Expression of Phosducin in a Phosducin-negative Cell Line Reveals Functions of a G

,

Phosducin is a member of the large group of proteins that bind to G-protein ␤␥-subunits (G ␤␥ ) and whose biological functions are often unknown. Human A431 cells do not contain detectable amounts of phosducin. We generated A431 cells expressing phosducin at a level of Ϸ1 pmol/mg of cytosolic protein, which is Ϸ10% of the phosducin level in brain. cAMP accumulation in response to ␤ 2 -adrenergic receptor agonists was enhanced at early times in phosducin-expressing cells, but reached a lower plateau than in control cells. Permeabilization of the cells with digitonin did not change this pattern, but allowed the introduction of specific inhibitors: antibodies to phosducin abolished all differences between the two cell lines. Inhibitors of the ␤-adrenergic receptor kinase abolished the differences at early time points. An almost complete loss of ␤ 2 -adrenergic receptor desensitization in the phosducin-expressing cells was also observed when intact cells were desensitized and receptor function was then determined in membrane preparations. Inhibition of protein kinase A accentuated the effects of phosducin, suggesting that also in vivo phosducin is regulated by this kinase. These data indicate that phosducin affects G-protein-mediated signaling in at least two ways: it dampens the overall responsiveness, and it impairs the rapid desensitization mediated by the ␤-adrenergic receptor kinase.
A large array of membrane receptors utilize GTP-binding G-proteins to transmit their signals across membranes. These G-proteins are heterotrimeric proteins consisting of ␣-, ␤-, and ␥-subunits (1)(2)(3). They couple activated seven-transmembrane helix receptors to divergent effectors such as adenylyl cyclase and other enzymes or to various ion channels. These signaling pathways are highly regulated systems. Such regulation has been described mostly at the receptor level, where multiple mechanisms are capable of reducing receptor function at various speeds and for various periods of time (4,5).
Classically, most functions of G-proteins have been assigned to the ␣-subunits. However, over the past years it has become clear that the ␤␥-subunits (G ␤␥ ) have multiple signaling functions of their own (6 -8). G ␤␥ have been shown to interact with a growing number of divergent proteins, and these G ␤␥ -binding proteins have recently gained much interest both from a structural and functional point of view. From a structural point, several of the G ␤␥ -binding proteins appear to contain a pleckstrin-homology (PH) domain (9), a domain characterized by a six-membered ␤-barrel closed by a terminal ␣-helix (10,11). In two cases of PH domain-containing proteins, the ␤-adrenergic receptor kinase, and Ras guanine nucleotide releasing factor, the G ␤␥ -binding region has been mapped to the C-terminal segment of the domain, encompassing the ␣-helix and extending beyond it (12,13). A short consensus sequence Gln/Asn-X-X-Glu/Asp-Arg/Lys has been proposed as the essential determinant of G ␤␥ coupling (14).
On the functional level it has been shown that G ␤␥ -binding proteins can inhibit G ␤␥ -mediated signaling in reconstituted systems, in overexpressing cells and in transgenic mice (15)(16)(17). These experiments underline the importance of G ␤␥ -mediated signaling. However, they do not elucidate the physiological roles of G ␤␥ -binding proteins, since the overexpression resulted in non-physiological levels, and since the overexpressed protein was truncated. A more relevant strategy to elucidate their function would be the knock-out of a G ␤␥ -binding protein. However, a much simpler approach is the reintroduction of such a protein, at physiological levels, into a background where it is not present.
We have chosen phosducin as a G ␤␥ -binding protein. Phosducin was initially discovered as a major retinal phosphoprotein which could be copurified with the ␤and ␥-subunits of G t , the retinal G-protein (18). Its expression had been thought to be restricted to the retina and the developmentally related pineal gland (19). Phosducin was subsequently purified from bovine brain and its mRNA was identified in many tissues, suggesting that it is widely distributed (20). Recombinant purified phosducin has been shown to inhibit the GTPase activity of several purified G-proteins; furthermore, addition of phosducin to cell membranes inhibited the stimulation of adenylyl cyclase by ␤-adrenergic receptors or by G s . From these data we concluded that phosducin might be a widely distributed Gprotein regulator (20). Similar inhibitory effects were also observed for G t (21).
In contrast to most other G ␤␥ -binding proteins, phosducin has no other known enzymatic or signaling function, and may thus be regarded as a "pure" G ␤␥ -binding protein. Furthermore, it has a high affinity for G ␤␥ , suggesting that G ␤␥ -binding might be its primary physiological role. We wished to explore the functions that phosducin might have when present at physiological levels. For these studies, we took advantage of the fact that we could not detect phosducin in human A431 cells, a cell line widely used to characterize ␤-adrenergic receptor/G s -mediated signaling, and generated cell lines stably expressing phosducin at physiological levels.

EXPERIMENTAL PROCEDURES
Vector Construction-A 738-base pair NcoI-BamHI fragment containing the coding region for bovine phosducin was excised from the vector pET-phd (20), blunted with the Klenow polymerase fragment, and ligated into the blunted ApaI and SmaI sites of the expression vector pBC-KS-dhfr (22). The resultant vector, termed pBC-phd-dhfr, contains the phosducin cDNA under the control of the strong cytomeg-* This work was supported by the Deutsche Forschungsgemeinschaft, the European Commission, and the Fonds der Chemischen Industrie. 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. alovirus immediate early promoter, and the mouse dihydrofolate reductase (dhfr) gene under the control of the weaker SV40 promoter.
Generation of Phosducin-expressing Cells-Human A431 cells, subclone E3 (provided by EJM Helmreich, University of Wü rzburg), were grown in Dulbecco's modified Eagle's medium containing 10% fetal calf serum, glutamine (2 mM), penicillin (100 units/ml), and streptomycin (0.1 mg/ml). They were transfected with 20 g of pBC-phd-dhfr and 2 g of pSV2-neo using the transfection reagent N- [1-(2,3-dioleoyloxy)propyl]-N, N, N-trimethylammonium methylsulfate (Boehringer Mannheim). Two days after transfection the cells were subjected to a first round of selection using 800 g/ml of geneticin (Life Technologies, Inc.). The resultant resistant cell pool was then selected a second time with 0.3 M methotrexate (in the continued presence of geneticin). Resistant clones were then isolated and maintained at 0.3 M methotrexate and 400 g/ml geneticin. They were screened for phosducin expression using Western blots (see below). Corresponding control cell clones were generated by transfection with pBC-KS-dhfr instead of pBC-phd-dhfr and two identical rounds of selection.
Antibodies, Immunoprecipitation, and Western Blots-Antisera were raised against purified recombinant phosducin (20) in rabbits and goats. Rabbit antibodies were affinity-purified by binding to phosducin immobilized on Affi-Gel 15 columns (Bio-Rad), followed by elution with 100 mM glycine-HCl, pH 2. An IgG fraction was prepared from the goat antiserum by caprylic acid precipitation (23).
Immunoprecipitation of phosducin from cytosolic fractions of bovine brain or from A431 cells was done as follows. Cytosolic fractions were obtained by centrifugation of Ultra-Turrax disrupted tissue at 200,000 ϫ g for 20 min. An aliquot of the supernatant was incubated with prewashed protein A-Sepharose (Pharmacia) for 30 min at 4°C. After centrifugation, 5 g of affinity purified rabbit anti-phosducin antibodies were added to the supernatant and incubated for 1 h. The antibodies were bound to another portion of protein A-Sepharose and pelleted by centrifugation. After washing five times in 50 mM Tris-HCl, pH 7.4, containing 1% Brij 96 (Sigma), 50 mM NaCl, 10 mM EDTA, 20 mg/liter benzamidine, and 20 M phenylmethylsulfonyl fluoride, the samples were resuspended in SDS sample buffer, heated to 95°C for 5 min to release bound protein, and subjected to Western analysis.
For Western blots, cells were harvested in phosphate-buffered saline and disrupted by sonication. Cytosolic and particulate fractions were separated by centrifugation at 200,000 ϫ g for 20 min. Samples containing 200 g of protein per lane were separated by electrophoresis on 12% SDS-polyacrylamide gels and blotted onto Immobilon (Millipore) membranes. Goat antibodies (1:2,000 dilution) and horseradish peroxidase-coupled secondary antibodies (Dianova) were used in conjunction with ECL reagents (Amersham) to develop the blots.
cAMP Accumulation in Intact A431 Cells-Phosducin-expressing and control A431 cells were grown in 6-well Falcon tissue culture plates to 60 -80% confluence. They were incubated in Dulbecco's modified Eagle's medium containing 100 M of the phosphodiesterase inhibitor IBMX 1 for 30 min. The medium was removed, and the same medium containing 10 M (Ϫ)-isoproterenol was added to cause maximal stimulation of the cells' ␤ 2 -adrenergic receptors. After various times, the incubation was stopped by aspiration of the medium and addition of 2 ml of boiling water. The cells were then scraped off the plates and were separated from soluble material by centrifugation at 5,000 ϫ g for 10 min.
The cAMP content in the supernatant was determined by radioimmunoassay according to Harper and Brooker (24). This was done using a rabbit antiserum raised against cAMP coupled via a succinyl moiety in the 2Ј-position to bovine serum albumin. Samples and standards were 2Ј-O-acetylated with acetic anhydride and then measured using 125 I-cAMP-2Ј-O-succinyltyrosylmethylester (DuPont NEN) as the radioligand.
cAMP Accumulation in Permeabilized A431 Cells-In order to gain access to their cytosol, A431 cells were permeabilized with digitonin by a procedure modified from the one described earlier (25,26). In brief, 2 ϫ 10 5 A431 cells/well were incubated in Dulbecco's modified Eagle's medium supplemented with 100 M IBMX as described above. Cells were then washed with phosphate-buffered saline, followed by two washes with 150 mM potassium glutamate, pH 7.1, 5 mM EGTA, 7 mM MgCl 2 , 5 mM glucose, 2 mM ATP (KGA buffer), and 100 M IBMX and then kept with 1 ml of this buffer per well. The cells were then permeabilized by stepwise addition of digitonin to a final concentration of 0.015%. The degree of permeabilization was verified in one well by staining with trypan blue. Rabbit phosducin antibodies (see above, 1:200 dilution), the ␤ARK inhibitors heparin (100 nM) and the peptide KTAIAKFERLQTVTNYFITSE (50 M; Ref. 27), or the PKA inhibitor peptide PKI (10 M) were added immediately after the permeabilization. 10 min later (Ϫ)-isoproterenol was added and cAMP accumulation was measured as described above for intact cells.
␤-Adrenergic Receptor Desensitization in Intact A431 Cells-Desensitization of ␤ 2 -adrenergic receptors in intact A431 cells was effected by incubating the cells for 10 min at 37°C in Dulbecco's modified Eagle's medium containing 10 M (Ϫ)-isoproterenol. The incubation was stopped by washing the cells three times in ice-cold phosphate-buffered saline. Subsequently, crude membranes were prepared and the extent of desensitization was determined in adenylyl cyclase assays as described below. The adenylyl cyclase assays used to detect desensitization were done with millimolar concentrations of Mg 2ϩ , conditions that favor the detection of ␤ARK-mediated versus PKA-mediated desensitization (26,28).
Adenylyl Cyclase Assays in A431 Cell Membranes-Cells were harvested in 50 mM Tris-HCl, pH 7.4, and disrupted with an Ultra-Turrax homogenizer. Crude membranes were prepared by centrifugation of the homogenate at 50,000 ϫ g for 30 min, followed by resuspension in 50 mM Tris-HCl, pH 7.4, and a similar centrifugation step. The final pellet was resuspended in 1 ml of 50 mM Tris-HCl, pH 7.4. Adenylyl cyclase activity in the crude membranes was determined essentially as described (29). Incubations contained up to 50 g of crude membrane protein, 50 mM Tris-HCl, pH 7.4, 4 mM MgCl 2 , 1 mM EDTA, 100 M [␣-32 P]ATP (0.2 Ci/tube), 100 M cAMP, 10 M GTP, 5 mM creatine phosphate, 0.4 mg/ml creatine kinase, and 1 mg/ml bovine serum albumin. Various concentrations of (Ϫ)-isoproterenol were used to activate ␤ 2 -adrenergic receptors; 100 M forskolin was used to achieve receptorindependent activation of adenylyl cyclase. Incubations were done at 37°C for 30 min.
Other Procedures-␤ 2 -Adrenergic receptors in A431 cells were quantitated in saturation experiments using (Ϫ)-[ 125 I]cyanopindolol in concentrations up to 200 pM as the radioligand. Activity of ␤ARK was determined using urea-stripped rod outer segments (Ͼ95% rhodopsin) as the substrate with purified recombinant ␤ARK-1 was used as a standard as described (30). In order to separate ␤ARK from other kinases and from phosducin, the cytosolic fractions were passed over small DEAE-Sephadex columns (31).
G-protein-␤-subunits were quantitated in Western blots using a ␤ common antibody kindly provided by Dr. G. Schultz, Berlin, and a G ␤␥ purified from bovine brain according to Sternweis and Robishaw (32) as standard.
Data Analysis-Accumulation of cAMP in intact cells was analyzed using a mono-exponential equation of the type: A ϭ A i (1 Ϫ e Ϫkapptt ) ϩ A o , with A denoting the concentration of cAMP, A o the basal concentration of cAMP, and A i the agonist-induced accumulation of cAMP. k app is the time constant of cAMP accumulation. The parameters were estimated by non-linear curve-fitting as described earlier (33).
Adenylyl cyclase activities were normalized to the maximal activities determined in the presence of 100 M forskolin. In order to quantitate desensitization, concentration-response curves of isoproterenol-stimulated adenylyl cyclase activity were analyzed as described earlier (34) using the following algorithm (35), with E denoting the effect, E m the maximum possible effect, A the agonist concentration, K a the dissociation constant of the agonist-receptor complex. is a parameter describing the signal transduction efficacy of the system, and was estimated individually for the two curves, whereas all the other parameters were shared. 0 denotes the value under control conditions, and is the value for the desensitized curve.

RESULTS
In order to find a phosducin-negative cell line suitable for expression of phosducin and for the analysis of its effects on G-protein-mediated signaling, we tested a variety of cell lines for the presence of phosducin in Western blots with and without prior immunoprecipitation. Using these techniques, which had a sensitivity well below 10 fmol of phosducin, we could not detect any phosducin in human A431 cells (see Fig. 1), a cell line with a well characterized ␤-adrenergic receptor/G s signaling system. A431 cells stably expressing phosducin were then generated by transfection with the expression vector pBC-phddhfr followed by selection with geneticin and methotrexate. A control cell line was generated by similar transfection and selection using the control vector pBC-KS-dhfr. Fig. 1A shows the expression of phosducin in the two cell lines as visualized in a Western blot. Using purified recombinant phosducin as a standard, the extent of expression was estimated at Ϸ1 pmol of phosducin/mg of cytosolic protein. Subcellular fractionation studies showed that almost all phosducin was present in the cytosol, whereas only very weak phosducin immunoreactivity was found in particulate fractions (data not shown). In the control cell line, phosducin could not be detected with our antibodies in Western blots when up to 200 g of cellular protein were loaded per lane (Fig. 1); since less than 10 fmol of phosducin could be detected in such blots, this suggests a level of Ͻ50 fmol/mg protein in the control cells.
The level of phosducin expression was then compared with the endogenous level found in brain. This was done using immunoprecipitation followed by Western blots with another antibody, because in experiments with brain tissue we could not obtain clear bands in direct Western blots. The efficiency of this immunoprecipitation is Ϸ20%. 2 Much more phosducin immunoreactivity was found in brain than in the phosducinexpressing A431 cells, whereas there was no signal from control A431 cells (Fig. 1B). A semiquantitative analysis using recombinant phosducin as a standard (not shown) gave phosducin levels of Ϸ10 pmol/mg of cytosolic protein in bovine brain. Thus, the level of phosducin expression in the A431 cells was well within the physiological range.
The expression of G ␤␥ (quantitated in Western blots) and ␤ 2 -adrenergic receptors (determined by radioligand binding) was not affected by phosducin expression (Table I). Likewise, total ␤ARK activity measured in cytosolic preparations with light-activated rhodopsin as the substrate was identical in control and in phosducin-expressing cells (Table I).
In order to investigate the effects of phosducin on G-proteinmediated signaling, we measured the cAMP accumulation caused by stimulation of ␤ 2 -adrenergic receptors (Fig. 2). These experiments were done in the presence of the phosphodiesterase inhibitor IBMX to inhibit cAMP degradation. The ␤ 2 -adrenergic receptor agonist isoproterenol at 10 M caused a more than 100-fold increase of the cellular cAMP content. This increase was clearly reduced in phosducin-expressing cells. At steady-state, the cAMP levels were almost 3 fmol/control cell compared to just above 1 fmol/phosducin-expressing cell. This indicates that phosducin exerts an inhibitory effect on the ␤ 2 -adrenergic receptor/G s /adenylyl cyclase system. However, the kinetics of isoproterenol-induced cAMP accumulation were more rapid in the phosducin-expressing cells. The time constant k app was 0.21 Ϯ 0.02 min Ϫ1 in control cells, but 0.45 Ϯ 0.05 min Ϫ1 in phosducin-expressing cells. At early times of stimulation, the cAMP levels were actually higher in the phosducin-expressing cells (Table II). Thus, in addition to an overall reduction of cAMP accumulation, phosducin appeared to cause an enhancement at early time points.
In order to probe the intracellular mechanisms of these effects, we measured cAMP accumulation in permeabilized cells. This was done to allow the introduction of various inhibitors of intracellular proteins into the cell interior. To simplify these assays, cAMP was determined at three different time points after addition of isoproterenol: after 20 s (higher levels of cAMP in phosducin-expressing cells), 2 min, and after 20 min (lower levels of cAMP in phosducin-expressing cells). Table II shows that the overall effects of phosducin were the same in digitoninpermeabilized cells. The absolute values of cAMP/cell were very similar in permeabilized and intact cells, and the early enhancement and the later inhibition caused by phosducin were also present, even though these effects were slightly smaller than in intact cells.
As a first step, phosducin antibodies were added to the permeabilized cells to verify the specificity of the observed effects. These antibodies essentially abolished the differences between control and phosducin-expressing cells (Table II, Fig. 3). This suggests that both the early enhancement and the late inhibi-  Panel B shows phosducin immunoprecipitated from cytosolic preparations (6 mg of protein) of bovine brain, and of control and phosducinexpressing A431 cells. The immunoprecipitation was done with affinity purified antibodies generated in rabbits, and the blots were developed with the IgG-fraction of an antiserum raised in goats.

TABLE I Levels of G-protein ␤-subunits, ␤ 2 -adrenergic receptors, and ␤ARK
activity in phosducin-expressing and control A431 cells The amounts of G-protein ␤-subunits and ␤ 2 -adrenergic receptors present in crude membrane fractions were determined in Western blots developed with ␤ common antibodies using purified brain ␤␥ preparations as the standard, and in radioligand binding experiments with 125 Icyanopindolol. ␤ARK activity was determined in DEAE-Sepharose preadsorbed cytosolic fractions using rhodopsin as the substrate and purified recombinant ␤ARK-1 as the standard. tion of cAMP accumulation were indeed due to the expressed phosducin.
While the lower cAMP levels in phosducin-expressing cells were expected since purified phosducin inhibited isoproterenolstimulated adenylyl cyclase activity in cell membranes (20), the initial enhancement of cAMP levels in phosducin-expressing cells was more surprising. One possibility is an interference with the ␤ARK. This kinase causes a very rapid (t 1 ⁄2 Ͻ15 s) dampening of the cAMP response to ␤-adrenergic receptor stimulation in A431 cells (36), and in experiments with purified reconstituted components phosducin impairs receptor phosphorylation by ␤ARK (15). Therefore, we added two different inhibitors of ␤ARK to the permeabilized cells: heparin, which is the most potent inhibitor of ␤ARK with an IC 50 value of Ϸ10 nM (25,37), and the peptide KTAIAKFERLQTVTNYFITSE which has an IC 50 value of 7 M in inhibiting ␤ARK-mediated receptor phosphorylation (27). Both compounds abolished the early enhancement of cAMP levels caused by phosducin (Table II, Fig. 3). The peptide revealed an inhibitory effect of phosducin already at early time points, while heparin, possibly due to a slower diffusion into the cells, only enhanced the inhibitory effects of phosducin at 2 min (Table II, Fig. 3).
These experiments suggested that the expressed phosducin impaired ␤ARK-mediated receptor desensitization. To investigate this hypothesis further we measured such desensitization directly in a two-step experiment. ␤ 2 -Adrenergic receptors were desensitized in intact cells by incubation with 10 M isoproterenol for 10 min. Subsequently, membranes were prepared and desensitization was assessed by measuring isoproterenol-stimulated adenylyl cyclase activity. The pattern of desensitization in control cells (Fig. 4A) was very similar to that described earlier for untransfected A431 cells (26): the maximal stimulation of adenylyl cyclase activity by isoproterenol was reduced by about one-third, and the desensitized curve was shifted to the right. Analysis of the two curves with the algorithm described under "Experimental Procedures" indicates a 77 Ϯ 6% desensitization in control cells. In contrast, there was much less desensitization in the phosducin-expressing cells: the maximum was decreased by less than 10%, and there was barely a shift of the concentration-response curve (Fig. 4B). The quantitative analysis gave a desensitization of 33 Ϯ 13%. These results confirm the notion that phosducin impaired desensitization of the ␤ 2 -adrenergic receptors.
In addition, the data presented in Fig. 4 show that isoproterenol-stimulated adenylyl cyclase activity was lower in membranes from phosducin-expressing cells than in those from control cells. The activity elicited by 10 M isoproterenol was 27% of the activity in the presence of 100 M forskolin, whereas in membranes from control cells it was 35%. Forskolin-stimulated activities were the same in membranes from the two cell lines (Fig. 4, legend), indicating that these effects of phosducin were exerted at the receptor/G-protein level.
Phosphorylation of phosducin by protein kinase A (PKA) has been shown to abolish the effects of phosducin on G-protein function (20,38). This suggests that in our experiments enough unphosphorylated phosducin must have remained present in order to exert effects on cAMP accumulation. The peptide PKAinhibitor PKI was added to permeabilized cells to determine whether inhibition of PKA activity would cause an enhancement of the effects of phosducin expression. Fig. 5 shows that PKI caused an exaggeration of the phosducin effects: it increased the initial enhancement of cAMP levels and lead to greater inhibition of cAMP accumulation at later time points. These data indicate that phosphorylation of the expressed phosducin by endogenous PKA resulted in significant inhibi-

Isoproterenol-induced cAMP-accumulation in digitonin-permeabilized phosducin-expressing and control A431 cells
Phosducin-expresing and control transfected cells were preincubated for 30 min with 100 M of the phosphodiesterase inhibitor IBMX and were then, where indicated, permeabilized with digitonin. Phosducin antibodies (1:200 dilution) or the ␤ARK inhibitors heparin (100 nM) or the ␤ARK-inhibitor peptide KTAIAKFERLQTVTNYFITSE (50 M) were added, 10 M (Ϫ)-isoproterenol was added 10 min later, and cAMP levels were determined after the indicated periods of time always in the continued presence of 100 M IBMX. Data are mean Ϯ S.E.; n ϭ 6; paired t test: *, p Ͻ 0.05; **, p Ͻ 0.01.  3. Modulation of the effects of phosducin expression on cAMP accumulation in digitonin-permeabilized A431 cells by inhibitors. cAMP accumulation in digitonin-permeabilized phosducinexpressing and control transfected cells was measured in the presence of phosducin antibodies, or the ␤ARK inhibitors heparin (100 nM) or the ␤ARK-inhibitor peptide KTAIAKFERLQTVTNYFITSE (50 M). Phosducin effects were plotted as the ratio of the respective cAMP levels in phosducin-expressing and control cells as given in Table II. Open circles, control (no inhibitors); inverted triangles, phosducin antibodies; squares, heparin; upright triangles, ␤ARK-inhibitor peptide. Data are means Ϯ S.E. of six independent experiments with duplicate determinations. tion of phosducin. At the same time, PKI resulted in significantly higher absolute cAMP levels of both cell types (see legend to Fig. 5), which is compatible with a role of PKA in negative feedback in the ␤-adrenergic receptor system (4,5). DISCUSSION Proteins which interact with G ␤␥ have recently attracted much interest. Some of these proteins appear to be regulated in their activity by G ␤␥ , while others use this interaction as a means of membrane targeting. However, almost all of these studies have been done either with isolated proteins, or with expression systems which result in highly unphysiological levels. Thus, we know very little about the physiological role of such G ␤␥ binding. Phosducin may be regarded as a pure G ␤␥binding protein because, in contrast to other G ␤␥ -binding proteins, it appears to be devoid of other functions so that it may be used to assess the effects of G ␤␥ binding per se.
An investigation of the effects of phosducin at physiological expression levels was made possible by the fact that A431 cells, which contain a well studied ␤-adrenergic receptor/G s /adenylyl cyclase system, express little if any endogenous phosducin. Phosducin was then expressed at Ϸ1 pmol/mg of protein, which is Ϸ10% of its level in brain, and Ϸ2% of that of G ␤␥ , which are presumed to represent the primary target of phosducin. Expression of phosducin had two apparently opposite effects: it inhibited cAMP accumulation in response to long-term (several minutes) ␤ 2 -receptor stimulation, and it accelerated the cAMP response to short-term ␤ 2 -receptor stimulation. The first effect, inhibition of ␤ 2 -receptor-mediated cAMP-production, is analogous to the inhibitory effects of phosducin in reconstituted systems which were mentioned above. We assume that they are due to an interaction of phosducin with G s , which results in inhibition of G s activation and, consequently, in inhibition of G s -mediated signaling (20). In fact, the reduction of cAMP levels seen here is similar to the extent of G-protein inhibition observed earlier (20). A small inhibitory effect of the phosducin could also be observed in the membranes prepared from these cells (Fig. 4), which is compatible with the presence of small amounts of phosducin in the membrane fraction.
The second effect, the initial acceleration of the cAMP response to isoproterenol, appears to be due to an inhibition of ␤ 2 -adrenergic receptor desensitization. We have shown earlier that phosducin can inhibit receptor phosphorylation by ␤ARK in reconstituted systems (15). This inhibition is thought to be due to the fact that phosducin competes with the kinase for the G-protein ␤␥-subunits which serve as membrane anchors for the kinase (39 -41). Such an inhibition of ␤ARK-mediated receptor phosphorylation should result in reduced ␤ARK-mediated receptor desensitization. Since ␤ARK-mediated desensitization of ␤ 2 -adrenergic receptors in A431 cells occurs with a t 1 ⁄2 of less than 15 s (36), the lack of such desensitization should be observed very soon after receptor stimulation. Indeed, the enhanced cAMP levels in phosducin-expressing cells were already seen 20 s after the addition of isoproterenol. Assessment of ␤ 2 -adrenergic receptor desensitization in two-step assays is compatible with this hypothesis.
The two effects of phosducin combined result in a "sharpening" of the cAMP signal in response to isoproterenol. The initial acceleration in cAMP production followed by a reduced plateau result in a more rectangular response pattern which is reflected in a higher time constant of cAMP accumulation in the phosducin-expressing cells. These effects of phosducin are further complicated by the fact that they are antagonized by PKA-mediated phosphorylation of phosducin. It has been shown that phosphorylation of phosducin by PKA can result in marked inhibition of phosducin's effects in reconstituted systems (15,20,38). Phosducin in our A431 cells is apparently phosphorylated by endogenous PKA, since addition of PKI results in an enhancement of the effects of phosducin. However, inhibition of phosducin's effects by PKA was obviously incomplete in these cells, since otherwise no long-term effects of phosducin would be observable. Fig. 6 shows a diagram illustrating the position that phosducin may have in feedback loops of G-protein-mediated signaling.
It is remarkable that phosducin at an expression level of Ϸ1 pmol/mg of protein can inhibit G-protein-mediated signaling even though the level of G-protein ␤␥-subunits is about 50 times higher. This suggests that in some way phosducin may act preferentially on activated G-proteins, or that it interacts specifically with ␤␥-subunits which may be involved in ␤ 2adrenergic receptor/adenylyl cyclase signaling. Using a series of defined G-protein ␤␥-subunits we have not been able to detect major differences in their affinities for phosducin (42). The former hypothesis is supported by the observation that in a reconstituted retinal system the inhibitory effects of phosducin on G t -mediated signaling increased with time (21). This may be interpreted as evidence for a preferential interaction of phosducin with ␤␥-subunits which dissociate from ␣-subunits in response to receptor activation. Signal amplification at the receptor3 G-protein step is only a fewfold in most hormonal receptor systems (1,2,43). Thus, effective dampening of such responses would be expected to occur when the levels of a G ␤␥ -binding protein are similar to those of the respective receptors activated in response to stimuli.
Taken together our data indicate that phosducin at physiological levels has distinct effects on G-protein-mediated signaling in intact cells. Such effects would, depending on the expression level and their affinity for G ␤␥ , be common to all G ␤␥binding proteins. These proteins may thereby become additional components of G-protein systems and add another level of complexity to the many mechanisms regulating such systems.
FIG. 6. Proposed functional role of phosducin in the ␤ 2 -adrenergic receptor/G s /adenylyl cyclase/PKA system. Arrows with a ϩ symbol denote activation, bars with a Ϫ symbol indicate inhibition. The signaling chain from the receptor (␤ 2 AR) to PKA is based on a series of activating steps. This activation is impaired by ␤ARK at the receptor level, and by phosducin at the G ␤␥ level. The inactivating step mediated by ␤ARK is greatly enhanced by G ␤␥ . This effect of G ␤␥ is also antagonized by phosducin. PKA-mediated phosphorylation of phosducin inhibits all of its effects in this system.