Isolation of Two Proteins with High Affinity for Guanine Nucleotides from Membranes of Bovine Brain*

Membranes from bovine brain bind relatively large quantities of guanosine 5’-(3-O-thio)triphosphate (GTPyS) with high affinity. The two proteins responsible for most of this activity were purified; they account for 1.5% of the membrane protein. The two proteins contain a subunits of either 39,000 or 41,000 Da, ,8 subunits of 36,000 or 35,000 Da, and a potential y subunit (11,000 Da). These structures are the same as a family of proteins that includes transducin and the regulatory proteins, Gs and GI, of adenylate cyclase. The 41,000- and 39,000-Da polypeptides can be ADP-ribosylated with islet-activating protein from Bordetella pertussis, bind guanine nucleotides specifically, and migrate through polyacrylamide gels with rates similar to the a subunits of GI and transducin, respectively. The 36,000- and 35,000-Da polypeptides are similar to the j3 subunits of GI and Gs. The y subunit is found whenever ,8 subunits are present. The 41,000- and 39,000-Da polypeptides (with ,8 and y) are desig-nated, respectively, GI and Go from brain. The a subunit of Go was isolated without the use of ligands known to dissociate other G proteins. Goa binds GTPyS reversibly in the absence of Mg2+ and is relatively stable in cholate. This isolated a subunit should be of great utility in elucidating

bovine rod cells. The purified protein contains two larger subunits (a, p) with molecular weights of 39,000 and 36,000, respectively (4)(5)(6)(7)(8). It also contains a smaller y subunit of about 10,000 Da (6,8). GS has been purified from rabbit liver (9,10) and turkey and human erythrocytes (11,12); it was reported to have a composition consisting of two major subunits (a, p) with molecular weights of 45,000 and 35,000. A second a subunit of 52,000 Da was also obtained in lower quantities from the liver. Procedures developed for the purification of GS from liver and erythrocytes have been slightly modified to purify GI from rabbit liver (13,14) and human erythrocytes (15). GI contains two major subunits (a, /3) with molecular weights of about 41,000 and 35,000. The presence of a potential y subunit of smaller size has been observed in GI from liver (14) and human erythrocytes (16). A y subunit is also reported to be present in Gs from human erythrocytes (16).
The structural homology of these three proteins is striking.
Furthermore, the p subunits from Gs, GI, and T have the same electrophoretic mobility in SDS gels, yield similar peptides upon proteolysis, and have identical amino acid compositions (17). The a subunits were also similar with respect to amino acid composition and proteolysis (17). All three proteins can be modified by ADP-ribosylation. Thus, the a subunits of Gs can be specifically modified by cholera toxin (9,18,19), and the a subunit of GI can be modified by IAP, the ADP-ribosylating toxin of Bordetellu pertussis (13,20). Transducin a can be modified by both IAP (21) and cholera toxin (22).
The mechanisms of action of Gs, GI, and T are also quite similar. GTP is required for the activation of all three proteins in membranes, and GTPase activities have been implicated in their function (1)(2)(3). The a subunits contain the binding site for guanine nucleotides (8,13,23,24), and activation of the purified proteins in vitro coincides with dissociation of the subunits (8,10,11,13,24). These and other experiments with purified subunits (8,25,26) have implicated the a subunits of these proteins as the active species in the regulation of adenylate cyclase by Gs and GI and cGMP-dependent phosphodiesterase by T. Additional experiments suggest that the /3 subunit may be a key element in the inhibition of from the other G proteins, and can be isolated from its other subunits in a stable and unliganded form. The function of this protein is not yet clear.

EXPERIMENTAL PROCEDURES
Membrane Preparations-Bovine brains were obtained from the heads of freshly slaughtered cattle and placed in ice-cold 10 mM Tris-C1, pH 7.5 (about 1 h). All further procedures were carried out at 0-4 "C. Cerebra were dissected crudely to remove any remaining brainstem and to excise large portions of white matter. The remaining cerebral tissue (-150-200 g/brain) was homogenized in a blender (medium speed) with 4 volumes of 10 mM Tris-C1, pH 7.5, 10% sucrose. The homogenate was filtered through 4 layers of cheesecloth, and the membranes were collected by centrifugation at 20,000 X g for 30 min. The membrane pellet was suspended in 5 volumes of 10 mM Tris-C1, pH 7.5, 10% sucrose with a Potter-Elvehjem homogenizer and collected by centrifugation at 20,000 X g for 60 min. The membranes were subjected to a second identical wash and then resuspended with the same solution to a protein concentration of about 20 mg/ml for storage a t -80 "C.
This procedure was also utilized to prepare membranes from brains of decapitated rats with the following changes; excision and cooling of brains was rapid (within 5 min after death) and buffers contained 0.5 mM phenylmethanesulfonyl fluoride.
Membranes from bovine heart (both atrial and ventricular muscle) were prepared at 0-4 "C as follows. Muscle tissue (250 g) was homogenized with 1,400 ml of 10 mM Tris-C1, pH 7.5,0.7 M KC1 in a blender (high speed) and passed through 2 layers of cheesecloth. After centrifugation at 8,000 X g for 30 min, the pelleted membranes were suspended in 800 ml of 10 mM Tris-C1, pH 7.5, and homogenized with a Polytron homogenizer, model PT-20, at a setting of 7 for 30 s. More dense membranes were separated from a lighter fraction by centrifugation at 8,000 X g for 20 min. This procedure was repeated three times, and the supernatants were collected each time. Supernatants from the three slow spins were centrifuged at 20,000 X g for 60 min to collect a fraction of less-dense membranes. These membranes were resuspended in the same buffer and stored at -80 "C; the membranes were enriched (4to 15-fold) for muscarinic receptor-binding activity (300-1,200 fmol/mg of protein) over the original pelleted material.
Plasma membranes from the cyc-(Gs-deficient) variant of the S49 lymphoma cell (29) and crude membranes from rat and rabbit liver (10) were prepared as described.
Purification of GTPyS-binding Proteins-The procedures utilized to purify the major GTPyS-binding proteins from bovine brain were essentially the same as those developed to purify Gs and GI from rabbit liver (9,10,13) and Gs from turkey erythrocytes (11). The detailed procedures that follow will clarify any changes from the methods published previously. All procedures were done at 0-4 "C.
Membranes (12 g of protein) from bovine brain were washed once with 1200 ml of TED buffer (20 mM Tris-C1, pH 8, 1 mM EDTA, 1 mM dithiothreitol) containing 100 mM NaCl. The washed membranes were suspended to 2 liters with TED buffer and a final concentration of 1% sodium cholate and incubated with stirring for 1 h. Extracted membranes were removed by centrifugation at 35,000 rpm for 60 min in Beckman 35 rotors. The Supernatant (extract) was applied to a 1liter column of DEAE-Sephacel (Pbarmacia) which had been equili- brated with 3 liters of TED/1.0% sodium cholate. The GTPySbinding proteins were then eluted from the DEAE with a linear gradient of NaCl(2 liters; 0-225 mM) in TED/1% cholate. The eluate was collected in fractions of 23 ml. The gradient was followed by further elution with 1 liter of 500 mM NaCl in TED/1.0% cholate. Fig. 1 details the activities measured in the eluted fractions and will be discussed further under "Results." Fractions containing the major peak of GTPyS-binding activity (fractions 64-74) were pooled and concentrated to 20 ml by pressure filtration through an Amicon PM-30 membrane. The material was applied to a 1.2-liter column of Ultrogel AcA 34 (LKB) and eluted overnight with TED/1.0% cholate/100 mM NaCl. Fractions of 14.5 ml were collected. The major peak of GTPyS-binding activity (frations 51-56; 87 ml) was diluted with 270 ml of TED/100 mM NaC1. This diluted pool was applied to a 100-ml column (2.2 X 26 cm) of heptylamine-Sepharose. The column was then washed with 100 ml of TED/0.25% cholate/300 mM NaCI. Elution of GTPyS-binding proteins was accomplished with a linear gradient (800 ml total) of TED/0.25% cholate/200 mM NaCl to TED/1.3% cholate/50 mM NaCl. Fractions of about 8 ml were collected in tubes that had been siliconized with Aquasil (Pierce Chemical Co.). Figs. 2 and 3 show the profiles of the activities that were eluted from these two columns.
Purified proteins from the heptylamine-Sepharose column were utilized directly for experiments or pooled appropriately and concentrated by filtration on an Amicon PM-30 membrane to about 1 mg/ ml. The proteins could then be stored at -80 "C or on ice for several weeks with little or no loss of binding activity.
The @ subunits of the G proteins from brain were obtained by further treatment of fractions eluted from heptylamine-Sepharose that contained GS activity; these procedures were described by Northup et al. (24).
Assays-G proteins were identified by their ability to bind GTP+. Samples were diluted into 10 mM NaHepes, pH 8, 1 mM EDTA, 1 mM dithiothreitol, 0.1% w/v Lubrol 12A9. In the standard assay, 20 pl of diluted sample were mixed with 20 pl of 50 mM NaHepes, pH 8, GTPyS, and [%]GTPyS (-106 cpm). The samples were incubated at 30 "C for 40-60 min. Alterations in these conditions are indicated in the descriptions of specific experiments. GTPyS bound to protein was determined by dilution of the samples with ice-cold filtration buffer (20 mM Tris-C1, pH 8, 100 mM NaCl, 25 mM MgC12) followed by rapid filtration of the samples through BAS5 nitrocellulose filters (Schleicher and Schuell). The filters were then washed 4 times with 2 ml of the same buffer (25). Filters were dried and dissolved in 10 ml of Liquiscint (National Diagnostics) for analysis of retained radioactivity.
Gs and the j3 subunits were assayed as described previously (10,27). One unit of Gs activity is defined as the amount of Gs that will reconstitute the production of 1 nmol of cAMP/min in cyc-membranes. Proteins were determined by staining with Amido Black as described by Schaffner and Weissman (30) with bovine serum albumin as the standard.
ADP-ribosylatwn of G Proteins by ZAP-The procedure used for labeling was the same as that described (13). Purified G proteins were ADP-ribosylated in a total volume of 80 pl containing 75 m M Tris-C1, pH 8.0, 1 mM EDTA, 1 mM dithiothreitol, 2.7 mM MgC12, 10 mM thymidine, 1 mM ATP, 5 p~ NAD, ["PINAD (l0,OOO cpm/pmol), 1 mg/ml dimyristoyl phosphatidylcholine, and 25 pg/ml IAP. In some cases, 100 p~ GTP was included, residual detergent from G protein samples did not exceed 0.015% Lubrol12A9 or 0.02% sodium cholate. The reaction was allowed to proceed for 60 min at 30 'C. At 60 min, an aliquot of 10 pl was diluted into 0.5 ml of 2% SDS, and the protein was precipitated by the addition of 0.5 ml of 30% trichloroacetic acid. Precipitated protein was collected on BA85 nitrocellulose filters and washed with a total of 20 ml of 6% trichloroacetic acid. The filters were dissolved in Liquiscint and analyzed for radioactivity. The remainder of the sample (70 pl) was treated with 20 pl of 10% SDS to halt further reaction. Samples were then treated with N-ethylmaleimide and sample buffer and subjected to analysis on SDS gels as described below.
SDS-Polyacrylamide Gels-Electrophoresis of polypeptides through 11 or 15% polyacrylamide gels was accomplished with the discontinuous system described by Laemmli (31). Samples were prepared as follows. Proteins in 20 pl of TED/l% cholate were mixed with 5 pl of 1 m M dithiothreitol and 5 pl of 10% SDS. After warming to 90 "C for 2-5 min, the samples were cooled, mixed with 10 pl of 10 m M N-ethylmaleimide, and incubated at room temperature for 15 min. Sample buffer (160 pl of 40 mM Tris-C1, pH 6.8, 1% SDS, 4% j3-mercaptoethano1, and 40% glycerol) was then added, and samples were heated at 100 "C for 5 min. The treatment with N-ethylmaleimide results in polypeptide bands of sharper clarity and, thus, better resolution; this is presumably due to prevention of the formation of intrachain disulfide bonds during electrophoresis for long periods of time at roam temperature. Molecular weight standards were obtained from Bio-Rad. Proteins were visualized with either Coomassie Blue or silver (32).
Materials-GTPyS and [=S]GTPyS were obtained from Boehringer Mannheim and New England Nuclear, respectively. [cy-Sap]ATP and [Y-~'P]GTP were synthesized as described by Johnson and Walseth (33); [TINAD was synthesized by the method of Cassel and Pfeuffer (19). Other nucleotides were obtained from a variety of sources and checked for purity by thin-layer chromatography. Lubrol 12A9 was purchased from IC1 and deionized prior to use. Heptylamine-Sepharose was prepared by the method of Shaltiel (34) with modifications (9). IAP was the most generous gift of Toshiaki Katada and Michio Ui, Hokkaido University, Sapporo, Japan.

RESULTS
We have used' GTPyS-binding activity as an assay to identify and purify GTP-dependent regulatory proteins in membranes from bovine brain. The low concentration of GTPrS (1 p~) and the filtration method used limit the assay to detection of proteins with a relatively high affinity (or slow dissociation rate) for the nucleotide. This assay has been used to characterize Gs (25) and GI (14, 28), two proteins that mediate stimulation and inhibition of adenylate cyclase, respectively. Table I compares the amount of binding of GTPyS in cholate extracts of crude membranes from rat and bovine brain with extracts of membranes from other tissues and S49 Membranes were prepared as described under "Experimental Procedures." Extracts were made in TED/l% cholate at 10 mg protein/ml. lymphoma cells. A striking feature is the high level of binding in the two preparations from brain. Much less binding was observed in crude membranes from rabbit liver, a tissue used for preparation of purified Gs and GI. Somewhat higher quantities of binding were observed in extracts from partially purified plasma membranes from bovine heart and S49 lymphoma cells; these levels are still considerably less than those seen in crude membranes from brain. This result contrasts with the amounts of Gs extracted from these membranes. The highest GS activity was found in partially purified plasma membranes from bovine heart. The crude membrane preparations from rabbit liver and bovine brain yielded lower but probably comparable amounts of Gs. The significance of the low yield of Gs from membranes of rat liver and rat brain is not known.
The significance of the high quantity of binding of GTPrS in membranes of bovine brain became more apparent when this activity was purified. Fig. 1 demonstrates that the majority of the extracted binding activity was bound to DEAE-Sephacel and eluted as a single peak (sotid circles) that was partially resolved from the peak of GS activity. The behavior of this peak of GTPyS-binding activity is similar to that reported for liver GI (assayed by ADP-ribosylation with IAP) by 14). The remainder of the binding activity eluted with higher salt concentrations but without any other obvious peaks to indicate the presence of another protein species in high quantity.
The peak of GTPyS-binding activity was pooled, concentrated, and subjected to filtration through Ultrogel AcA 34 (Fig. 2). Again, one major peak of GTPyS-binding activity was eluted at about the same position as residual Gs in the preparation. However, the broadened descending portion of the binding peak suggests some heterogeneity of the proteins responsible for this binding activity; the material that eluted more slowly was not included in the pool of the binding peak (see "Experimental Procedures") and has not been characterized further at this time.
The result of chromatography of the pool of GTPyS-binding activity through heptylamine-Sepharose is shown in Fig.  3. T w o peaks of GTPyS-binding activity were obtained, and both were resolved from Gs activity. Analysis of the polypeptide composition of fractions that contain GTPyS-binding activity is shown in Fig. 4. The first peak (fractions 34-39) was composed to a large degree of one major polypeptide with a molecular weight of 39,000. The specific activity of -30 nmol/mg of protein across this peak (Fig. 3) suggests that 1 mol of nucleotide binds/mol of polypeptide. Essentially pure 39,000-Da polypeptide could be obtained by passing this front peak through the heptylamine-Sepharose column a second time.
The second peak of GTPyS-binding activity had a lower  specific activity (-15 nmol/mg of protein) and contained three major polypeptides with molecular weights of 41,000, 39,000, and 36,000. A smaller polypeptide of 35,000 Da was also observed; the amount of this polypeptide was variable; some preparations yielded more equivalent amounts of the 35,000-and 36,000-Da polypeptides. The specific activity of this second peak is consistent with the presence of heterodimers of a and /3 subunits as has been observed for purified Gs (10) and GI (13). The existence of larger complexes with multiple binding sites is unlikely in view of the dimeric size observed during gel filtration (Fig. 2). A summary of the purification of these proteins is shown in Table 11. About 8% of the total binding activity in membranes was obtained in the two peaks obtained from heptylamine-Sepharose. Purifications of 80-and 40-fold were obtained for the two respective peaks. While a yield of 8% may n (2 GTP-binding proteins from bovine brain with G I from rabbit liver and T from bovine rod outer segments. Samples of purified G proteins were prepared as described under "Experimental Procedures" and subjected to electrophoresis through a 0.75-mm slab of 11% polyacrylamide. Lanes 1-6 were stained with Coomassie Blue; lanes 7-12 were stained with silver. Applied samples were: j3 subunit purified from bovine brain, seem low for such a simple purification procedure, it should be noted that 14% of the extracted activity was recovered as purified protein and that the peaks of GTPyS-binding activity have been deliberately pooled to eliminate some of the fractions containing Gs. The total yield of 19 mg of purified Gprotein was still possible due to the high concentration of these proteins in the crude membranes. Analysis of the results of the purification and the initial binding activity suggest that the a subunits of these two GTP-binding proteins constitute about 1% of the total protein in the crude membranes from brain. How do these purified proteins relate to similar GTPbinding proteins? Fig. 5 compares the polypeptide structure of the proteins purified from brain with GI purified from rabbit liver and transducin purified from bovine rod outer segments. The 41,000-Da polypeptide from brain has the same mobility as G,(Y from rabbit liver; the 39,000-Da polypeptide has a mobility similar to the a subunit of transducin. The major /3 subunit (36,000 Da) in the brain preparations shows a similar mobility to the /3 subunit of transducin and the larger /3 subunit of liver GI. The existence of a j3 subunit doublet in preparations of Gs from liver has been observed previously (10); the current method of sample preparation allows the consistent resolution of these two polypeptides, and both are observed routinely in preparations of Gs and GI from this laboratory.' The significance of the two /3 subunits is not clear, and any specific segregation with specific a subunits has not been determined. For purposes of discussion, we will refer to the separate / 3 subunits as /336 and ps5. Further experimentation will be required to determine whether 8% and j335 are products of the same gene, whether they differ functionally, or whether both polypeptides actually possess the activities defined for the /3 subunit (27).
These preparations were also examined for the presence of a y subunit. Fig. 6 shows a 15% acrylamide gel, which allows 'The doublet of j3 subunits is also observed routinely in the laboratory of Alfred G. Gilman (personal communication).  Fig. 5). All preparations were derived from bovine brain. detection of polypeptides of lower molecular weight. A potential y subunit was observed in all preparations that contained @ subunits; it was not found in fractions containing pure a subunit. The apparent association of the y and @ subunits mimics the behavior of the y and p subunits of transducin (8) and GI from liver.3 The polypeptide from brain has a molecular weight of about 11,000 in this SDS gel system and has the same apparent size as the suggested y subunit in liver GI (14). The significance of multiple polypeptides in the y region in some preparations of the / . ? subunit (Fig. 6) is not known. The comparative analysis of these purified proteins on SDS gels has led us to identify, tentatively, the 41,000-Da polypeptide as the a subunit of GI in brain. The 39,000-Da polypeptide does not migrate with GI nor does it have activities associated with Gs. Therefore, we have simply labeled this protein Goa for the "other" GTP-binding protein. A more meaningful designation will have to await determination of a definite function for this protein. Go is thus the designation for the oligomeric complex of the 39,000-Da polypeptide with 8, and possibly y, subunits. Direct evidence for the interaction of Goa with 0 will be shown in Fig. 9.
A common feature of the a subunits of GI and transducin is a site for ADP-ribosylation by NAD and IAP, a toxin from B. pertussis. Both a subunits obtained from bovine brain are G. M. Bokoch, and A. G. Gilman, personal communication. substrates for such ADP-ribosylation. Fig. 7 shows an autoradiograph of an SDS gel analysis of various proteins that were treated with IAP and ["PINAD. The a subunits of Go and GI from brain proved to be excellent substrates with about 1 mol of ADP-ribose incorporated/mol of a subunit. GI from liver and transducin labeled less well, with stoichiometries of only 0.3 and 0.6 mol of ADP-ribose/mol of a subunit, respectively. The brain preparation that contained both the 41,000-and 39,000-Da subunits (lanes E and F) is shown with Coomassie Blue stain in the outer lanes of Fig. 7. The extent of labeling of Goa and G p in this preparation was consistent with the protein content and suggests that both polypeptides in brain are labeled equally well. The exclusion of GTP during the reaction had no effect on the ADP-ribosylation of Goa by IAP; this behavior is similar to the modification of GI and transducin by IAP. Attempts to ADP-ribosylate these proteins with cholera toxin and NAD have failed. ADP-ribosylation factor, a protein required for ADP-ribosylation of purified Gs, was included in these experiments (36,37).
The behavior of Go through purification and its ADPribosylation by IAP but not cholera toxin closely resembles GI. Therefore, one explanation for the existence of Goa (39,000) in brain membranes could be proteolytic cleavage of G1a (41,000). Analysis of the products and time course of digestion of the two polypeptides with trypsin ( Fig. 8) suggests that this is not the case. Goa, GI (enriched in the 41,000-Da polypeptide) from bovine brain, @ from bovine brain, and GI from rabbit liver were exposed to trypsin for varying lengths of time under identical conditions (in the presence of MgZ+ and GTPyS). Goa was digested rapidly to a polypeptide of 38,000 Da; this polypeptide was remarkably stable as shown by its continued presence even after 6 h of exposure to trypsin. Samples at 0.3 mg/ml protein in 25 mM Tris-C1, p H 8, 1 mM EDTA, 1 mM dithiothreitol, 50 mM NaCI, 30 mM MgSO., 10 p~ GTPyS, and 1% sodium cholate (liver GI also contained 0.04% Lubrol) were incubated for 1 h a t 30 "C. At zero time, an equal volume of tosylphenylalanyl chloromethyl ketone trypsin (0.3 mg/ml) was added and digestion allowed to proceed at 30 "C. At the times indicated, aliquots of the digestion mixtures were removed and mixed with soybean trypsin inhibitor (8 times the amount of trypsin by weight). Aliquots of these samples were then processed and analyzed by SDS-gel electrophoresis and silver stain as described under "Experimental Procedures." Samples of brain Goa, brain GI, and liver GI were the same as those shown in Fig. 5. Purified 6 subunit from brain was the same sample as shown in Fig. 6; this preparation of is actually enriched in the component of higher mobility (8%).
GIa from liver was also digested rapidly to polypeptides of 39,000 and 38,000 Da. However, these polypeptides did not display the same stability as the 38,000-Da fragment from Goa and were digested to polypeptides of about 30,000 Da; total digestion occurred within 6 h. To eliminate the possibility that GI from brain and GI from liver have different susceptibilities to trypsin, a preparation of brain GI was digested. GIa (41,000 Da) from brain appeared to be digested with a time course similar to that of Gla from liver. Again, intermediate digestion products were observed in the 30,000-Da region; digestion is almost complete at 6 h. The small amount of 38,000-Da product observed at 6 h was consistent with the digestion of Goa that was present in this preparation.
Under these same conditions, the /3 subunits are digested with extreme rapidity, and no visible products were detected except near the dye front.
The different susceptibility of the two a subunits to digestion with trypsin strongly suggests that the two polypeptides are distinct entities. It seems unlikely, although not impossible, that a proteolytic cleavage of 41,000 to 39,000 Da would generate a product that is much more stable to the effects of trypsin and is not capable of producing detectable intermediates in the range of 30,000 Da. If a cleavage of this nature is the explanation for Goa, the event would have to be well controlled, since the yields of the two subunits were highly consistent for several preparations. Furthermore, the 39,000-Da polypeptide does not appear to be just a random proteolytic product of Gra, since it is also present in preparations from other sources. The major GTP-binding proteins from membranes of rat brain and bovine heart have been purified to about 50% of homogeneity (5-8 nmol of GTPyS bound/mg of protein). Both preparations contained major constituents at 41,000 and 39,000 Da as well as the /3 subunits (data not shown). This indicates that Goa exists in membranes from different tissues and species.
Goa offers some unique opportunities for studying the mechanisms of regulation of the GTP-binding proteins. This is demonstrated by the time course of association of GTP+ with isolated Goa that is shown in Fig. 9. Binding occurred in the absence or presence of Mg2' and was rapid relative to rates reported for GS (10,25) and GI (14) from rabbit liver. If the purified /3 subunit was added to this reaction in the absence of M e , the binding was markedly slowed and reduced in extent. The rate of binding to the resolved a subunit was similar both in the presence and absence of M$+; however, the extent of binding was increased by the divalent cation. The addition of / 3 subunit to samples of a that have reached near-equilibrium levels of binding caused a rapid reversal of this binding in the absence of M P ; the amount of bound GTPyS was reduced to levels that were observed when /3 was included initially. GTPrS binding in the absence of Mg2+ could also be reversed by adding excess guanine nucleotide (data not shown). No reversal was observed in the presence of Mg2' (the apparent rise in the level of binding upon addition of / 3 may reflect a stimulation by /3 of the rate of binding of GTPyS to a very high-affinity site in the presence of M P ) . If Mg2' was then added to samples of Goa and /3 that had no divalent cation (at 60 min), bound GTPyS increased at a rate similar to that observed for association to the free subunit.
This experiment demonstrates several properties of purified

Goa. 1) Purified Goa does interact well with /3 subunits. 2)
The binding of GTPyS to this protein is readily reversible under some conditions, and it suggests that a real equilibrium is observed in the absence of M P . 3 added (0,0, A). GDP > ITP >> GMP, ATP, CTP, UTP. At 10 nM GTPyS (KO = 6 nM for these conditions), the concentrations of nucleotides required for half-maximal inhibition were: GTP, 0.3 KM; GDP, 0.5 p~; ITP, 8.0 KM; GMP, ATP, CTP, UTP, >1 mM. Thus, Goa is a binding protein specific for guanine nucleotides. The affinity of Goa for GTPyS is variable with conditions; the apparent binding constant in the presence of M$+ appears to be less than 1 nM, and this would be a higher affinity than has been observed for the dimeric forms of Gs (25) or GI (14). Finally, a GTPase activity has been observed with this subunit; the rate is low (0.1-0.3 mol of Pi produced/ min/mol of Goa), but the reaction is catalytic.

DISCUSSION
We have reported the purification of two major GTPbinding proteins from membranes of bovine brain. The isolation procedures are essentially the same as those used to purify GS and GI from rabbit liver (9,10,13,14). This confirms further the utility of the procedures for the isolation of these GTP-binding proteins from other tissues. The ease of purification is the direct result of the high quantity of the two proteins in membranes from brain. The purification profiles indicate that about 50% of the GTPyS-binding activity (0.2 nmol/mg) observed in these membranes is due to the two purified proteins (labeled Go and GI). Therefore, with an assumed molecular weight of about 80,000, the two proteins together account for more than 1.5% of the total protein of the membrane. Twenty mg of purified protein was obtained from 12 g of membranes even with stringent pooling of peak activities. Other experiments suggest that a more generous collection of fractions containing binding activity will result in similar purification of the proteins with a greater yield. Thus, bovine brain is an excellent source for obtaining the large quantities of protein that will facilitate structural and biochemical studies of these and related G proteins.
The structure of the purified G proteins from bovine brain and their activity as substrates for ADP-ribosylation by IAP identify them as probable members of a family of membraneassociated proteins that bind GTP and mediate the regulation of intracellular functions by external stimuli. Like Gs, GI, and T, the G proteins from brain have at least two subunits (a, p) and possibly a third ( 7 ) . Direct comparison of these polypeptides on SDS gels suggests that the larger a subunit from brain is GIa; demonstration that this protein can act as an inhibitor of adenylate cyclase will be required to confirm this. The function of the smaller a subunit (Goa) remains unknown. Its migration in SDS gels with a mobility very similar to that of the a subunit of transducin is provocative. Initial experiments, however, indicate that the two proteins are different. Go aggregates in the absence of detergent, and attempts to remove Go from membranes with GTPrS have failed (the 3-5% release of guanine nucleotide binding activity that was observed may indicate that some solubilization of GI or Go occurs, but this contrasts strongly with the efficient 60-70% removal of transducin from membranes of rod outer segments under the same conditions). Therefore, while Go and T may be related, their physical behavior differs.
While Goa could be a proteolytic product of Gla, our data suggest that this is unlikely. Digestion of the two polypeptides with trypsin results in quite different patterns of product formation. While it is possible that GI that has been clipped by a protease may be in an altered conformation that is more resistant to digestion, this seems unlikely. It is even less probable that the clipped polypeptide would not be digested to any intermediates resembling those observed during digestion of the original protein. It should also be noted that Gla is rapidly reduced to a polypeptide with a size similar to G1a prior to further digestion by trypsin; this intermediate does not show the stability of GIa to further proteolytic cleavage.
Additional evidence that suggests that Go is not a random proteolytic product of GI is the appearance of a 39,000-Da polypeptide in preparations of partially purified G proteins from bovine heart and rat brain. In the latter case, care was taken to remove and chill the brains swiftly and to prepare membranes in the presence of at least one protease inhibitor and EDTA. Finally, the consistent appearance of both the 41,000-and 39,000-Da polypeptides in several preparations suggests that proteolysis is either not involved or else is very reproducible.
The possible presence of Goa will have to be considered in the evaluation of two types of experiments that are utilized to examine the G proteins in membranes. First, the use of ADP-ribosylation for detection or quantitation of G I In * membranes may not provide valid measurements due to the ability of Goa to be ADP-ribosylated by IAP and the similar mobilities of Goa and G I~ on SDS gels. Labeling of Goa can, therefore, be easily misinterpreted as labeling of GIa. Second, the ability of GO to bind guanine nucleotides readily must be considered in studies that examine binding of these nucleotides to membranes. Since the interaction of Go with receptors and its function in the membranes is unknown, the action of this protein may obscure or mimic effects of GI.
Go should be useful for studies of the interaction of guanine nucleotides with G proteins. This is the first of these proteins that has been purified in a highly stable form in the presence of cholate and the absence of effector molecules (i.e. GTP, F-, MgZ+). Storage of the protein in cholate (a detergent that can be removed easily) is desirable for many potential methods of reconstitution. One reason for the stability of Go is probably the high concentration of the molecule that can be maintained during purification; conditions that promote dissociation of subunits of other G proteins (including dilution) have resulted in increased inactivation rates of the proteins (38). This, however, cannot account for the stability of the purified a subunit of Go; this stability appears to be an intrinsic property of the protein from brain.
The a subunit of Go (39,000) can be obtained in a pure form without the aid of activating agents such as M e , guanine nucleotide analogues, or fluoride. This is probably due to a lower affinity between the a and p subunits of Go under the conditions utilized for chromatography through heptylamine-Sepharose. The 41,000-Da polypeptide (presumed GIa from brain) does run as a dimer with p under these same conditions; this parallels the behavior of liver GI (13,14). The ability to obtain Goa without the intercession of small regulatory ligands provides us with the opportunity to study this protein in a native state. We can study the interaction of the a subunit of this G protein with guanine nucleotides in the absence of other subunits and without having to depend on reversing the effects of activating ligands to which the protein was exposed previously. Thus, initial kinetics can be studied readily. We have already observed several properties of Goa that relate to the mechanism of G-protein activation. One of these is the association of GTPyS with Goa in the absence of M$+. The binding is readily reversible and can be influenced by the simple addition of p subunit. This experiment suggests that a real equilibrium is being observed directly (this has not been the case for preparations of GS and GI). Addition of p subunit then appears to reduce the affinity of the a subunit for GTPyS in the absence of M$+. M$+ prevents the effects of the @-subunit and apparently causes Goa to have a very high affinity for the nucleotide. These observations coincide with those of the dimeric form of GI and Gs (10,14,25). While confirming these data, we have also demonstrated how well this protein lends itself to experimental manipulation. We expect that more extensive studies will yield new and useful information on the mechanisms utilized by Go and other members of this family of GTPbinding proteins. At this time, we can only speculate on the function of GO. Its relationship to Gs, GI, and transducin suggests that it is involved in the conversion of extracellular signals to intracellular regulation. Further experiments are required to determine the prevalence of Go in other tissues, the exact relationship of Go to GI and transducin, and the potential function of Go as a regulator of adenylate cyclase or as a modulator of some other cellular process under hormonal control.