Mechanism of Assembly of G Protein βγ Subunits by Protein Kinase CK2-phosphorylated Phosducin-like Protein and the Cytosolic Chaperonin Complex*

Phosducin-like protein (PhLP) is a widely expressed binding partner of the G protein βγ subunit complex (Gβγ) that has been recently shown to catalyze the formation of the Gβγ dimer from its nascent polypeptides. Phosphorylation of PhLP at one or more of three consecutive serines (Ser-18, Ser-19, and Ser-20) is necessary for Gβγ dimer formation and is believed to be mediated by the protein kinase CK2. Moreover, several lines of evidence suggest that the cytosolic chaperonin complex (CCT) may work in concert with PhLP in the Gβγ-assembly process. The results reported here delineate a mechanism for Gβγ assembly in which a stable ternary complex is formed between PhLP, the nascent Gβ subunit, and CCT that does not include Gγ. PhLP phosphorylation permits the release of a PhLP·Gβ intermediate from CCT, allowing Gγ to associate with Gβ in this intermediate complex. Subsequent interaction of Gβγ with membranes releases PhLP for another round of assembly.

Eukaryotic cells employ heterotrimeric G proteins to transduce a wide variety of hormonal, neuronal, and sensory signals that control numerous physiological processes. As a result, malfunctions in G protein pathways contribute to many diseases (1)(2)(3), and therapeutic agents targeting G protein-coupled receptors represent the single largest class of current pharmaceuticals (4). There are three fundamental steps in the propagation of a G protein-mediated signal. First, a ligand binds a receptor, resulting in a change in the packing of the seven transmembrane ␣-helices found in all G protein-coupled receptors. Second, the activated receptor catalyzes exchange of GDP for GTP on the ␣ subunit of a heterotrimeric G protein (G␣) 2 on the intracellular surface of the receptor. GTP binding causes G␣ to dissociate from the G protein ␤␥ subunit complex (G␤␥). Third, the G␣⅐GTP and G␤␥ complexes control the activity of effector enzymes and ion channels that regulate the intracellular concentration of second messengers (cyclic nucleotides, inositol phosphates, and Ca 2ϩ ) and the plasma membrane electrical potential (mainly via K ϩ channels). Changes in these properties in turn orchestrate the cellular response to the stimulus (5).
Phosducin-like protein (PhLP) is a member of the phosducin gene family (6 -8) that is believed to participate in G protein signaling by virtue of its ability to bind the G␤␥ dimer with high affinity (9 -11). Many in vitro and overexpression experiments have shown that PhLP binding to G␤␥ blocks its ability to interact with G␣ or effectors (9,10,(12)(13)(14). From these experiments, it was suggested that the physiological role of PhLP was to down-regulate G protein signaling by sequestering G␤␥. However, the results of several recent studies have seriously challenged this model. Specifically, disruption of the PhLP1 gene in the chestnut blight fungus Cryphonectria parasitica (15) and in the soil amoeba Dictyostelium discoideum (7) yielded the same phenotype as the disruption of the G␤ gene. Moreover, PhLP deletion blocked G protein signaling in Dictyostelium (7). In another study, the duration of opiate desensitization was prolonged in mice in which PhLP expression in the brain was inhibited by antisense oligonucleotide treatment (16). All of these observations are the exact opposite of what would be predicted by the G␤␥ sequestration model.
Insight into an alternative function of PhLP has come from the observation that PhLP interacts with the cytosolic chaperonin containing tailess complex polypeptide 1 (CCT), an essential molecular chaperone that mediates the folding of actin, tubulin, and other proteins into their native structures (17). PhLP was shown to interact with CCT as a regulator and not as a folding substrate. In addition, the cryoelectron microscopic structure of the PhLP⅐CCT complex (18) shows that PhLP binds CCT at the top of the CCT apical domains positioned above the folding cavity in a manner analogous to prefoldin, a CCT co-chaperone that binds nascent actin polypeptide chains and delivers them to CCT for folding (19). Coupling these observations with the fact that yeast G␤ (20) and other proteins with seven ␤-propeller structures similar to G␤ (21)(22)(23) interact with CCT suggests that PhLP might function as a co-chaperone in the folding of G␤. Indeed, recent findings show that PhLP does act as an essential chaperone for assembly of the G␤␥ dimer, demonstrating that the physiological function of PhLP is not to down-regulate G protein signaling by sequester-ing G␤␥ but to support G protein signaling by catalyzing G␤␥ dimer formation (24). However, the role of CCT in this process remains unclear (24,25).
Phosphorylation of PhLP by protein kinase CK2 (CK2) plays an important role in PhLP function. A major site of CK2 phosphorylation occurs within a sequence of three consecutive serines (residues 18 -20) near the N terminus (14). Phosphorylation of serines 18 -20 was required for PhLP-mediated G␤␥ assembly, for when these residues were substituted with alanine, PhLP was unable to catalyze G␤␥ dimer formation (24). The mechanism by which phosphorylation at these sites enhances G␤␥ dimer formation is not known; therefore, the effects of CK2 phosphorylation on PhLP function were investigated. The results of these studies provide evidence for a mechanism of PhLP-mediated G␤␥ assembly that involves the formation of a ternary complex between PhLP, the nascent G␤ polypeptide, and CCT. PhLP phosphorylation is required for the release of PhLP⅐G␤ from the CCT complex and the subsequent association of G␥ with G␤ to form the G␤␥ dimer.

EXPERIMENTAL PROCEDURES
Cell Culture-HEK-293 and CHO cells were cultured in Dulbecco's modified Eagle's medium/F-12 (50/50 mix) growth media with L-glutamine and 15 mM HEPES, supplemented with 10% fetal bovine serum (HyClone). The cells were subcultured regularly to maintain active growth but were not used beyond 20 -25 passages.
Preparation of cDNA Constructs-Wild-type human PhLP, PhLP ⌬1-75, and Pdc with 3Ј c-myc and His 6 tags in the pcDNA3.1/myc-His B vector (Invitrogen) were prepared as described (24). Serine-to-alanine variants of human PhLP at positions 18,19,20,25, and 296 were constructed in the pcDNA3.1/myc-His B vector by employing a PCR-based strategy and utilizing unique endonuclease restriction sites near the substitution site as described (24). The constructs were then subcloned into the bacterial expression vector pET15b (Novagen) as described (26). The integrity of all constructs was confirmed by sequence analysis. The N-terminally hemagglutinin (HA)-tagged G␥ 2 , FLAG epitope-tagged G␤ 1 and untagged G␤ 1 cDNAs also in the pcDNA3.1 vector were obtained from the UMR cDNA Resource Center.
Protein Expression and Purification-Wild-type and CK2 phosphorylation site variants of human PhLP in the pET15b vector were transformed in Escherichia coli DE3 cells and were purified using nondenaturing Ni 2ϩ affinity chromatography as described previously (11). The purified proteins were concentrated and exchanged into 20 mM HEPES, pH 7.2, 150 mM NaCl by ultrafiltration and were stored in 50% glycerol at Ϫ20°C. Purified G␤ 1 ␥ 1 and retinal rod outer segment membranes were prepared also as described previously (11). Protein concentrations were determined using Coomassie Plus protein assay reagent (Pierce), and the purity was determined to be ϳ95% by SDS-PAGE.
Assay of PhLP Binding to CCT-The binding of CK2 phosphorylated or unphosphorylated PhLP and its CK2 phosphorylation site variants to CCT was measured by adding 5 nM purified PhLP to 5% rabbit reticulocyte lysate in phosphate-buffered saline containing 1.2% IGEPAL CA-630 (Sigma) and 0.6 mM phenylmethylsulfonyl fluoride in a total volume of 300 l. Binding was allowed to proceed for 30 min at 23°C after which the PhLP was immunoprecipitated by incubating for 30 min at 4°C with 2.1 g of anti-c-myc antibody (clone 9E10, Biomol) followed by the addition of 30 l of a 50% slurry of Protein A/G Plus-agarose beads (Santa Cruz Biotechnology) and another 30-min incubation at 4°C with constant mixing. The beads were washed with the phosphate-buffered saline/IGEPAL buffer, and proteins were solubilized with 20 l of 2ϫ SDS-PAGE sample buffer. 15 l of each sample was resolved on a 10% SDS-PAGE gel and was immunoblotted with a 1:1000 dilution of rabbit polyclonal anti-CCT⑀ antiserum (18) followed by a 1:2000 dilution of a goat anti-rabbit horseradish peroxidase-conjugated secondary antibody (Calbiochem). Immunoblots were developed with the ECL Plus chemiluminescence reagent (GE Healthcare) and visualized with a Storm 860 PhosphorImager (Amersham Biosciences). The band intensities were quantified using Image-QuaNT software (Amersham Biosciences).
Assay of PhLP Inhibition of G t Binding to Rhodopsin-Lightinduced binding of 0.2 M 125 I-labeled G t ␣ and 0.2 M G t ␤␥ to membranes containing 1 M rhodopsin Ϯ 2 M CK2 phosphorylated or unphosphorylated PhLP was measured as described previously (11).
Mass Spectrometric Analyses-Tryptic peptides of CK2-phosphorylated PhLP were generated and analyzed as described previously (26). Briefly, molecular ions in the effluent from a C18 capillary chromatography that corresponded to the predicted masses of phosphopeptides from PhLP were fragmented by collision-induced dissociation in a Q-ToF mass spectrometer (LC/MS/ MS). Fragmentation spectra were obtained using either automated or manual parent ion selection. Data were analyzed using BioAnalyst software (Applied Biosystems, Framingham, MA).
Electrophoretic Mobility Determinations-CHO cells were plated in 6-well plates so that they were 70 -80% confluent the next day. The cells were then transfected with 1 g of either wild-type PhLP-myc or one of the CK2 phosphorylation site variants using Lipofectamine Plus reagent according to the manufacturer's protocol. The cells were harvested 48 h later in 200 l of immunoprecipitation buffer (24), and the PhLP-myc was immunoprecipitated from the lysate with 3 g of anti-cmyc antibody and 30 l of Protein A/G beads as described previously (17,24). The final precipitate was solubilized in 40 l of 2ϫ SDS-PAGE sample buffer, and 10 l of each sample was resolved on 10% SDS-PAGE gels. The gels were immunoblotted with a 1:1000 dilution of the anti-c-myc antibody and developed as described above.
G␤␥ Expression Measurements-HEK-293 cells were plated in 6-well plates so that they would be 70 -80% confluent the following day. They were then co-transfected with 1 g of each of the PhLP-myc, HA-G␥ 2 , and G␤ 1 cDNAs using Lipofectamine Plus reagent. In all experiments involving multiple transfections, the total amount of cDNA was held constant by adding empty vector. After 48 h, the cells were washed and solubilized in 200 l of immunoprecipitation buffer. The G␤ 1 ␥ 2 complexes were immunoprecipitated from 100 l of the lysate with 1.5 g of anti-HA (clone 3F10, Roche Applied Science) antibody as described previously (17,24). The complexes were solubilized in 40 l of 2ϫ SDS sample buffer, and 10 l was resolved on 10% Tris-glycine SDS-PAGE gels for G␤ 1 , or 20 l was resolved on 16.5% Tris-Tricine SDS-PAGE gels for G␥ 2 . For G␤ detection, the gels were immunoblotted with a 1:2000 dilution of a G␤ 1 antibody as described (27). For HA-G␥ 2 , the membranes were blocked for 1 h in 2% bovine serum albumin in Tris-buffered saline, followed by incubation for 1 h in a 1:500 dilution of anti-HA antibody in the blocking buffer and then 1 h in a 1:2000 dilution of goat anti-rat horseradish peroxidaseconjugated secondary antibody (Calbiochem). The immunoblots were developed as described above.
Radiolabel Pulse-Chase Assay-The pulse-chase assay was performed and quantified as described previously (24). A similar protocol was used to measure the rate of release of nascent G␤ from CCT. Six-well plates of HEK-293 cells were co-transfected with 1.0 g of FLAG G␤ 1 , HA-G␥ 2 or PhLP-myc variants as indicated. After a 10-min pulse, the radiolabel was chased for the times indicated and the cells were harvested in 220 l of immunoprecipitation buffer. The extract was divided into two 95-l samples, and 2.5 l of 1 g/l anti-CCT⑀ antibody (Serotec) was added to one sample and 3.0 l of 1 g/l anti-FLAG antibody was added to the other sample. The immunoprecipitation and analysis of the radiolabeled proteins co-immunoprecipitating with CCT were carried out as described (24). The G␤ 1 band was clearly separated from the other radiolabeled bands, facilitat-ing its quantification. The amount of G␤ 1 in the CCT immunoprecipitate was divided by that in the FLAG-G␤ 1 immunoprecipitate to determine the fraction of the total G␤ 1 bound to CCT. These values were expressed as a percentage of the 30-min time point to readily compare the rates of G␤ dissociation from CCT. The data were fit to a first order dissociation rate equation using the KaleidaGraph graphics software to determine the dissociation rate constant k. From the k values, the half-life was calculated by the equation, t1 ⁄ 2 ϭ ln 2/k.
Assay of G␤ Binding to CCT-For CCT binding experiments involving G␤ 1 overexpression, HEK-293 cells were plated in 6-well plates and transfected with 1.0 g of FLAG-G␤ 1 , HA-G␥ 2 , Pdc, or PhLP cDNAs as indicated in Fig. 5 (A and C). Alternatively, the transfections were performed with 0.5 g of FLAG-G␤ 1 , 1.0 g of PhLP variants, and 1.5 g of HA-G␥ as indicated in Fig. 6A. After 48 h, cells were lysed and extracts were immunoprecipitated with 2.5 g of anti-CCT⑀ antibody (Serotec), the immunoprecipitates were resolved on SDS-PAGE gels and immunoblotted for FLAG-G␤ 1 , PhLP-myc, Pdc-myc, or HA-G␥ 2 using the indicated antibodies as described above. G␤ 1 bands were quantified as described above, and intensities were calculated as a percentage of the control as indicated.
For binding experiments involving endogenous G␤, HEK-293 cells were grown in 100-mm dishes and transfected with 6.0 g of PhLP variant cDNAs as indicated in Fig. 5B. Cells were lysed in 1.2 ml of buffer, and 1 ml was immunoprecipitated with 10 g of anti-CCT⑀ antibody and 60 l of protein A/G beads. Endogenous G␤ 1 was detected with the anti-G␤ 1 antibody. For binding experiments involving endogenous PhLP, HEK-293 cells were grown in 6-well plates and transfected with 1.0 g of FLAG-G␤ 1 cDNA as indicated in Fig. 5C. Extracts from two wells were pooled, and 200 l was immunoprecipitated with 3 g of anti-CCT⑀ antibody. Other immunoprecipitation and immunoblotting procedures were as described above.

Effects of CK2 Phosphorylation of PhLP on CCT and G␤␥
Binding-To begin to assess the impact of CK2 phosphorylation on PhLP function, the effects of phosphorylation on the binding of PhLP to its two known binding partners, G␤␥ and CCT, were determined in vitro. Purified recombinant human PhLP was readily phosphorylated by CK2, resulting in a marked reduction in the mobility of the PhLP protein band in SDS-PAGE gels (Fig. 1A). The entire PhLP band was shifted, indicating that phosphorylation was 100% complete under the conditions used. The effects of CK2 phosphorylation on CCT binding FIGURE 1. Effects of CK2 phosphorylation on PhLP binding to CCT and G␤␥. A, the decrease in mobility of PhLP in SDS-PAGE upon CK2 phosphorylation is shown. PhLP was phosphorylated by CK2 in vitro (P-PhLP) and was analyzed on a 10% gel along with unphosphorylated PhLP. B, the effects of CK2 phosphorylation on the binding of PhLP to CCT and G␤␥ are shown. Binding was measured by immunoprecipitation of PhLP coupled with detection of the co-immunoprecipitating CCT⑀ or G␤ by immunoblotting. A representative immunoblot is shown. The graph gives the average intensity Ϯ S.E. of the CCT bands relative to the unphosphorylated sample from eight separate experiments. C, the effects of CK2 phosphorylation on the ability of PhLP to inhibit G␤ 1 ␥ 1 -assisted binding of 125 I-labeled G t ␣ to membranes containing light-activated rhodopsin were determined. The graph gives the average Ϯ S.E. from three separate experiments.

Mechanism of G Protein ␤␥ Dimer Assembly
was assessed by measuring the ability of PhLP to co-immunoprecipitate CCT from rabbit reticulocyte lysate (18). Human PhLP bound CCT with a high affinity, as evidenced by the fact that addition of only 5 nM PhLP was sufficient to co-immunoprecipitate readily detectible amounts of CCT from the reticulocyte lysate. Under these conditions, CK2 phosphorylation increased the co-immunoprecipitation of CCT by 7-fold ( Fig.  1B). In contrast, CK2 phosphorylation had no effect on the ability of PhLP to co-immunoprecipitate purified G␤ 1 ␥ 1 (Fig.  1B) or to inhibit association of G␤ 1 ␥ 1 with G t ␣ and light-activated rhodopsin (Fig. 1C), indicating that phosphorylation did not change the binding of PhLP to G␤ 1 ␥ 1 .
Mass Spectrometric Analysis of the CK2 Phosphorylation Sites of PhLP-The phosphorylation sites that could potentially be responsible for the increase in PhLP binding to CCT upon CK2 phosphorylation were identified by mass spectrometry. PhLP was phosphorylated by CK2 in vitro and digested with trypsin, and the resulting peptide fragments were analyzed by electrospray tandem mass spectrometry. Fig. 2A shows the collisioninduced dissociation (CID) spectrum of a doubly charged parent ion with an m/z ratio of 1198.5, corresponding to the mass of a tryptic peptide containing the C-terminal residues 287-301 of PhLP plus one phosphate. This CID spectrum showed robust peaks for both the b and y ions corresponding to the sequence of the 287-301 peptide. Of these, the y16 ion had an m/z equal to the loss of a phosphate, and the formation of a dehydroalanine at one of the two serines in this fragment, indicating that either Ser-293 or Ser-296 was phosphorylated in the parent ion. The b2, b4, b6, and b9 ions all had m/z ratios corresponding to the mass of their unphosphorylated fragments, suggesting that Ser-288 and Ser-293 were not phosphorylated. Therefore, the phosphate most likely resided on Ser-296. Accordingly, Ser-296 is within a strong consensus site for CK2 phosphorylation with negatively charged residues at the ϩ1 and ϩ3 positions (28).
One other tryptic fragment with m/z values corresponding to a phosphorylated species was detected and analyzed by CID. This peptide consisted of PhLP residues 13-32 plus one and two phosphates. The spectrum for the singly phosphorylated species yielded few b and y ions, none of which were phosphorylated. However, there were sufficient fragments to confirm the identity of the peptide (Fig. 2B). The same result was obtained with the doubly phosphorylated species. The CID spectrum confirmed the identity of the peptide but did not show any phosphorylated fragments (Fig. 2C). Hence, the four serines of this peptide, serines 18 -20 and 25, could all be considered as potential CK2 phosphorylation sites. Of the four, Ser-20 and Ser-25 are within CK2 consensus sites, and phosphorylation of Ser-20 would create a strong consensus site for CK2 phosphorylation of Ser-19 in a doubly phosphorylated species. Similarly, phosphorylation of Ser-19 would make Ser-18 a good CK2 site, although no triply phosphorylated species of the 18 -32 peptide were detected. Together, the mass spectrometric data suggest five potential CK2 phosphorylation sites on PhLP: Ser-18, Ser-19, Ser-20, Ser-25, and Ser-296.
Contribution of Specific CK2 Phosphorylation Sites to the PhLP-CCT Interaction-To identify which of these sites is responsible for the phosphorylation-dependent increase in PhLP binding to CCT, each of the five serines identified above was substituted with alanine in various combinations. The resulting PhLP variants were CK2-phosphorylated, and their binding to CCT was determined as described in Fig. 1. Substitution of one residue within the serine 18 -20 sequence caused only minor reductions in the phosphorylation-induced increase in binding, whereas substitution of two residues within this sequence resulted in reductions in the phosphorylationinduced binding from 7-fold to ϳ4-fold (Fig. 3). Replacement of all three serines within this phosphorylation site caused a further reduction in the phosphorylation-induced increase to 3-fold, indicating that multiple phosphorylation events within the serine 18 -20 site were responsible for much of the observed increase in PhLP binding to CCT upon CK2 phosphorylation.
With regard to the Ser-25 and Ser-296 sites, replacement of both residues caused a similar modest decrease in binding as was seen with dual substitution within the serine 18 -20 site, while replacement of either Ser-25 or Ser-296 along with all three of the Ser-18, Ser-19, and Ser-20 residues was required to completely block the phosphorylation-induced increase in binding. These results show that each of the five serines identified by mass spectrometry can contribute to the phosphorylation-induced increase in PhLP binding to CCT and that no other CK2 phosphorylation sites are involved in this process, suggesting that all the major CK2 phosphorylation sites were identified in the mass spectrometric analysis.
It is important to note that the serine to alanine replacements did not change the binding of unphosphorylated PhLP to CCT significantly (Fig. 3A), indicating that the alanine substitutions did not affect the folding of the PhLP variants nor did they modify the PhLP-CCT interaction significantly. Thus, the loss of the phosphorylation-induced increase in binding with the multiple alanine substitutions could be attributed to an inability of the variants to be phosphorylated by CK2.
Identification of Specific CK2 Phosphorylation Sites in Cells-To assess whether the CK2 phosphorylation sites of PhLP identified in vitro were also phosphorylated in vivo, the decrease in electrophoretic mobility upon CK2 phosphorylation was exploited to detect PhLP phosphorylation events in living cells. In this experiment, PhLP serine to alanine substitution variants with a C-terminal myc tag were transfected into CHO cells. Cells were extracted, and the PhLP was immunoprecipitated and immunoblotted with an antibody to the tag to distinguish the variants from the endogenous PhLP. Substitution of a serine that normally would be phosphorylated by CK2 in wild-type PhLP would be expected to result in an increase in mobility of that PhLP variant. Wild-type PhLP showed decreased mobility of the entire band when compared with an unphosphorylated PhLP standard, indicating that all of the transfected PhLP was phosphorylated in the CHO cells (Fig. 3C,  upper panel). The PhLP S20A variant showed two bands, a higher band with the same mobility as wild-type PhLP and a lower band Representative immunoblots for the phosphorylated (P) and unphosphorylated (NP) variants are shown. B, the -fold increase in CCT bound upon CK2 phosphorylation of PhLP was calculated by dividing the CCT⑀ band intensity of the phosphorylated sample by that of the unphosphorylated sample. The graph gives the average increase Ϯ S.E. from three to five separate experiments. C, the shift in electrophoretic mobility of PhLP upon CK2 phosphorylation was used to determine which of the putative CK2 sites were phosphorylated in cells. CHO cells were transfected with the indicated PhLP variants with C-terminal myc epitope tags. After 48 h, the cells were harvested and extracts were immunoprecipitated and immunoblotted with an antibody to the myc tag. Phosphorylation of the variants was determined by the shift in mobility of the PhLP band compared with wild-type PhLP-myc or purified, unphosphorylated PhLP-myc. D, the electrophoretic mobility of wild-type PhLP and the indicated PhLP variants after CK2 phosphorylation in vitro was also determined to compare the mobility shifts in vitro with those observed in cells. PhLP variants were analyzed by SDS-PAGE gels as in Fig. 1. with increased mobility. The ratio of the intensities of the two bands was ϳ2 to 1, with the higher band having the greater intensity. The PhLP S18A and S19A showed a very small amount of the lower band, whereas S25A and S296A showed no lower bands. The presence of both bands in the S20A variant suggests that phosphorylation of Ser-18 or Ser-19 may be partially impaired when position 20 cannot be phosphorylated. These results indicate that Ser-20 is phosphorylated in cells and that other phosphorylation events might also occur within the serine 18 -20 sequence.
A similar analysis was done with double and triple serine to alanine substitutions (Fig. 3C, middle panel). The S18A/S19A/ S20A variant showed a single lower band compared with wildtype PhLP but that was still higher than the unphosphorylated control. The S18A/S20A and S19A/S20A variants also showed a major lower band, with almost no higher band corresponding to wild-type PhLP. The S18A/S19A variant showed both the higher and lower bands, confirming phosphorylation at Ser-20 and indicating that it is sufficient for the mobility shift. The two bands also suggest that Ser-20 phosphorylation may be impaired in the absence of serine or phospho-serine at position 18 or 19. In the case of the S25A/S296A variant, there was a single higher band with the same mobility as wild-type PhLP, demonstrating that the large decrease in mobility is not a result of Ser-25 or Ser-296 phosphorylation, but rather it stems from at least one phosphorylation event in the Ser-18, Ser-19, and Ser-20 sequence.
The mobility of PhLP variants substituted at four and all five sites was also determined. The S18A/S19A/S20A/S25A variant causes an additional increase in the mobility of the PhLP band to the same mobility as unphosphorylated PhLP, whereas the S18A/S19A/S20A/S296A variant did not increase the mobility beyond that of serines 18 -20. The variant in which all five sites were substituted has the same mobility as the S18A/S19A/ S20A/S25A variant and unphosphorylated PhLP. These data show that Ser-25 is also phosphorylated in cells, at least in the absence of phosphorylation at serines 18 -20, and that Ser-25 phosphorylation causes a small decrease in PhLP mobility. The lack of change in mobility with substitution of Ser-296 did not permit a conclusion to be made about the phosphorylation of this site in cells. Either phosphorylation at Ser-296 did not occur or it did not change the mobility of PhLP in SDS gels.
A very similar pattern of electrophoretic mobility shifts was observed with in vitro CK2 phosphorylation of the PhLP S/A variants (Fig. 3D). S18A/S19A/S20A showed increased mobility compared with wild-type PhLP, and the S18A/S19A/S20A/ S25A/S296A variant had the same mobility as unphosphorylated PhLP. The similarities in mobility of the PhLP variants between the in vitro phosphorylation and that found in cells argue that CK2 is responsible for PhLP phosphorylation in vivo, in agreement with previous data indicating that CK2 was the physiologically relevant kinase (14). Importantly, in the absence of CK2 phosphorylation, the S/A variants all had the same mobility as unphosphorylated wild-type PhLP (data not shown), indicating that the differences in mobility were not caused by the alanine substitutions. Together, these data make a strong case for CK2 phosphorylation events within the serines 18 -20 and 25 sites in vivo.

Effects of Specific CK2 Phosphorylation Sites on G␤␥ Expression and Dimer
Assembly-It has recently been reported that substitution of all three serine residues in the serine 18 -20 sequence blocked the ability of PhLP to enhance the cellular expression of G␤␥ (24,25). To further investigate this phenomenon, the CK2 phosphorylation site variants of PhLP were coexpressed with G␤ 1 and G␥ 2 in HEK-293 cells, and the effects on G␤␥ expression were measured by immunoprecipitating the G␥ 2 from cell extracts and immunoblotting for both G␤ 1 and G␥ 2 . Co-expression of the single phosphorylation site variants did not change G␤␥ expression significantly compared with wild type, nor did co-expression of the S18A/S19A or the S25A/ S296A double variants (Fig. 4, A and B). However, co-expression of the S18A/S20A or the S19A/S20A double variants inhibited G␤ expression by 60 -70% and G␥ expression by ϳ50% compared with wild type (Fig. 4A). Likewise, the S18A/ S19A/S20A triple variant inhibited G␤ and G␥ expression by 70 -80% and 60 -70%, respectively (Fig. 4, A and B). Further substitution of Ser-25 and Ser-296 caused no further decreases in G␤␥ expression (Fig. 4B). These data clearly show that phosphorylation of at least one serine within the serine 18 -20 sequence is important for PhLP to assist in the expression of G␤␥, with Ser-20 phosphorylation contributing the most to this process. They also show that phosphorylation of Ser-25 and Ser-296 plays no additional role in G␤␥ expression. Moreover, the significant reduction in G␤␥ expression by several of the PhLP serine 18 -20 variants to levels below those observed with the empty vector indicate that these variants block the ability of endogenous PhLP to support G␤␥ expression and are thus acting as dominant negative inhibitors of G␤␥.
The reason for enhanced G␤␥ expression in the presence of CK2-phosphorylated PhLP is that phosphorylated PhLP increases the rate of G␤␥ dimer assembly (24). To determine which phosphorylation sites are critical for PhLP-mediated G␤␥ assembly, the ability of the PhLP CK2 phosphorylation site variants to catalyze G␤␥ dimer assembly was determined. All of the double and triple variants of the serine 18 -20 sequence were compromised in their ability to assist in G␤␥ dimer formation compared with wild-type PhLP (Fig. 4C). The S18A/ S19A variant was the least compromised, as reflected by an assembly half-life of 99 min compared with 42 min for wildtype PhLP, whereas the S18A/S19A/S20A variant was the most compromised, with a half-life of 284 min (Fig. 4C). The S19A/ S20A and S18A/S20A variants showed intermediate half-lives of 153 and 128 min, respectively. In contrast, the S25A/S296A variant was as effective as wild-type PhLP in promoting G␤␥ assembly with a half-life of 46 min. These G␤␥ assembly results are qualitatively similar to the G␤␥ expression data. However, there is one significant quantitative difference in the S18A/ S19A variant between the G␤␥ expression and the G␤␥ assembly data. G␤␥ expression was only slightly reduced by the S18A/ S19A variant, whereas the rate of G␤␥ assembly was reduced by Ͼ2-fold. This difference can be explained by the 48-h time period over which G␤␥ expression was measured. It appears that the 2-fold reduction in the rate of G␤␥ assembly is sufficient to maintain the steady-state G␤␥ levels achieved in the 48-h expression period to near those found in the presence of wild-type PhLP, whereas the larger reduction in assembly FIGURE 4. Effects of PhLP phosphorylation on G␤␥ expression and assembly. A, cellular expression of G␤␥ dimers was determined in the presence of PhLP S18A/S19A/S20A variants. HEK-293 cells were transfected with G␤ 1 , HA-tagged G␥ 2 , and the indicated variants. The cells were harvested, and extracts were immunoprecipitated with an antibody to the HA tag. The amount of HA-G␥ 2 and co-immunoprecipitated G␤ 1 was determined by immunoblotting with anti-HA and anti-G␤ 1 antibodies. A representative blot is shown. The graph gives the average G␤ 1 and HA-G␥ 2 amounts Ϯ S.E. relative to wild-type PhLP from three separate experiments. Cells in the empty sample were transfected with pcDNA3.1 vector with no PhLP cDNA. B, similar experiments were performed with PhLP variants S25A and S296A separately and in combination with S18A/S19A/S20A. The data are also combined from three separate experiments. C, the rate of nascent G␤ 1 ␥ 2 dimer formation in the presence of CK2 phosphorylation site variants of PhLP was determined using a radiolabel pulse-chase assay. Time measurements indicate the sum of the 10-min pulse and the variable chase periods. A representative gel is shown. Band intensities were quantified, and molar ratios of G␤ 1 to HA-G␥ 2 were calculated and plotted. Lines represent a fit of the data from three separate experiments to a first-order rate equation. Values for t1 ⁄2 are shown next to the graph. observed with the other Ser-18, Ser-19, and Ser-20 variants is not. Together, the G␤␥ assembly and expression data indicate that two phosphorylation events in the serine 18 -20 sequence are required for PhLP to be fully active in catalyzing G␤␥ assembly. Phosphorylation at one of the three sites results in partial activity, with Ser-20 phosphorylation conferring the most activity. The results also show that phosphorylation of Ser-25 or Ser-296 has no bearing on G␤␥ assembly.
G␤ Binds CCT in a Ternary Complex with PhLP-The correlation between the increase in binding of PhLP to CCT upon phosphorylation of serines 18 -20 (Fig. 3B) and the necessity of phosphorylation of serines 18 -20 for full activity in G␤␥ assembly (Fig. 4C) suggests that the effects of PhLP phosphorylation on assembly may occur through CCT. However, a role for CCT in G␤␥ assembly has not been established (24). If CCT does participate in the assembly process, then it must interact with G␤, G␥, or both. An interaction between G␤ and CCT has been observed in yeast protein interaction screens, but no such interaction has been reported in mammalian cells. Therefore, the binding of G␤ and G␥ to CCT was assessed by co-immunoprecipitation of overexpressed G␤ or G␥ in HEK-293 cells. G␤ co-immunoprecipitated with CCT robustly, to a similar extent as overexpressed PhLP, whereas overexpressed Pdc, which does not bind CCT, was not found in the CCT immunoprecipitate (Fig. 5A). Thus, G␤ appears to be specifically interacting with CCT under overexpression conditions. In contrast, overexpressed G␥ did not co-immunoprecipitate with CCT (Fig.   5A). To determine whether the interaction also occurred with endogenous amounts of G␤, the experiment was also done without overexpressing G␤. Co-immunoprecipitation of G␤ with CCT was also observed with endogenous G␤, confirming the results of the overexpression experiments (Fig. 5B).
The manner in which PhLP binds CCT at the top of the apical domains without entering the folding cavity (18) suggests that PhLP, G␤, and CCT might form a ternary complex in the process of G␤␥ folding. If such a ternary complex does exist, then PhLP would be predicted to increase the binding of G␤ to CCT and vice versa. To test this possibility, the effects of PhLP or G␤ overexpression on the binding of the other to CCT was measured. As predicted, G␤ overexpression increased the binding of endogenous PhLP to CCT (Fig. 5C). However, PhLP overexpression unexpectedly caused a small but reproducible decrease in G␤ binding to CCT (Fig. 5B). It is possible that this decrease in G␤ binding to CCT might be caused by PhLP-catalyzed G␤␥ assembly and release of the G␤␥ dimer from CCT. To test this possibility, the effects of two PhLP variants that do not support G␤␥ assembly on G␤ binding to CCT were also tested. One variant was PhLP S18A/S19A/S20A, and the other was a truncation variant in which residues 1-75 had been removed (PhLP ⌬1-75) (24). Both of these variants bind CCT, but they block G␤␥ assembly in a dominant negative manner (24). Overexpression of either of these variants increased endogenous G␤ binding to CCT dramatically (Fig. 5B). Thus, it appears that in the absence of serine 18 -20 phosphorylation, PhLP forms a ternary complex with G␤ and CCT that cannot progress in the assembly process. It is interesting to note that the PhLP⌬1-75 variant binds G␤␥ very poorly (24), yet it is still able to stabilize the complex between G␤ and CCT. This observation indicates that PhLP ⌬1-75 may do so, more through interactions with CCT than through interactions with G␤.
PhLP Phosphorylation Is Required for the Release of G␤ from CCT and Interaction with G␥-To further investigate the apparent correlation between the destabilization of the PhLP⅐G␤⅐CCT ternary complex by PhLP phosphorylation and the requirement for PhLP phosphorylation in G␤␥ assembly, the effects of G␥ on ternary complex formation with several PhLP variants was measured. G␤ was overexpressed in HEK-293 cells with G␥ and PhLP variants as indicated, and the amount of G␤ co-immunoprecipitating with CCT was measured (Fig. 6A). Co-expression of G␥ caused a decrease in G␤ binding to CCT that was intensified by the co-expression of FIGURE 5. G␤ binds CCT in a ternary complex with PhLP. A, binding of G␤ to CCT was detected by coimmunoprecipitation. HEK-293 cells were transfected with FLAG-G␤ 1 , PhLP, Pdc, or HA-G␥ 2 , and cell extracts were immunoprecipitated with an antibody to CCT⑀ to bring down CCT complexes. The immunoprecipitates were immunoblotted for G␤ 1 , PhLP, Pdc, or G␥ 2 . B, the effects of PhLP on the binding of endogenously expressed G␤ to CCT were measured by co-immunoprecipitation. HEK-293 cells were transfected with wildtype PhLP, the PhLP S18A/S19A/S20A or ⌬1-75 variants, or empty vector. Cell extracts were immunoprecipitated with the anti-CCT⑀ antibody and immunoblotted for endogenous G␤ 1 . A representative immunoblot is shown. Bars in the graph represent the average Ϯ S.E. of the G␤ band intensity relative to the empty vector control from four separate experiments. C, the effects of G␤ on the binding of endogenously expressed PhLP to CCT were also measured by co-immunoprecipitation. HEK-293 cells were transfected with G␤ 1 , CCT was immunoprecipitated as in panel B, and samples were immunoblotted for endogenous PhLP. A representative immunoblot is shown. Bars in the graph represent the average Ϯ S.E. of the PhLP band intensity relative to the empty vector control from three separate experiments.
wild-type PhLP. In striking contrast, G␤ binding to CCT was greatly enhanced by co-expression of PhLP ⌬1-75 and was completely insensitive to co-expression of G␥. Co-expression of PhLP S18A/S19A/S20A also enhanced G␤ binding to CCT significantly, and G␥ had much less of an effect on binding than with wild-type PhLP. Interestingly, the effects of PhLP ⌬1-75 and S18A/S19A/S20A on G␤ binding to CCT in the presence of G␥ were quantitatively very similar to their effects on G␤␥ assembly. PhLP ⌬1-75 completely blocked G␤␥ assembly (24) and G␥-mediated dissociation of G␤ from CCT, whereas PhLP S18A/S19A/S20A decreased the rate of G␤␥ assembly by 15-fold (24) and G␥-induced dissociation of G␤ from CCT by 9-fold (compare the G␤␥ PhLP-WT sample to the G␤␥ PhLP S18A/S19A/S20A sample in Fig. 6A). From these data, it appears that PhLP phosphorylation contributes to G␤␥ assembly by enhancing the ability of G␥ to release G␤ from the ternary complex.
There are two possible mechanisms by which phosphorylated PhLP could contribute to G␥-mediated release of G␤ from CCT. Both involve a conformational change in the ternary complex upon PhLP phosphorylation. First, PhLP phosphorylation could induce a conformation that allows G␥ to access G␤ in the ternary complex and form the G␤␥ dimer. The G␤␥ would then be released from CCT. Second, phosphorylation could induce a conformation that releases PhLP⅐G␤ from CCT, thereby freeing the G␥ binding site on G␤ for G␤␥ association to occur. To distinguish between these two mechanisms, the effects of G␥ and PhLP overexpression on the rate of dissociation of G␤ from CCT were measured. In this experiment, cells co-expressing G␤ with G␥, PhLP, or PhLP S18A/S19A/S20A were pulsed with [ 35 S]methionine for 10 min to label the nascent polypeptides and then were chased with unlabeled methionine. At the times indicated, the cells were lysed and CCT was immunoprecipitated. The co-immunoprecipitating proteins were separated by SDS-PAGE, and the amount of 35 S in the G␤ band was quantified. In the absence of PhLP or G␥ co-expression, the dissociation rate of nascent G␤ from CCT was very slow, with a t1 ⁄ 2 of ϳ8 h. PhLP co-expression increased the rate by 4-fold to a t1 ⁄ 2 of ϳ2 h. In contrast, PhLP S18A/S19A/S20A co-expression did not increase the dissociation rate (Fig. 6B). When G␥ was co-expressed with G␤, the dissociation rate increased by Ͼ2-fold to a t1 ⁄ 2 of ϳ3 h, whereas, when both G␥ and PhLP were co-expressed, the t1 ⁄ 2 increased even further to ϳ2 h, the same value observed in the absence of G␥ overexpression (Fig. 6C). When PhLP S18A/S19A/S20A was co-expressed with G␥, there was essentially no G␤ dissociation, similar to what was seen in the absence of G␥ overexpression (Fig. 6C).
These effects of G␥, PhLP, and PhLP S18A/S19A/S20A on the dissociation rates are consistent with their effect on the steadystate binding of G␤ to CCT (Fig. 6A) and further demonstrate that PhLP phosphorylation is required for the release of G␤ from the ternary complex. These findings are able to distinguish between the two potential mechanisms mentioned above. For example, the enhanced rate of dissociation of G␤ from CCT upon PhLP overexpression in the absence of G␥ overexpression (Fig. 6B) is consistent with the second mechanism in which a phosphorylated PhLP⅐G␤ complex would be released prior to G␥ binding to G␤. This result would not be expected in the first mechanism in which G␥ binding would be required for release of G␤ from CCT. Similarly, the observed lack of increase in the G␤ dissociation rate upon co-expression of G␥ with PhLP would be predicted by the second mechanism but not by the first. On the other hand, the increased release of G␤ from CCT upon G␥ overexpression in the absence of PhLP overexpression is consistent with the first mechanism, but this result could also be explained by the second mechanism if the endogenous PhLP were acting catalytically to release G␤ from CCT for association with G␥. In this case, the dissociation process would be drawn forward by the formation of the G␤␥ dimer and its association with G␣ and cell membranes (Fig. 7).
To further assess the role of G␥ in the release of G␤ from CCT, the possible association of G␥ with G␤ and PhLP in CCT complexes was determined. G␥ was co-expressed with the indicated combinations of G␤ and the PhLP variants, the CCT complexes were immunoprecipitated, and the samples were immunoblotted for G␥. G␥ was not found in any of the CCT immunoprecipitates (Fig. 6D), despite the fact that G␤ and PhLP could be readily found under these conditions (see Fig. 5).
Thus, it appears that G␥ does not interact with CCT in any of its complexes with G␤ and PhLP.
Together, the data in Fig. 6 indicate that PhLP phosphorylation results in the release of a PhLP⅐G␤ complex from CCT that can then associate with G␥ to form the G␤␥ dimer. This conclusion is also supported by the previously reported observation that PhLP forms a stable complex with G␤ that does not include G␥ (24).

DISCUSSION
A model for G␤␥ Assembly-Recent studies have shown that PhLP acts as an essential chaperone in the assembly of G␤␥ dimers by binding the G␤ subunit and thereby allowing G␥ to associate with G␤ (24,25). Phosphorylation of PhLP at serines 18 -20 by CK2 was required for G␤␥ assembly to FIGURE 6. PhLP phosphorylation is required for the release of G␤ from CCT and interaction with G␥. A, the effects of PhLP phosphorylation and G␥ co-expression on G␤ binding to CCT were measured by co-immunoprecipitation. HEK-293 cells were transfected with FLAG-G␤ 1 , HA-G␥ 2 , and the PhLP variants as indicated. Cell extracts were immunoprecipitated with an antibody to CCT⑀ and then immunoblotted for FLAG-G␤. A representative immunoblot is shown. Bars in the graph represent the average Ϯ S.E. of the G␤ band intensity relative to the G␤/PhLP-WT sample from three separate experiments. B, the effects of PhLP phosphorylation on the rate of G␤ release from CCT were measured using a radiolabel pulse-chase assay. HEK-293 cells were transfected with FLAG-G␤ 1 and the indicated PhLP variants. The pulse-chase assay was performed as in Fig. 4C. After the chase times indicated, cell extracts were immunoprecipitated with antibodies to CCT⑀ or G␤ 1 . Proteins were resolved by SDS-PAGE, and radiolabeled bands were detected using a PhosphorImager. The G␤ band intensities were quantified, and ratios of nascent G␤ 1 in the CCT immunoprecipitate versus the total nascent G␤ in the G␤ immunoprecipitate were calculated and plotted as a percentage of the ratio at the first time point. Lines represent a fit of the data from 3 separate experiments to a first-order rate equation. Values for t1 ⁄2 are shown below the graph. C, the effects of G␥ on PhLP-mediated release of nascent G␤ 1 from CCT were measured as in panel B in HEK-293 cells co-expressing HA-G␥ 2 in addition to FLAG-G␤ 1 and PhLP. D, the ability of G␥ to bind CCT was assessed by co-immunoprecipitation. HEK-293 cells were transfected with FLAG-G␤ 1 , HA-G␥ 2 , or PhLP variants as indicated. Cells extracts were immunoprecipitated with an antibody to CCT⑀ and then immunoblotted for G␥. A representative blot is shown. The Std lane in the CCT IP panel was the lysate from the G␥-transfected cells.
occur, yet the means by which phosphorylation of serines 18 -20 contributes to assembly was unknown. Moreover, CCT had been implicated in the assembly process, but the results were conflicting (18,24,25). The current study provides evidence for a molecular mechanism describing both the role of CCT and PhLP phosphorylation in G␤␥ assembly (Fig. 7). There are five important steps in this mechanism: 1) the nascent G␤ polypeptide binds CCT. This is a stable complex that releases G␤ very slowly in the absence of PhLP. 2) PhLP binds forming a ternary complex. If PhLP is not phosphorylated, then the ternary complex forms in a stable conformation that does not release PhLP⅐G␤, and the G␤␥ assembly process is blocked. However, if PhLP is dually phosphorylated within the serine 18 -20 sequence, then the ternary complex assembles in a conformation that readily releases the PhLP⅐G␤ dimer. 3) PhLP⅐G␤ dissociates from CCT. The structure of the Pdc⅐G t ␤␥ complex shows that Pdc binds G␤ on the opposite face as G␥ (29), predicting that the G␥ binding site on G␤ would be free in the PhLP⅐G␤ dimer. 4) G␥ binds G␤ form-ing a PhLP⅐G␤␥ complex. This complex is stable with a 100 nM binding affinity (11). However, both the G␣ binding site and the membrane association surface of G␤␥ overlap extensively with the PhLP binding site (11); therefore in the presence of G␣ and membranes, PhLP would be expected to be released from G␤␥. 5) G␤␥ associates with G␣ and/or the endoplasmic reticulum membrane and is transported to the plasma membrane (30). PhLP is then free to catalyze another round of G␤␥ assembly.
This model readily explains the dominant negative effect of the PhLP S18A/S19A/S20A and PhLP ⌬1-75 variants. These variants form PhLP⅐G␤⅐CCT ternary complexes that do not release PhLP⅐G␤ for G␥ binding. Such stable ternary complexes would also block the endogenous, phosphorylated PhLP from forming competent ternary complexes capable of releasing PhLP⅐G␤ for G␥ binding. Previous explanations of the dominant negative effect of PhLP S18A/S19A/S20A, which postulated that unphosphorylated PhLP would block G␤ and G␥ association with CCT (25) or that unphosphorylated PhLP FIGURE 7. CK2 phosphorylation-dependent release model of G␤␥ assembly. A model is proposed in which nascent G␤ forms a ternary complex with CCT and PhLP. If PhLP is not phosphorylated, the ternary complex is stable and PhLP⅐G␤ is not released from CCT. If PhLP is phosphorylated, the ternary complex is destabilized, possibly by electrostatic repulsion between the phosphates in the serine 18 -20 phosphorylation site and negatively charged residues on the CCT␣ or -⑀ apical domains. Once released, the PhLP⅐G␤ complex binds G␥, forming the G␤␥ dimer. The dimer then associates with G␣ and membranes in a manner yet to be defined. In the process, PhLP is released to catalyze another round of dimer formation. The approximate position of the serine 18 -20 phosphorylation site is depicted by a red oval marked 'P.' The relative amount of positive and negative charges on the CCT apical domains that contact the PhLP N-terminal domain is also indicated. See text for details.
would form a PhLP⅐G␤ complex that would not accept G␥ (24), are incomplete.
Phosphorylation-induced Conformational Changes-One apparent inconsistency in the data is that CK2 phosphorylation of PhLP increased its binding to CCT in the absence of G␤ (Fig. 1), yet PhLP phosphorylation was necessary for the release of PhLP⅐G␤ from CCT in the presence of G␤ (Figs. 4 -6). The difference between these observations may stem from differences in the structures of the PhLP⅐CCT and PhLP⅐G␤⅐CCT complexes. Clues regarding the nature of the phosphorylationdependent changes in these structures may be gleaned from the cryoelectron microscopic studies of the unphosphorylated PhLP⅐CCT complex (18). In this complex, PhLP was shown to interact in two distinct conformations at the top of the CCT toroid, contacting only the CCT apical domains (18). In one conformation, the N-terminal phosphorylation site of PhLP was in close proximity to the CCT␣ and -⑀ apical domains and in the other conformation the phosphorylation site was in close proximity to the CCT and -␤ apical domains. The binding surfaces of all eight apical domains are dominated by charged and polar residues (31) with the CCT␣ and -⑀ binding surfaces having a high distribution of negative charge, whereas the CCT binding surface exhibits an extensive positively charged patch. The serine 18 -20 phosphorylation site of PhLP is harbored within a sequence ( 18 SSSDEDESD 26 ) that is already very negatively charged. The addition of phosphates within this sequence would create an extremely high concentration of negative charge that would interact effectively with the positively charged patch of CCT. In the absence of G␤, phosphorylation could favor the conformation that brings the PhLP phosphorylation site in close proximity to the CCT apical domain, increasing the binding of PhLP to CCT. In the presence of G␤, it is possible that interactions with G␤ may limit the ability of PhLP to rotate on the top of the CCT toroid. Thus, the phosphorylation site may be fixed in close proximity to the CCT␣ and -⑀ apical domains, causing electrostatic repulsion between the negative charges on the CCT␣ and -⑀ binding surfaces and the PhLP phosphorylation site. This repulsion might destabilize the ternary complex and allow the release of the PhLP⅐G␤ complex. Further studies will be required to test the validity of this structural model.
Regulation of CK2 Phosphorylation of PhLP-Given the essential role of CK2 phosphorylation of PhLP in G␤␥ dimer formation, an important issue yet to be addressed is the regulation of this phosphorylation event. CK2 is a constitutively active kinase with many protein substrates (32). Determination of which substrates are phosphorylated and when phosphorylation occurs appears to be controlled by regulated expression and assembly of the CK2 ␣ 2 ␤ 2 tetramer and by the association of different CK2 binding partners (32). In the case of PhLP, CK2 phosphorylation occurs within the first 30 min of its synthesis (data not shown), and it remains completely phosphorylated under the cell culture conditions used here (Fig. 2). It is not clear from the current data whether phosphorylation occurs prior to or after association of PhLP with CCT (Fig. 7). In mouse tissues, PhLP was also completely phosphorylated in brain and heart but was mostly unphosphorylated in the adrenal gland (14). It is possible that G␤␥ assembly is a continuous process in some cell types, whereas in others assembly is highly regulated, only occurring under certain conditions that promote CK2 phosphorylation of PhLP.
These investigations into the mechanism of PhLP-mediated G␤␥ assembly and its regulation by CK2 phosphorylation suggest that PhLP and its interactions with G␤ and CCT could be targeted by therapeutics to control the levels of G␤␥ expression and thus the degree of G protein signaling within the cell, perhaps providing additional tools to treat the myriad of G proteinlinked diseases.