate cyclase in this tissue. The protein is homogenous by some criteria after only a 50-fold purification from a cholate extract of brain membranes. Thus, Ni accounts for approximately 2% of the protein of these extracts or 1% of the membranes. Pertussis toxin catalyzes the introduction of 1 mol of ADP-ribose/mol of protein. Thus, essentially all of the protein in the purified material is a pertussis toxin substrate. Very little further purification was achieved by a variety of techniques including sucrose gradient centrifugation, DEAE-Affi-Gel blue, and repeated DEAE-Sephacel and hydroxyapatite columns. The protein migrates as a single boundary with a sedimentation coefficient of 5.1 in the analytical ultracentrifuge and its behavior in sedimentation equilibrium experiments is that of a single species M, 82,000, with the presence of a small amount of either aggregation or high molecular weight impurities. On SDS-polyacrylamide gel electrophoresis, the protein shows the 3-subunit structure expected of Ni (5, lo), namely a (MI 40,000), / 3 (MI 35,0001, and y (M, < 10,000). Only minor traces of other bands are visible, corresponding to contamination levels of less than 5%. Importantly for studies on the regulation of adenylate cyclase, the Ni preparation is free of contamination with N, as judged by the low stimulation of adenylate cyclase activity in cycmembranes by the preparation.
Close examination of the SDS gel patterns shows that the a band actually consists of a doublet corresponding to M, of 40,000 (a1) and 39,000 (az). Although both bands are ADPribosylated with pertussis toxin, they can be distinguished by an antibody raised against bovine transducin. This antiserum, which recognizes Ni a from human erythrocytes binds to our a1 but not the az band.3 We do not know precisely how different the two polypeptides are but we have been able to effect a partial separation of the two with hydroxyapatite columns. The preparations relatively enriched either in aI or az are indistinguishable in the amounts required to halfmaximally restore opiate inhibition of adenylate cyclase to pertussis toxin inactivated NG 108-15 membranes. Thus, they represent different forms of Ni. Bovine brain Ni is a GTPase, albeit a relatively weak one. However, the turnover number of 0.3 is within a factor of 2-3 of the activity expected based on low K, GTPase in brain membranes. Possibly, suitable reconstitution with lipids and/ or inhibitory receptors would produce a stimulation of the GTPase activity. Experiments are in progress to test these possibilities. The stimulatory GTP-binding protein, N., isolated from rabbit liver displays a GTPase activity which, in the absence of lipids and receptors is approximately one-fifth that of our Ni preparations (34). Reconstitution with /3 adrenergic receptors resulted in a hormone-dependent GTPase activity with a turnover number of approximately 1 (34). The relationships among the GTPase activities associated with Ni, N,, and that of the intact adenylate cyclase system need further study. The time course of Ni GTPase near its pH P. Gierschik, G. Milligan, W. A. Klee, and A. Spiegel, unpublished observation.
optimum of 6.3 shows a burst of GTP hydrolysis at early (<1 min) times which might suggest the initial accumulation of a phosphorylated intermediate which then breaks down more slowly.
Interestingly, the ionic dependences of Ni as a GTPase are not the same as those of intact membranes. With purfied Ni, Na+ stimulates GTPase activity whereas Mg2+ inhibits; the two in conjunction cancel each others effect. The fact that EDTA, but not EGTA, inhibits Ni GTPase suggests that low concentrations of Mg2+ are necessary for activity. In membranes, Na+ and Mg2+ have little effect alone but the combination of the two ions inhibits GTPase. Furthermore, opiate inhibition of adenylate cyclase and stimulation of GTPase are completely dependent on Na+ and Mg2+ (33) (Fig. 1 ) . Fractions 1-10 and 11-20 wePe Pooled separately and then further processed independently but Identically.
1 r a R a n ""mber