Chloride effects on Gs subunit dissociation. Fluoroaluminate binding to Gs does not cause subunit dissociation in the absence of chloride ion.

The stimulatory guanine nucleotide binding protein (Gs) is heterotrimeric (abg), and mediates activation of adenylyl cyclase by a ligand-receptor complex. The a subunit of Gs (Gsa) has a guanine nucleotide binding site, and activation occurs when tightly bound GDP is displaced by GTP. Together, GDP and fluoroaluminate (AlF4 ) form a transition state analog of GTP that activates Gs. The work of other investigators suggests that AlF4 2 causes subunit dissociation when it activates Gs. We have observed that in solution AlF4 2 did not cause Gs subunits to dissociate unless NaCl was also present. The effect of NaCl was concentration dependent (10–200 mM). Omitting F, Al, or Mg prevented the NaClinduced dissociation of Gs subunits. Na2SO4 could not substitute for NaCl in causing subunit dissociation, but KCl could, suggesting that the anion was responsible for the effect. Gs subunit reassociation occurred when the concentration of Cl was reduced even though the concentrations of AlF4 2 and Mg were maintained. The absence of Cl did not prevent AlF4 2 binding to Gsa. We have concluded that AlF4 , a ligand which is capable of activating G proteins, can bind to Gs in solution without causing subunit dissociation.

It has been more than 35 years since F Ϫ was first identified as being an activator of adenylyl cyclase (Rall and Sutherland, 1958). In the interim the heterotrimeric (␣␤␥) stimulatory G protein (G s ) 1 was identified as mediating activation of the enzyme by hormone-receptor complexes, by GTP and its analogs, and by F Ϫ (Howlett et al., 1979). In order for G s to activate adenylyl cyclase, GDP which is tightly bound to the guanine nucleotide binding site of the ␣ subunit (G s ␣) must be displaced by GTP or a nonhydrolyzable GTP analog such as GTP␥S or Gpp(NH)p. This process is facilitated by the hormone-receptor complex which explains its role in the activation of G s . The intrinsic GTPase activity of G s ␣ provided a means for terminating activation of the adenylyl cyclase (Gilman, 1987;Birnbaumer, 1990;Simon et al., 1991;Clapham and Neer, 1993). The mechanism by which F Ϫ activated G s remained a mystery until it was recognized that Al 3ϩ or Be 2ϩ was also required for the activation (Sternweis and Gilman, 1982). It is believed that together Al 3ϩ and F Ϫ form fluoroaluminate (AlF 4 Ϫ ), a phosphate analog, that binds to the guanine nucleotide binding site of G s ␣ and together with bound GDP mimics the effects of GTP (Chabre, 1990). Following the activation of G s by GTP analogs or AlF 4 Ϫ it is thought that G s ␣ dissociates from the G protein ␤␥ subunit complex (G␤␥). This dissociation is believed to be a critical part of the activation process. Contributing to the subunit dissociation hypothesis are data suggesting that activation of G s by AlF 4 Ϫ is accompanied by G s subunit dissociation Northup et al., 1983;Kahn and Gilman, 1984). However, we have recently reported that AlF 4 Ϫ does not cause G s subunits to dissociate (Toyoshige et al., 1994). In an attempt to reconcile our findings with those of other investigators we have discovered that Cl Ϫ is required for AlF 4 Ϫ -induced G s subunit dissociation. Here we report the experimental results of our investigation.
Experimental Treatment and Immunoprecipitation of G s -G s was prepared from bovine brain and stored at an estimated concentration of 450 g of G s /ml (Roof et al., 1985). For experiments, samples containing approximately 80 ng of bovine brain G s were incubated in 2 l of solution A (20 mM HEPES (pH 8.0), 1 mM EDTA, and 1 mM DTT) containing 0.1% Lubrol PX as described previously (Toyoshige et al., 1994) except that the incubation time was reduced from 2 to 1 h, and buffered solutions were prepared with the free acid of HEPES so that the pH could be adjusted with NaOH. The latter change was made in order to avoid the unintentional addition of Cl Ϫ ion. The incubations also included NaCl, Na 2 SO 4 , KCl, NaF, AlCl 3 , MgCl 2 , and/or MgSO 4 at the concentrations designated in the figures. Subsequently, the samples were either immunoprecipitated or used for zonal sedimentation.
For immunoprecipitation the samples were diluted to 100 l and treated as described previously (Toyoshige et al., 1994). During immunoprecipitation the salt concentrations were maintained, decreased or increased as indicated in the figures. Immunoprecipitates were assayed for G s subunits by SDS-polyacrylamide gel electrophoresis and immu-* 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.
noblotting (Toyoshige et al., 1994). In addition to the ␣ and ␤ subunits of G s , the immunoblots show a protein with slower electrophoretic mobility than G s ␣. This is the heavy chain of the RM/1 antibody. When percent dissociation of G s is reported, it is based on zero percent being defined as the amount of G␤ present when G s was immunoprecipitated following incubation in solution A containing 2 mM MgSO 4 and 0.1% Lubrol PX.
For zonal sedimentation samples of bovine brain G s incubated as described above were diluted to 100 l so that, with the exception of G s , the concentration of components in the solution were unchanged. The samples were then transferred onto the top of sucrose density gradients for zonal sedimentation experiments. G s from S49 membranes was also used for some zonal sedimentation experiments. Samples containing 250 g of membrane protein were diluted with one volume of solution A, and centrifuged at 43,000 ϫ g for 30 min at 4°C to collect the membranes. The membrane pellets were suspended in 100 l of solution A containing 0.1% Lubrol PX and the combination of salts indicated in the figure legend. The samples were incubated for 30 min at 30°C, and centrifuged at 43,000 ϫ g for 30 min at 4°C to remove insoluble material. The supernatant was then transferred onto the top of a sucrose density gradient for zonal sedimentation experiments as follows.
Zonal Sedimentation of G s on Sucrose Density Gradients and Adenylyl Cyclase Assays-Samples of bovine brain or S49 G s that had been prepared for zonal sedimentation were layered onto the top of 5 ml of linear density gradients of 5 to 20% sucrose made in solution B (20 mM HEPES (pH 8.0), 1 mM EDTA, 0.1 mM DTT, and 0.025% Lubrol PX) and containing other salts as described in the figure legend. The gradients were centrifuged at 50,000 rpm in an SW 50.1 rotor for about 20 h ( 2 t ϭ 2.0 ϫ 10 12 rad 2 /s). After centrifugation the gradients were divided into about 32 fractions of equal volume (approximately 160 l).
Sample volumes of 30 l from each fraction were then used to reconstitute adenylyl cyclase in the membranes of G s deficient S49 cyc Ϫ cells. Just prior to use the S49 cyc Ϫ membranes were thawed and diluted with a large volume of solution A containing 2 mM MgSO 4 . The membranes were recovered by centrifugation and suspended at a membrane protein concentration of 2.5 mg/ml in solution A containing 2 mM MgSO 4 . Ten l of the suspended S49 cyc Ϫ membranes were added to each sample, and after reconstitution of adenylyl cyclase, effector stimulated enzyme activity was assayed as described previously (Toyoshige et al., 1994). Effectors included 10 mM AlF 4 Ϫ for samples originally treated with AlF 4 Ϫ , and 30 M GTP␥S plus 20 M isoproterenol for samples of G s that were not activated before zonal sedimentation. No effector was added when samples of G s were activated with GTP␥S before zonal sedimentation.
Proteolysis of in Vitro Translated G s ␣-pBluescript II SKϩ containing the cDNA for the long form of rat olfactory G s ␣ was prepared and used to produce [ 35 S]methionine-G s ␣ by in vitro transcription and translation as described previously (Warner et al., 1996). After translation the sample was divided into two 25-l aliquots. One sample received 0.5 l (0.5 g) of brain G␤␥ and the other received an equivalent volume of solution containing 20 mM HEPES (pH 8.0), 0.1 mM EDTA, 1 mM DTT, and 0.25% Lubrol-PX. A solution of Lubrol PX was added to make the final detergent concentration 0.1% and to increase the reaction volume to 30 l. Samples were incubated at 30°C for 1 h, and then any NaCl that might have been present was removed by changing the solution for 30 l of solution C (20 mM HEPES (pH 8.0), 2 mM MgSO 4 , 1 mM EDTA, and 0.1 mM DTT) with a Centricon 30 ultrafiltration unit. Since Lubrol PX is retained during filtration, the final concentration of detergent remained 0.1%. Three-l samples were then adjusted to 12.5 l with solution C containing NaCl and/or AlF 4 Ϫ so that the final concentrations were 100 and 10 mM, respectively. The samples were incubated for 30 min at 30°C before adding 2.5 l of solution containing 150 g of trypsin/ml. The trypsin was made up in solution C containing 0.025% Lubrol PX with the appropriate concentration of NaCl and/or AlF 4 Ϫ . Samples that did not receive trypsin received the same volume of the appropriate solution without the protease. The samples were incubated for 15 min at 30°C before stopping the reactions by adding 1 l of solution containing 4 mg of soybean trypsin inhibitor/ml. The samples were prepared for electrophoresis by adding a solution containing ␤-mercaptoethanol and sodium dodecyl sulfate, but the samples were not heat-denatured (Jackson and Hunt, 1983) before applying them to a 10% sodium dodecyl sulfate-polyacrylamide gels.

RESULTS
RM/1 antiserum raised against a synthetic decapeptide corresponding to the carboxyl-terminal of G s ␣ has been used by us (Toyoshige et al., 1994) and others (Simonds et al., 1989;Morris et al., 1990) to immunoprecipitate G s . The amount of antiserum used for immunoprecipitation is of consequence since either too little or too much will decrease the efficacy of precipitation (Morris et al., 1990). It is reported that RM/1 will immunoprecipitate 30 to Ͼ90% of the G s present in detergent extracts of cell membranes (Simonds et al., 1989;Morris et al., 1990). Using conditions described previously (Toyoshige et al., 1994) we found that RM/1 precipitated 62% of the detected G s ␣ from preparations of bovine brain G s (Fig. 1). In the absence of RM/1 there was no detectable precipitation of G s ␣. Sample to sample variation in the amount of G s ␣ precipitated was Ϯ 12% (standard deviation for n ϭ 10 from a representative experiment), and there was no significant difference 2 between the amount of G s ␣ precipitated from samples of dissociated and undissociated G s (see for example Figs. 2 through 6). Similar results have been reported by other investigators (Morris et al., 1990). By using RM/1 to immunoprecipitate G s , we found that in the absence of NaCl, AlF 4 Ϫ was unable to cause G s subunit dissociation (Fig. 2). NaCl caused a concentration dependent dissociation of G s subunits in the presence of 2 mM MgSO 4 and AlF 4 Ϫ . Dissociation was easily detectable with 10 mM NaCl, and was nearly complete when it was 200 mM. In the presence of 2 mM MgSO 4 and AlF 4 Ϫ , 150 mM NaCl caused 76 Ϯ 2% dissociation when compared with samples of G s incubated in the absence of NaCl. In order to better understand what was required for G s subunit dissociation we varied the ion composition of the solutions during incubation and immunoprecipitation. Omitting AlCl 3 (Fig. 3A) or NaF (Fig. 3B) during the incubation and subsequent immunoprecipitation prevented G s subunit disso- FIG. 1. Efficacy of immunoprecipitation of G s with RM/1 antiserum. A sample of bovine brain G s was immunoprecipitated with RM/1 antiserum as described under "Experimental Procedures." The supernatant remaining after immunoprecipitation was concentrated with a Centricon 30 ultrafiltration unit that was treated with bovine serum albumin to block nonspecific binding of G s . The solute and Lubrol PX concentrations of the immunoprecipitated sample were adjusted to make them the same as that in the supernatant sample. As a control preimmune serum (PI) was substituted for RM/1 antiserum. All of the samples received ␤-mercaptoethanol and sodium dodecyl sulfate, and the proteins in the immunoprecipitate and supernatant were separated on 10% polyacrylamide gels. Detection of G s ␣ was accomplished as described under "Experimental Procedures." The preparation of bovine brain G s used for these studies contained both the short (G s ␣ 45 ) and long (G s ␣ 52 ) forms of G s , but the former predominated. The other bands in the immunoprecipitate and supernatant are the heavy chain of IgG and bovine serum albumin, respectively. ciation. G s subunit dissociation was also prevented if, at the time of immunoprecipitation, the NaF concentration was reduced by dilution to 200 M even though the concentration of AlCl 3 was maintained and 150 mM NaCl was added (Fig. 3B, last lane). Although not all combinations were tested, it appeared that the presence of NaF, AlCl 3 , and NaCl was required only during the immunoprecipitation to cause G s subunit dissociation (Fig. 3, A and B). MgSO 4 was also required since its omission prevented G s subunit dissociation in the presence of NaCl and AlF 4 Ϫ (Fig. 4). G s subunit dissociation occurred when KCl was substituted for NaCl but not when Na 2 SO 4 was substituted (Fig. 5). Reducing the NaCl concentration was sufficient to allow dissociated G s subunits to reassociate even though the concentrations of MgSO 4 and AlF 4 Ϫ were maintained, and the G s subunit concentration was reduced by 50fold as a consequence of dilution (Fig. 6).
To investigate AlF 4 Ϫ binding to G s ␣ we prepared [ 35 S]methionine-G s ␣ by in vitro transcription and translation of the cDNA for rat olfactory G s ␣. This technique produced the G s ␣ with a molecular mass of 52 kDa as well as two shorter products with molecular masses of 40 and 36 kDa (Fig. 7). The latter two proteins resulted respectively from initiation of translation of this cDNA at the codons for methionines 60 and 110. Only traces of the in vitro translation proteins were able to survive tryptic digestion in the absence of AlF 4 Ϫ . In the presence of AlF 4 Ϫ , a 37-kDa fragment was protected from proteolysis. The ability of AlF 4 Ϫ to protect this fragment was improved by the presence of NaCl although there was significant protection in the absence of this salt suggesting that AlF 4 Ϫ binding to G s ␣ did not require NaCl.
Since sedimentation on sucrose density gradients has been used to show that AlF 4 Ϫ causes G s subunit dissociation, similar experiments were performed for these studies. G s was incubated with AlF 4 Ϫ in the presence or absence of NaCl and subjected to zonal sedimentation. AlF 4 Ϫ was present throughout the sucrose gradients, and NaCl was included or omitted so as to be consistent with the way G s was treated before application to the sucrose gradients. In the absence of NaCl, G s from S49 cell membranes (Fig. 8A), and from bovine brain (Fig. 8B) sedimented as a heterotrimer despite the presence of AlF 4 Ϫ . For experiments with G s from bovine brain, it was necessary to substitute Na 2 SO 4 for NaCl when the latter was omitted from the gradients. This prevented the purified G s from aggregating, and sedimenting to the bottom of the centrifuge tube. However, the addition of Na 2 SO 4 to gradients reduced the sedimentation rate of heterotrimeric G s when compared with sedimentation through gradients that did not contain NaCl or Na 2 SO 4 (compare the sedimentation of heterotrimeric G s in Panels A and B of Fig. 8). The reduced rate of sedimentation was due, either in ؊ . Bovine brain G s was incubated as described under "Experimental Procedures" with the indicated concentrations of NaCl in the presence of 10 mM AlF 4 Ϫ and 2 mM MgSO 4 . The concentrations of NaCl and fluoroaluminate were maintained during the immunoprecipitation of G s ␣ with RM/1 antiserum. The amount of precipitated G s ␣ and G␤ was determined as described under "Experimental Procedures." FIG. 3. G s subunit dissociation requires the simultaneous presence of NaCl, NaF, and AlCl 3 at the time of immunoprecipitation. During the incubation step bovine brain G s was treated with 2 mM MgSO 4 as described under "Experimental Procedures" in the presence or absence of 150 mM NaCl, 10 mM NaF, or 10 M AlCl 3 as indicated in the figure. Samples were subsequently diluted at which time the concentration of these salts were either maintained, increased by addition, or decreased by dilution as indicated for immunoprecipitation. The concentration of MgSO 4 was maintained at 2 mM during the immunoprecipitation step. a At the time of immunoprecipitation the concentration of NaF was 200 M.

FIG. 4. G s subunit dissociation in the presence of AlF 4
؊ and NaCl requires magnesium ion. Bovine brain G s was incubated and immunoprecipitated as described under "Experimental Procedures" in the presence of 10 mM AlF 4 Ϫ , 150 mM NaCl, and/or 2 mM MgSO 4 as indicated in the figure. The concentration of each salt was maintained during the incubation and immunoprecipitation.

FIG. 5. Cation and anion effects on G s subunit dissociation.
Bovine brain G s was incubated and immunoprecipitated as described under "Experimental Procedures." Solutions for both the incubation and immunoprecipitation contained 2 mM MgSO 4 as well as 10 mM AlF 4 Ϫ plus either 150 mM NaCl, KCl, or Na 2 SO 4 as indicated.
FIG. 6. G s subunit reassociation can occur in the presence of AlF 4 ؊ . Bovine brain G s was incubated with 2 mM MgSO 4 as described under "Experimental Procedures." The incubations were done in the absence or presence of 10 mM AlF 4 Ϫ and 150 mM NaCl as indicated in the figure. Subsequently, the samples were diluted for immunoprecipitation, but the addition of 1 l of RM/1 antiserum was postponed for 1 h while the samples were incubated at 30°C. The dilutions were made in such a way that the MgSO 4 and AlF 4 Ϫ concentrations were not changed, and the NaCl concentration was either maintained or reduced to 3 mM (150/3) by dilution as indicated. After antiserum was added the immunoprecipitation was completed as described under "Experimental Procedures." part or entirely, to increased density caused by adding Na 2 SO 4 to the solutions used for making gradients.
In the presence of both AlF 4 Ϫ and NaCl, G s from S49 cells sedimented at the same rate as the free G s ␣-GTP␥S subunit (Fig. 8A) when the same conditions employed by other investigators were used (Kahn and Gilman, 1984). These conditions included the addition of 10 mM MgCl 2 during incubation and zonal sedimentation. When MgCl 2 was replaced with 2 mM MgSO 4 , G s from the S49 cell membranes sedimented at a rate intermediate between the free G s ␣-GTP␥S subunit and heterotrimeric G s (Fig. 8A). The same phenomenon was observed for G s from bovine brain even in the presence of 10 mM MgCl 2 (Fig.  8B), and the peak of G s activity was broad, extending to regions of the gradient where both the free G s ␣-GTP␥S subunit and heterotrimeric G s sedimented.

Heterotrimeric G proteins are activated when AlF 4
Ϫ and GDP form an analog similar to the transition state created when G␣ hydrolyzes GTP (Chabre, 1990;Sondek et al., 1994;Coleman et al., 1994). The activation is thought to be followed by dissociation of G␣ from G␤␥, and dissociation is considered necessary in order for G␣ to interact productively with its effector molecule. The activation of G s by AlF 4 Ϫ can be accompanied by subunit dissociation (Kahn and Gilman, 1984). However, we have reported that activation of G s by AlF 4 Ϫ / in the presence of 2 mM MgCl 2 did not cause G s subunit dissociation (Toyoshige et al., 1994). We speculated that the difference between our recent results and the earlier results of other investigators were a consequence of experimental design. In an effort to reconcile these differences, and to gain a better understanding of the effects of AlF 4 Ϫ and other ions on G s subunit interaction we conducted the experiments described in this article.
We found that dissociation of G s subunit in the presence of AlF 4 Ϫ required the simultaneous presence of 10 mM NaF, 10 M AlCl 3 , 2 mM MgSO 4 , and 10 -200 mM NaCl. Higher concentrations of NaF did not cause subunit dissociation in the absence of NaCl (data not shown), and dilution of NaCl to 3 mM allowed the G s subunit to reassociate even though the concentrations of MgSO 4 and AlF 4 Ϫ were maintained. The effects of NaCl are apparently due to Cl Ϫ . Other investigations (Higashijima et al., 1987) indicate that Cl Ϫ causes conformational changes in G proteins when they bind activating ligands, but not when they bind GDP. Recently, we have investigated the ability of ligands other than AlF 4 Ϫ to cause G s subunit dissociation. In the presence of low concentrations of Mg 2ϩ (2 mM or less) neither GTP nor GTP␥S caused G s subunit dissociation (Toyoshige et al., 1994, Basi et al., 1996. However, high concentrations of MgCl 2 (Toyoshige et al., 1994) and MgSO 4 (data not shown) do cause G s subunit dissociation both in the absence and presence of guanine nucleotides. While GTP and GDP were equally effective in attenuating the Mg 2ϩ -induced dissociation of G s sub- FIG. 7. Protection of in vitro translated G s ␣ from tryptic digestion by AlF 4 ؊ in the presence and absence of NaCl. [ 35 S]Methionine-G s ␣ was prepared by in vitro translation, and incubated in the presence or absence of G␤␥ as described under "Experimental Procedures." Subsequently, a Centricon 30 ultrafiltration unit was used to exchange the solution containing the [ 35 S]methionine-G s ␣ for one without NaCl and having 2 mM MgSO 4 . Then, 10 mM AlF 4 Ϫ and/or 100 mM NaCl was added as indicated, and the samples were incubated for 30 min at 30°C before digestion with trypsin as described under "Experimental Procedures." The proteolytic products were analyzed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis and autoradiography. The acute arrow indicates the core 37-kDa fragment of G s ␣ that is protected from proteolysis by AlF 4 Ϫ .
FIG. 8. Zonal sedimentation of G s through sucrose density gradients in the presence and absence of NaCl. G s from wild type S49 membranes (Panel A) or bovine brain (Panel B) was incubated as described under "Experimental Procedures." Samples were incubated with AlF 4 Ϫ in the presence (E, Ç) or absence (q) of NaCl and either 2 mM MgSO 4 (Ç, q) or 10 mM MgCl 2 (E). When samples of bovine brain G s (Panel B) were incubated in the absence of NaCl (q), 50 mM Na 2 SO 4 was substituted for the NaCl. Controls were as follows: 1) heterotrimeric G s was prepared by incubating G s with 2 mM MgSO 4 in the absence (Panels A) or presence (Panel B) of 50 mM Na 2 SO 4 and 2) the free G s ␣-GTP␥S subunit was prepared by incubating G s with 100 M GTP␥S, 120 mM MgSO 4 , and 100 mM NaCl. Before layering experimental and control samples of bovine brain G s onto the gradients, their volume was increased to 100 l in such a way that the concentration of components in the solutions, except for G s , were not changed. The volume for experimental and control samples of G s from S49 membranes was 100 l and therefore did not need adjusting before layering them onto gradients. The sucrose density gradients were made up in solution B and contained the same concentration of salts and effectors present in the samples except for the gradients used for the G s ␣-GTP␥S controls. These gradients were made without GTP␥S and they contained 2 mM MgSO 4 and 100 mM NaCl. After centrifugation, the gradients were divided into fractions and aliquants were assayed for their ability to reconstitute adenylyl cyclase in S49 cyc Ϫ membranes as described under "Experimental Procedures." The peak of reconstituted activity for G s ␣-GTP␥S and heterotrimeric G s are indicated by the labeled arrows. The first fraction represents the top of the gradient.