Interactions of phosducin with the subunits of G-proteins. Binding to the alpha as well as the betagamma subunits.

The high affinity interactions of phosducin with G-proteins involve binding of phosducin to the G-protein betagamma subunits. Here we have investigated whether phosducin interacts also with G-protein alpha subunits. Interactions of phosducin with the individual subunits of Go were measured by retaining phosducin-G-protein subunit complexes on columns containing immobilized anti-phosducin antibodies. Both the alpha and the beta subunits of trimeric Go were specifically retained by the antibodies in the presence of phosducin. This binding was almost completely abolished for both subunits following protein kinase A-mediated phosphorylation of phosducin and was reduced, more for alpha than for beta subunits, by the stable GTP analog guanosine 5'-(3-O-thio)triphosphate. Isolated alphao was also retained on the columns in the presence of phosducin but not in the presence of protein kinase A-phosphorylated phosducin. Likewise, purified G-protein betagamma subunit complexes as well as purified alpha subunits of Go and Gt were precipitated together with His6-tagged phosducin with nickel-agarose; this co-precipitation occurred concentration-dependently, with apparent affinities for phosducin of 55 nM (Gbetagamma), 110 nM (alphao), and 200 nM (alphat). In functional experiments, the steady state GTPase activity of isolated alphao was inhibited by phosducin by approximately 60% with an IC50 value of approximately 300 nM, whereas the GTPase activity of trimeric Go was inhibited by approximately 90% with an IC50 value of approximately 10 nM. Phosducin did not inhibit the GTP-hydrolytic activity of isolated alphao as measured by single-turnover assays, but it inhibited the release of GDP from alphao; the rate constant of GDP release was decreased approximately 40% by 500 nM phosducin, and the inhibition occurred with an IC50 value for phosducin of approximately 100 nM. These data suggest that phosducin binds with high affinity to G-protein betagamma subunits and with lower affinity to G-protein alpha subunits. We propose that the alpha subunit-mediated effects of phosducin might increase both the extent and the rapidity of its inhibitory effects compared with an action via the betagamma subunit complex alone.

Heterotrimeric GTP-binding proteins (G-proteins) comprise a family of regulatory proteins that couple transmembrane receptors for a variety of neurotransmitters, hormones, and other stimuli to their intracellular effectors (for reviews see Refs. [1][2][3][4]. G-proteins are composed of ␣, ␤, and ␥ subunits. Upon activation by agonist, receptors couple to their G-proteins and promote the exchange of bound GDP for GTP. The resultant conformational change induces dissociation of the GTPbound ␣ subunit (G ␣ ) 1 from the ␤␥ subunit complex (G ␤␥ ). In this dissociated state G ␣ activates effectors such as adenylyl cyclases, other enzymes, or ion channels. Additionally, the free ␤␥ subunits also interact with and regulate effectors (reviewed in Refs. 2 and 5). The intrinsic GTPase activity of G ␣ then leads to hydrolysis of the bound GTP, and this enables reassociation of GDP-bound G ␣ with G ␤␥ . The resulting reformation of the trimeric G-protein terminates the signal.
A similar GTPase switch function occurs in the small molecular weight monomeric GTP-binding proteins. For these proteins a large array of regulatory proteins have been identified that regulate various steps in the GTPase cycle (6). More recently, regulatory proteins have also been discovered for the heterotrimeric G-proteins. Among these are the growth cone protein GAP-43 and the RGS (regulators of G-protein signaling) family members that have been shown to activate the GTPase activity of several G-protein ␣ subunits (7-10) and a number of proteins that contain the structural motif of the pleckstrin homology domain and that bind to G-protein ␤␥ subunits (11), most notably the ␤-adrenergic receptor kinases (12)(13)(14)(15).
Phosducin is another type of regulator of G-protein signaling that is thought to act via the ␤␥ subunits of G-proteins (16,17). It was initially discovered as a major retinal phosphoprotein that could be copurified with the ␤ and ␥ subunits of transducin, G t (18). Its expression had been thought to be restricted to the retina and the developmentally related pineal gland (18 -20), but more recently phosducin has been shown to be ubiquitously expressed (16,21).
The molecular mechanisms of the interaction of phosducin with G-proteins are not clear. The co-purification of phosducin from the retina with the ␤␥ subunit complex of G t (18) indicated high affinity interactions between these proteins, and both the N and the C terminus of phosducin appear to contain high affinity binding sites for the ␤␥ subunit complex (22)(23)(24)(25). From these functional as well as structural observations it has been concluded that phosducin interacts exclusively with the ␤␥ subunits of G-proteins. Indeed, studies on the effects of phosducin and phosducin-like protein on G t suggest that phosducin acts by "trapping" free ␤␥ subunits; the GDP-bound ␣ subunits would then be unable to find free ␤␥ subunits for reassociation and would therefore stay inactive (17,26,27).
In contrast to the many studies on phosducin-G ␤␥ interactions, potential interactions with G-protein ␣ subunits have not been investigated. Such direct effects on ␣ subunits may be proposed from the observation that the extent of inhibition of G o GTPase is often larger than would be expected from antagonism of the function of ␤␥ subunits alone. A possible solution to this problem would be a direct interaction of phosducin with ␣ subunits. Therefore, the present study was undertaken to define the nature of interactions of phosducin with the subunits of G-proteins, using mostly G o as a model system.

EXPERIMENTAL PROCEDURES
Protein Purification-Recombinant phosducin was expressed in Escherichia coli as described earlier using the plasmid pET-phosducin (16). Following lysis of the bacteria by freeze thawing and precipitation of DNA with 2% streptomycin and centrifugation at 50,000 ϫ g for 10 min, the supernatant was applied to a MonoQ column (Amersham Pharmacia Biotech) and eluted with a 0 -500 mM NaCl gradient in 10 mM Tris-HCl, pH 7.4. Peak fractions of phosducin eluting at Ϸ250 mM NaCl were concentrated to Ϸ1 ml and were then further purified by gel filtration on Superdex 200 (Amersham Pharmacia Biotech). The purity of the preparations was Ͼ95% as determined by SDS-polyacrylamide gel electrophoresis and staining with Coomassie Blue. Mock preparations from E. coli transformed with "empty" vector were used as controls as described earlier (16).
Alternatively, phosducin was expressed in E. coli with a C-terminal His 6 -tag and purified on Ni-NTA-columns (Qiagen) as recently described (25). This resulted in apparently pure preparations as judged by Coomassie-stained SDS-polyacrylamide gels.
G o and its resolved ␣ o and ␤␥ subunits were purified from bovine brain according to the methods described by Sternweis and Robishaw (28). The amounts of G o or ␣ o were determined by [ 35 S]GTP␥S binding as described by Freissmuth and Gilman (29). Contaminating ␤␥ subunits were removed from the ␣ o preparation by a second round of heptylamine-Sepharose chromatography. The absence of contaminating ␤ subunits from these ␣ o preparations was verified in Western blots using a ␤ common antibody kindly provided by Dr. G. Schultz (Department of Pharmacology, Free University, Berlin, Germany). The ␣ and ␤␥ subunits of G t were purified to apparent homogeneity from bovine retina according to a protocol adapted from Gierschik et al. (30).
Phosphorylation of Phosducin-The catalytic subunit of protein kinase A was purified to apparent homogeneity from bovine heart by the method of Sugden et al. (31). 1 g of purified phosducin was incubated with 2 units of the PKA catalytic subunit in a buffer containing 10 mM Tris-HCl, pH 6.5, 5 mM MgCl 2 , 0.1 mM EDTA, 70 mM KCl, 0.5 mM dithiothreitol, 0.1 mM ATP, 5 mg/ml bovine serum albumin, 0.1% Tween 80, 0.2 mM benzamidine, 0.2 mM phenylmethylsulfonyl fluoride, and 1 g/ml leupeptin for 30 min at 0°C and 30 min at 30°C. At the end of the phosphorylation procedure, PKA inhibitor peptide was added to a final concentration of 10 M. In preliminary experiments we verified that this was sufficient to completely suppress further PKA activity. Control samples were treated identically, but the PKA catalytic subunit was absent during the incubations. The extent of phosphorylation was determined by addition of a small amount of [␥-32 P]ATP and measurement of the incorporation of 32 P into phosducin; phosphorylated phosducin used in the present experiments contained 0.8 -1.0 mol phosphate/mol phosducin.
Binding of G-proteins and Phosducin to Anti-phosducin Antibody Columns-An antiserum against phosducin was generated by immunization of rabbits with purified recombinant phosducin, and phosducinspecific antibodies were isolated by affinity chromatography (21,32). 2 ml of antiserum were incubated with 1 ml of wet protein A-Sepharose (Amersham Pharmacia Biotech) for 1 h at room temperature under gentle rocking. The resin was washed twice with 10 volumes of 0.2 M sodium borate, pH 9.0, and was suspended in 10 volumes of 0.2 M sodium borate, pH 9.0. The cross-linker bis(sulfosuccinimidyl)suberate (Pierce) was then added to a concentration of 1 mg/ml. After shaking for 30 min at room temperature, the reaction was stopped by washing the resin with several volumes of 0.2 M ethanolamine, pH 8.0, followed by incubation with 0.2 M ethanolamine, pH 8.0, for 2 h at room temperature. The resin was then washed with 5 volumes of 100 mM glycine-HCl, pH 2.5, and 40 volumes of phosphate buffered saline.
To study binding of G-proteins or their subunits to phosducin, equal molar amounts of phosducin and G o (or ␣ o as indicated) were preincubated for 10 min at room temperature and then added to 0.5 ml of the anti-phosducin resin together with 200 l of 20 mM Hepes, pH 7.4, 5 mM MgSO 4 , 0.1% Lubrol. The suspension was incubated under gentle rocking for 12 h at 4°C and the resin was subsequently washed with 5 ml of phosphate buffered saline/0.5% Tween 20, followed by a buffer change to 10 mM Tris-HCl, pH 6.8 (5 ml). The entire washing procedure was completed in less than 5 min. Phosducin and proteins bound to phosducin were then eluted with 1 ml of 100 mM glycine-HCl, pH 2.5. After neutralization and drying, these proteins were resolved by SDSpolyacrylamide gel electrophoresis and identified by Western blotting with specific antibodies against the ␣ or ␤ subunit (Signal Transduction Laboratories). Peroxidase-coupled secondary antibodies and enhanced chemoluminescence reagents (Amersham) were used to develop the blots.
Co-precipitation of Phosducin-His 6 and G-protein Subunits with Ni-NTA-Agarose-As an alternative to antibody immobilization of phosducin in the phosducin/G-protein binding assays we used immobilization of hexahistidine-tagged phosducin on Ni-NTA-agarose. Purified phosducin-His 6 (20 nM) was incubated with various concentrations of G ␤␥ or ␣ o purified from bovine brain, or ␣ t purified from bovine retina, in phosphate-buffered saline containing 0.05% cholate (140 mM NaCl, 30 mM KCl, 6.5 mM Na 2 HPO 4 , pH 7.3) for 30 min at 4°C. The mixture was centrifuged at 14,000 rpm for 10 min, and 40 l of Ni-NTA resin (Qiagen) were added to the supernatant to bind phosducin-His 6 (plus associated proteins). After 10 min at 4°C, the resin was pelleted by centrifugation and washed twice in the same buffer with intervening short centrifugations. G-protein subunits retained together with phosducin-His 6 were detected by taking up the beads in SDS-sample buffer followed by SDS-polyacrylamide gel electrophoresis and Western blotting as above.
Determination of GTPase Activity-The steady state GTPase activity of G o or ␣ o was measured as described earlier by determining the formation of 32 P i from [␥-32 P]GTP (33). 0.1 pmol (2 nM) of G o or ␣ o was assayed in buffer containing 0.1% Lubrol and 25 mM MgSO 4 . Phosducin was present at various concentations. The incubations lasted for 30 min at 30°C and were terminated by addition of 500 l of 1% charcoal in 2 mM NaH 2 PO 4 , pH 2.
The catalytic activity of ␣ o was measured in single-turnover assays adapted from Freissmuth and Gilman (29). Data Analysis-Kinetic data were fitted assuming exponential functions as described earlier (34). Concentration response curves were fitted to the Hill equation, and binding data were analyzed with a nonlinear binding analysis program as described (34,35).
The concentration response curve of the GTPase inhibition of G o by phosducin (see Fig. 6) was also analyzed assuming an interaction of phosducin with both the ␣ subunit and the ␤␥ subunit complex.
where A denotes the GTPase activity (in the percentage of the control activity), I ␣ and I ␤␥ denote the maximal inhibition caused via the ␣ or ␤␥ subunits, respectively, K ␣ and K ␤␥ denote the affinities of phosducin for the ␣ or ␤␥ subunits, respectively, and P is the concentration of phosducin. All data are derived from at least three independent experiments and represent means Ϯ S.E. unless stated otherwise.

RESULTS
To demonstrate directly the binding of phosducin to the subunits of G o , we incubated phosducin with G o or ␣ o in solution and then immobilized phosducin to detect which subunits were bound to phosducin. This was done by covalently coupling phosducin-specific antibodies raised in rabbits (32) to protein A-Sepharose, which allowed specific retention of phosducin (and proteins bound to phosducin) on small columns. In preliminary experiments we verified that the antibodies showed no cross-reactions with any of the subunits of G o , so that the G o subunits should only be retained on the columns if they were bound to phosducin. Phosducin (plus associated proteins) was then eluted from the columns with 100 mM glycine-HCl, pH 2.5, and the eluates were analyzed for the presence of ␣ and ␤ subunits of G o in Western blots. (We did not analyze the presence of ␥ subunits due to a lack of high affinity antibodies and also because the ␤␥ subunit complex does not appear to disso-ciate under physiological conditions.) Fig. 1A shows that using trimeric G o the ␣ as well as ␤ subunits were retained on the column in the presence of phosducin and that there was virtually no retention of either ␣ or ␤ subunits when mock preparations were used instead of phosducin. The retention of the ␣ as well as the ␤ subunits was sensitive to phosphorylation of phosducin by PKA (Fig. 1B). For both subunits, PKA-mediated phosphorylation decreased the amount of retention by more than 80% as determined by densitometric analysis. This corresponds roughly to the 80 -100% stoichiometry of phosducin phosphorylation that was achieved with PKA in these assays (see "Experimental Procedures") and suggests that phosphorylated phosducin binds the subunits of G o weakly if at all.
Preincubation of G o with GTP␥S to dissociate ␣ o from the ␤␥ subunits markedly reduced the amount of retained ␣ o but caused only a minor reduction of retained ␤ subunits (Fig. 1C). This result is compatible with the notion that phosducin binds to the ␤␥ subunit complex with high affinity, whereas binding to ␣ o is of low affinity. However, this experiment does not indicate whether the residual binding of ␣ o is due to incomplete dissociation of G o by GTP␥S or whether it does indeed represent direct low affinity binding of ␣ o to phosducin.
Direct low affinity binding of ␣ o to phosducin was then demonstrated with isolated ␣ o . The preparations of ␣ o used in these experiments were devoid of residual ␤␥ subunits by the criterion of Western blots (data not shown). Incubation of such ␣ o with phosducin led to the retention of ␣ o on the column, but, as expected for a low affinity interaction, the amounts of ␣ o retained were small (Fig. 2). Again, this binding was essentially abolished when PKA-phosphorylated phosducin was used (Fig. 2).
To confirm direct binding of ␣ subunits to phosducin, a second strategy involving another method to purify and to immobilize phosducin was used. To this end, a hexahistidine tag was added to the C terminus of phosducin, and phosducin-His 6 was purified to apparent homogeneity via binding of this hexahistidine tag to Ni-NTA-agarose. When 50 nM phosducin-His 6 were co-incubated with various concentrations of purified G ␤␥ and then pelleted with Ni-NTA-agarose, G ␤␥ were co-precipitated as detected by Western blots (Fig. 3A). Much less G ␤␥ was precipitated in the absence of phosducin, indicating a relatively low nonspecific binding of G ␤␥ to the resin. A semiquantitative analysis of this experiment by densitometry (Fig. 3B) revealed that the binding of G ␤␥ to phosducin-His 6 was saturable with an apparent affinity of 25 nM. Analysis of four similar experiments gave an average apparent affinity of 55 Ϯ 22 nM.
Analogous experiments were then done with purified ␣ o (Fig.  4). Again, there was a phosducin-His 6 -dependent precipitation of ␣ o with Ni-NTA-agarose. A semiquantitative analysis showed saturable binding of ␣ o to phosducin-His 6 with an ap- parent affinity of 85 nM (Fig. 4), and seven similar experiments gave an average value of 110 Ϯ 12 nM.
Because G t and its subunits can be purified better than G o (and also because phosducin was initially discovered in the visual system) such experiments were also carried out with ␣ t , which had been purified to apparent homogeneity from bovine retina (Fig. 5). These experiments gave essentially similar results as those obtained with ␣ o , but higher concentrations of ␣ t were required. The apparent affinity was 160 nM in the experiment shown in Fig. 5 and 200 Ϯ 53 nM in a total of four separate experiments.
These data provide direct evidence for binding of phosducin not only to the ␤␥ subunit complex but also to G-protein ␣ subunits. Functional assays were then used to determine the effects of such binding to ␣ subunits, again using ␣ o because of its substantial intrinsic GTPase activity. Fig. 6 shows the effects of phosducin on the steady state GTPase activity of ␣ o and trimeric G o . Phosducin inhibited the steady state GTPase activity of ␣ o in a concentration-dependent manner; maximal inhibition was Ϸ60%, and the IC 50 value was somewhat higher (Ϸ300 nM) than expected from the direct binding assays. As already shown earlier for trimeric G o (16) no such inhibition was seen with PKA-phosphorylated phosducin (data not shown). These results demonstrate a direct inhibitory effect of phosducin on the GTPase activity of ␣ o in agreement with its direct binding.
The GTPase activity of trimeric G o under the same conditions (i.e. basal activity in 0.1% Lubrol), was inhibited by phosducin up to Ϸ90% with an IC 50 value of Ϸ10 nM (Fig. 6). Assuming that this inhibition is caused both by a direct effect of phosducin on ␣ o and the well known effect via G ␤␥ , we also analyzed this inhibition curve, assuming that it has a ␤␥-dependent and an ␣-dependent component as described under

Phosducin-G ␣ Interaction
from the direct effects of phosducin on ␣ o ; likewise, the extent of inhibition (Ϸ19/(100 -73), i.e. Ϸ70%) is similar to the Ϸ60% inhibition seen in the GTPase assays with ␣ o alone. This twocomponent fit is statistically significantly better (p Ͻ 0.01 by F-test) than the simple one-component fit, but the differences between the two curves are very small.
The GTPase cycle contains two main steps: hydrolysis of GTP to GDP and release of GDP (followed by very rapid binding of GTP). The effects of phosducin on the first step in ␣ o were assayed in single-turnover experiments. In these assays, [␥-32 P]GTP is bound to G o in the absence of Mg 2ϩ , conditions that suppress the enzymatic activity of G-proteins. The addition of Mg 2ϩ plus a large excess of a stable GTP analog then allows the monitoring of the hydrolysis of the bound [␥-32 P]GTP. Fig. 7 shows that even a high concentration (2 M) of phosducin had no appreciable effect on this catalytic step. GTP hydrolysis occurred with a rate constant of 4.8 Ϯ 0.4 min Ϫ1 in the absence and 3.9 Ϯ 0.4 min Ϫ1 in the presence of phosducin. The amount of [␥-32 P]GTP hydrolyzed was almost identical under the two conditions (0.51 Ϯ 0.01 pmol/pmol G o under control conditions versus 0.56 Ϯ 0.02 in the presence of phosducin). These data indicate that the GTP-hydrolytic step of G o is not significantly affected by phosducin.
However, phosducin did impair the GDP release step of isolated ␣ o . The presence of 0.5 M phosducin reduced the rate constant of GDP release from isolated ␣ o from 0.23 Ϯ 0.01 to 0.14 Ϯ 0.01 min Ϫ1 (Fig. 8). Concentration response curves for this inhibitory effect were done by monitoring the amount of GDP remaining bound to ␣ o 5 min after initiation of the release reaction (Fig. 9). Phosducin caused a concentration-dependent increase in the amount of bound GDP, and this effect was half-maximal at Ϸ100 nM. This concentration is in agreement with the apparent affinity of phosducin for ␣ o determined in the direct binding assays (Fig. 4B). DISCUSSION Phosducin is a widely distributed inhibitor of the GTPase activity of trimeric G-proteins (16,17,21,22,26). In addition to inhibition of the enzymatic activity of G-proteins themselves, it has also been shown to inhibit G-protein-mediated activation of adenylyl cyclase in the ␤-adrenergic receptor system (16) and of cGMP phosphodiesterase in the rhodopsin system (17). In intact cells its overexpression impairs cAMP production induced by stimulation of ␤-adrenergic receptors (32). In the present study we have addressed the role of the G-protein subunits in this process and in particular whether in addition to the well demonstrated interactions with G-protein ␤␥ subunits phosducin might also have effects on the ␣ subunits.
Our data indicate that phosducin can indeed interact directly with G-protein ␣ subunits. This was shown in direct binding experiments as well as in functional assays. In the direct binding experiments two different factors were used to immobilize phosducin: specific affinity-purified antibodies and a C-terminal hexahistidine tag. These two strategies first permitted the use of two different initial purification procedures for phosducin and second should result in very different kinds of nonspecific binding of the G-protein subunits. With the antibody method no detectable amounts of G-protein subunits were re- Phosducin-G ␣ Interaction tained with the use of mock preparations from E. coli instead of phosducin (Fig. 1A), indicating first that the antibodies showed no significant reactivity toward any of the G-protein subunits and second that the retention of G-protein subunits was not due to a contaminating protein. We conclude that these assays did indeed measure the interaction of phosducin with G-protein subunits. With the hexahistidine tag method there was detectable nonspecific precipitation of both G ␣ and G ␤␥ (Figs. 4 -6), presumably due to the fact that the washing conditions could not be chosen as harsh as in the antibody method. However, this nonspecific binding was relatively low and allowed the clear detection of saturable specific binding of G ␤␥ , ␣ o , and ␣ t to phosducin. The demonstration of direct binding to ␣ t as well as ␣ o is important for two reasons. First, ␣ t can be purified to apparent homogeneity, indicating that no other proteins are required to effect phosducin-G ␣ interactions. Second, it indicates that phosducin might affect G-protein ␣ subunits also in the visual system.
The analysis of these binding assays by densitometry can be only semiquantitative due to the nonlinearity of the detection method. However, the apparent affinities correlate fairly well with those determined in functional assays; the affinities of phosducin for G ␤␥ reported by various groups using various methods is in the range of 8 -80 nM (16, 22, 23, 36 -38). The value found in our direct binding assay was 55 nM, which is in the upper range of the reported functional values. It is about 10-fold better than the previously obtained values utilizing G ␤␥ directly immobilized on microtiterplates, which presumably resulted in distortion of G ␤␥ (37).
The apparent affinity of phosducin for ␣ o determined in the direct binding assay was 110 nM, whereas in the functional assays IC 50 values of 100 nM (GDP release) and 300 nM (GTPase) were obtained; the analysis of the inhibition of G o -GTPase with a ␤␥and an ␣-dependent component gave an IC 50 value of the ␣-component of 120 nM. These affinities agree within the accuracy of the methods used.
The affinity of phosducin for ␣ o is lower than that for the ␤␥ subunit complex. However, an affinity in the range of 100 -300 nM is still below the concentration of phosducin in most tissues, which is about 1 M (21). Thus, this affinity is sufficient to allow phosducin-G ␣ interactions in vivo. Likewise, the high concentrations of phosducin in the retina (18 -21) are above the levels required for the low affinity interaction with ␣ t . Furthermore it appears that these interactions do indeed contribute to the inhibition of G-protein function by phosducin; the G o GTPase activity was inhibited by phosducin by up to 90% (Fig.  6). Under the conditions of these assays (in particular 0.1% Lubrol and high Mg 2ϩ ), G ␤␥ causes a 2-4-fold activation of the GTPase activity of ␣ o (39). Thus, if phosducin acted only to "trap" G ␤␥ , it should inhibit the GTPase activity of G o by at most 50 -75%. This is clearly less than the observed inhibition of 90% and suggests an additional mode of inhibition. The analysis of this inhibition curve with two components (␣ and ␤␥) did indeed result in a significantly better fit. Even though the improvement is only small, this two-component fit is entirely compatible in qualitative and in quantitative terms with the other data obtained here; it gives a G ␤␥ -mediated inhibition by 73% with an IC 50 value of 7 nM, compatible with the ranges given above, and a G ␣ -mediated inhibition of the remaining activity by 70% with an IC 50 value of 120 nM, compatible with the affinities and the extent of inhibition seen in assays with isolated ␣ o .
The functional effects of phosducin on ␣ o consist in an inhibition of GDP release but no effect on the catalytic step of the GTPase cycle. Thus, the effects of phosducin on G o appear to be composed of a direct inhibition of GDP release from ␣ o , and an antagonism of the effects of the ␤␥ subunit complex on the function of ␣ o . Because the ␤␥ subunits have stimulatory effects on ␣ o under activating conditions, such as in the presence of high Mg 2ϩ concentrations, mastoparan, or active receptors (39), the direct and the G ␤␥ -mediated effects of phosducin on ␣ o are additive under the conditions of stimulated G o activity.
The effects of phosducin on G ␣ should not only cause an increased inhibition of G-protein function compared with G ␤␥mediated effects alone, but they should also increase the rapidity of this inhibition. This is because trapping of G ␤␥ subsequent to G-protein activation would leave GTP-bound G ␣ free to interact with effectors and disrupt the G-protein cycle only after the first round of GTP hydrolysis. In contrast, a direct effect on G ␣ might already affect signaling in this first cycle of G-protein activation. Taken together these data suggest that the interactions of phosducin with ␣ o are of functional relevance.
Direct interactions of phosducin with G-protein ␣ subunits were not observed by Lee et al. (17) and Yoshida et al. (26) in their investigations on transducin (G t ). These authors found no comigration of phosducin and ␣ t on gel filtration columns and no effect of phosducin on the binding of ␣ t to rod outer segment membranes. There are two possible explanations for this discrepancy. First, we found that the affinity of phosducin for ␣ t is lower than that for ␣ o . Second, the assays used by these authors, which involve physical separation of proteins by chromatography or by centrifugation and washing, may be less sensitive to interactions of low affinity than the assays used here. In fact, gel filtration chromatography of phosducin has been shown to involve interactions with the matrix (26) that might well result in disruption of low affinity phosducin-␣ subunit interactions, and the affinity of phosducin for G t reported in binding assays with rod outer segment membranes (26) is Ϸ10-fold lower than that found in our assays.
In our hands, phosphorylation of phosducin by PKA impaired binding to G ␤␥ as well as to ␣ o . This was seen both in direct binding and in GTPase assays. The data about the effect of phosphorylation on the phosducin/transducin-␤␥ interaction are conflicting, depending on the assay used. Phosphorylated phosducin no longer coeluted with transducin-␤␥ from gel filtration columns but still inhibited transducin-␤␥ binding to rod outer segment membranes (26). It was concluded that phosphorylation might not alter the affinity of phosducin for transducin-␤␥ but rather affect the character of the interaction (26). However, Hawes et al. (22) observed a loss of G ␤␥ binding upon phosphorylation of phosducin similar to our data. Because phosphorylation of phosducin would be required to impair G ␤␥ binding by the C-terminal as well as the N-terminal binding sites, it is plausible to assume that it involves a major alteration of the structure of phosducin (24). Under these circumstances it is not surprising that phosphorylation of phosducin would also impair its interactions with G ␣ .
The molecular mechanisms of the interaction between phosducin and G ␣ remain to be elucidated. The crystal structure of phosducin complexed with the ␤␥ subunits of G t (24) suggests that phosducin binds with its C-terminal domain, particularly with an essential stretch of a few essential amino acids close to the C terminus (25), to the side of the ␤-propeller, whereas its less well defined N terminus is stretched out on the face of the propeller covering sites where G ␤␥ interacts with G ␣ . Detailed studies will be required to elucidate G ␣ -binding sites in phosducin. Furthermore, we do not know whether the interactions of phosducin with G o involve two separate binding events (one with G ␤␥ and another one with ␣ o ), two-step binding (first to G ␤␥ and then to the ␣␤␥ trimer), or a single composite reaction to form a tetrameric phosducin-␣␤␥ complex.
In summary, we believe that our data support interactions of phosducin with G ␤␥ as well as G ␣ . At low concentrations, phosducin appears to act preferentially by binding to G ␤␥ and by neutralizing G ␤␥ effects on G o activation. At higher concentrations, a direct inhibition of ␣ o causes additional inhibition of G-protein function. Both types of interactions occur with affinities in the lower range of physiological phosducin concentrations in many tissues. Direct effects of phosducin on G ␣ are predicted to have two major effects compared with the previously presumed exclusive action via G ␤␥ ; they would increase the extent as well as the rapidity of its inhibitory effects. Our observations suggest that complex interactions of phosducin with G-protein subunits play a role under physiological circumstances. These interactions provide an additional level to the many mechanisms (40) that regulate transmembrane signaling via G-protein-coupled receptors.