Phosducin-like Protein Regulates G-Protein βγ Folding by Interaction with Tailless Complex Polypeptide-1α

Phosducin-like protein (PhLP) exists in two splice variants PhLPLONG (PhLPL) and PhLPSHORT (PhLPS). Whereas PhLPL directly inhibits Gβγ-stimulated signaling, the G βγ-inhibitory mechanism of PhLPS is not understood. We report here that inhibition of Gβγ signaling in intact HEK cells by PhLPS was independent of direct Gβγ binding; however, PhLPS caused down-regulation of Gβ and Gγ proteins. The down-regulation was partially suppressed by lactacystine, indicating the involvement of proteasomal degradation. N-terminal fusion of Gβ or Gγ with a dye-labeling protein resulted in their stabilization against down-regulation by PhLPS but did not lead to a functional rescue. Moreover, in the presence of PhLPS, stabilized Gγ subunits did not coprecipitate with stabilized Gβ subunits, suggesting that PhLPS might interfere with Gβγ folding. PhLPS and several truncated mutants of PhLPS interacted with the subunit tailless complex polypeptide-1α (TCP-1α) of the CCT chaperonin complex, which is involved in protein folding. Knock-down of TCP-1α in HEK cells by small interfering RNA also led to down-regulation of Gβγ. We therefore conclude that the strong inhibitory action of PhLPS on Gβγ signaling is the result of a previously unrecognized mechanism of Gβγ-regulation, inhibition of Gβγ-folding by interference with TCP-1α.

The two splice variants of phosducin-like protein (PhLP) 1 differ in the length of their N terminus and their expression pattern. The long form (PhLP L ) is a ubiquitously expressed protein and binds G-protein ␤␥-subunits (G␤␥) and thereby inhibits G␤␥-mediated functions (1)(2)(3)(4)(5). The extended N terminus of PhLP L (83 amino acids) contains a highly conserved G␤␥-binding motif, which plays the crucial role in binding and regulating G␤␥-subunits (4, 6 -11). In contrast, the short splice variant PhLP S , which has a more restricted expression, lacks this motif and did not seem to exert a major G␤␥ inhibition, when tested with purified proteins. However, we recently demonstrated that PhLP S showed a more pronounced G␤␥ inhibition than PhLP L in transiently transfected HEK 293 cells (12). Although PhLP L is the more abundantly expressed splice variant in most tissues, it is expressed at high levels in some tissues (e.g. adrenal gland) and has been suggested to play a role in regulation of catecholamine release (12,13). This suggests that PhLP S has an effect on G␤␥ that is not mediated by direct binding to G␤␥. Therefore, we set out to investigate this (potentially indirect) mechanism of G␤␥ inhibition by PhLP S in intact cells. We report that transfection of PhLP S was associated with down-regulation of transfected and endogenously expressed G␤␥ and that this down-regulation involved a proteasome-dependent pathway. It was recently reported that PhLP might inhibit the function of the cytosolic chaperonin complex (chaperonin complex containing TCP subunits (CCT)), which is involved in the folding of several proteins (14). Such an interaction might result in misfolding of G-protein subunits and might, thereby, provide an indirect mechanism of G-protein inhibition. In fact, we observed that PhLP S interacted with a subunit of the CCT complex and thereby appears to prevent proper folding of G␤␥ complexes.
Construction of Expression Vectors-All of the cDNAs used in these studies were subcloned into pcDNA3 (Invitrogen). The cDNAs for PhLP L , PhLP S , and PhLP L A18 -20 have been described previously (2,12). The construction of deletion mutants and Trp-66 to Val mutants of PhLP L was done by PCR and confirmed by automated sequencing. In order to generate G␤ or G␥ subunits resistant to N-terminal destabilization, the dye-labeling protein O 6 -alkylguanine-DNAalkyltransferase (AGT) (15), was subcloned N-terminal of either the G␤ or G␥ subunit in pcDNA3 by PCR and also confirmed by automated sequencing.
For binding assays, HEK cell extract was prepared by disrupting cells of one 10-cm dish in 1 ml of buffer (150 mM NaCl, 20 mM imidazol, 10 mM 2-mercaptoethanol, 2 mM MgCl 2 , 1 mM phenylmethylsulfonyl fluoride, 0.1% Nonidet P-40, and Tris-Cl, pH 7.5), followed by centrifugation (20,000 ϫ g, 10 min). Protein content of the supernatant was determined by the Bradford method. Binding was then performed by adding this supernatant from G␤␥-transfected HEK cells (100 g of protein) to 400 nM His 6 -tagged PhLP L , PhLP S , or GST (phosphorylated or control-treated). For certain binding experiments, cell extract of PhLP isoform-transfected cells was combined with cell extract of G␤ 1 ⅐His 6 -G␥ 2 -expressing cells. After incubation (37°C for 30 min), reactions were diluted with 5 volumes of ice-cold binding buffer supplemented with 50 l of Ni 2ϩ -NTA-agarose (Qiagen) and rotated at 4°C for 10 min. The beads were then washed, eluted with SDS-Laemmli buffer, and analyzed by SDS-PAGE and Western blotting (anti-G␤ antibody or anti-PhLP-CT antibody as indicated).
Co-immunoprecipitations-For the co-immunoprecipitation of PhLP constructs and endogenous TCP-1␣, HEK 293 cells were transiently transfected with the indicated cDNA, and, 42 h later, cells were lysed in PBS, 0.5% Nonidet P-40, 1 mM phenylmethylsulfonyl fluoride (lysis buffer). Precipitation was performed with the PhLP-CT antibody precoupled to protein G-Sepharose (Amersham Biosciences). After washing four times in lysis buffer, samples were separated by SDS-PAGE and transferred to polyvinylidene difluoride membranes. Detection of bound TCP-1␣ was performed with anti-TCP-1␣ antibody (rat; Calbiochem).
For co-immunoprecipitation of AGT-G␤ 1 and AGT-G␥ 2 , cells were transfected with AGT-G␤ 1 or AGT-G␥ 2 and were lysed and treated as before. Then AGT-G␤ 1 was precipitated with the G␤ antibody precoupled to Affi-Gel-10 (Bio-Rad) according to the manufacturer's protocol. Elution from the resin was done with 100 mM sodium citrate, pH 3.0. After neutralization, samples were separated by SDS-PAGE and analyzed by Western blotting with the G␥ 2 antibody.
Cell Culture and Transient Transfections-Human embryonic kidney (HEK) 293 or HEK 293-TSA cells were grown in Dulbecco's modified Eagle's medium, 10% fetal calf serum and were kept in a 7% CO 2 humidified atmosphere. Usually, cells were transfected by using the CaPO 4 method (18) on 10-cm dishes at 70% confluence with a constant total amount of DNA. In assays with endogenous G␤␥, cells were transfected at 40% confluence, and measurements were performed 66 h after transfection. In the case of membrane current measurements in HEK 293 cells stably expressing GIRK1/4, transfections were performed with the Effectene transfection kit (Qiagen) according to the manufacturer's protocol using 0.2 g of CD8 cDNA as transfection reporter system (19), 0.2 g of ␣ 2A -adrenergic receptor cDNA, and 0.6 g of PhLP S cDNA or empty plasmid. Transfection efficiency was usually between 60 and 80% and was equal for the different PhLP isoforms as judged by transfection of different green fluorescent protein-tagged constructs. Control of protein expression by Western blotting was performed as described (12).
Design and Use of siRNA for TCP-1␣ Knock-down-Two siRNAs directed against human TCP-1␣ were designed with the help of the MWG design tool (available on the World Wide Web at www.mwgbiotech.com) and the RNA sequence according to the Refseq human data base at NCBI (accession number NM_030752). The targeted sequences were 5Ј-GAA GUU GGA GAU GGA ACU ATT (positions 272-292) and 5Ј-GAA GUG GUA CAG GAG AGA ATT (positions 1067-1087). Both sequences were run on a Blast search and found to interact only with human TCP-1␣. They were produced (MWG) as preannealed duplex RNAs (sense and antisense each ending with dTdT). For control experiments, siRNA directed against GFP was designed as described recently (5Ј-CGU AAA CGG CCA GUU CTT, positions 64 -84 (20)). Transfection of HEK 293-TSA cells with the siRNAs was performed with Lipofectamine 2000 (Invitrogen) according to the manufacturer's protocol and analyzed 42 h later. In pilot experiments, both specific anti-TCP-1␣ siRNAs (in concentrations between 40 and 80 pmol/transfection in a 12-well plate) were found to be effective in reducing the amount of TCP-1␣ by roughly 50 -70% as determined by Western blots (data not shown). In all further experiments, they were subsequently used in combination (at 40 pmol/well each; for simplification named siTCP-1␣). siRNA against GFG was used in parallel at a concentration of 80 pmol/well.

Measurement of GIRK Channel Currents and Inositol
Phosphate Generation-Membrane currents were recorded under voltage clamp conditions, using conventional whole cell patch clamp techniques exactly as described recently (21). I K was measured as an inward current using a holding potential of Ϫ90 mV, and voltage ramps (from Ϫ120 to ϩ60 mV in 500 ms, every 10 s) were used to determine current-voltage (I-V) relationships. All measurements were performed at room temperature, and summarized results are presented as mean Ϯ S.E. Student's t test was done to test for significance of differences. Inositol phosphate measurements were performed in triplicates and results were analyzed as means Ϯ S.E. of at least three independent experiments. Analysis of variance and post-test comparison (Bonferroni) were done as appropriate.
RNA Preparation, Reverse Transcription, and Quantitative PCR-For total RNA preparation, cells seeded on 6-well plates were lysed and treated according to the instructions of the RNeasy Mini Kit (Qiagen) and after DNase digestion. RNA preparations were routinely controlled on a denaturing formaldehyde gel. One g of total RNA of each sample was reverse-transcribed with the Superscript III Reverse Transcriptase-Kit (Invitrogen) using oligo(dT) to produce first strand cDNA. For PCR of the human G␤ 1 cDNA, the primers were chosen to distinguish the cDNA product (95 bp) from a potentially contaminating genomic DNA product (324 bp): (a) 5Ј-CTTGCTGGGTACGAC-GACTT-3Ј and (b) 5Ј-AGCTGACGCGGTTGTCAT-3Ј. For PCR of the rat PhLP cDNA the primers were (a) 5Ј-ATGGAGCGGCTGATCAAAAAG and (b) 5Ј-CAAATTCTTGTCCATCATACC-3Ј, resulting in a 144-bp product. Real time PCR (in triplicates) was performed on an ABI PRISM Sequence Detection System 7700 (Applied Biosystems) with Sybr Green (Cambrex) as fluorescent, 6-carboxy-X-rhodamine (Molecular Probes, Inc., Eugene, OR) as reference dye and Hot-Taq (Eppendorf) as polymerase. A standard curve was generated with several dilution steps of the cDNA from control cells, and relative amounts of cDNAs were then calculated assuming a PCR efficiency of 100%. As controls, real time PCR of human glyceraldehyde-3-phosphate dehydrogenase and ␤-actin were performed and were not significantly different between groups.

Effects of Phosphorylation of PhLP L by CK2 on G␤␥ Binding-
We have recently demonstrated that PhLP L is constitutively phosphorylated by CK2 and that this phosphorylation inhibited the effect of PhLP on G␤␥-mediated inositol phosphate generation in intact HEK 293 cells (12). The homologous protein phosducin is also phosphorylated at the N terminus by several different kinases, and these phosphorylations lead to a loss in affinity toward G␤␥ subunits (11,(22)(23)(24)(25)(26). We therefore tested whether CK2-mediated phosphorylation affected directly G␤␥-binding of PhLP L . Recombinant C-terminally His 6tagged PhLP L was purified from E. coli by Ni 2ϩ -NTA affinity purification and was phosphorylated with recombinant CK2. Completion of phosphorylation was monitored and controlled by a mobility shift of the Coomassie-stained PhLP L band in SDS-PAGE (Fig. 1A, upper panel). Equimolar concentrations (400 nM) of PhLP L , phosphorylated PhLP L , PhLP S -His 6 , or GST-His 6 were incubated with lysates from HEK 293 cells overexpressing G␤␥ subunits. The His 6 -tagged proteins were then precipitated with Ni 2ϩ -NTA-agarose, and the amounts of coprecipitated G␤␥ subunits were monitored by Western blotting with a G␤-specific antibody (Fig. 1A, lower panel). These experiments showed that G␤␥ subunits were coprecipitated in a complex with PhLP L as well as with phosphorylated PhLP L , but there was no change in the amount of coprecipitated G␤␥ after phosphorylation of PhLP L . In contrast, PhLP S bound to G␤␥ only weakly. These findings were confirmed in the effects of PhLP on pertussis toxin-catalyzed ADP-ribosylation of Gprotein ␣ o subunits (G␣ o ). This ADP-ribosylation is stimulated by the presence of G␤␥ subunits. As depicted in Fig. 1B, the G␤␥-stimulated ADP-ribosylation of G␣ o was inhibited by PhLP L and by phospho-PhLP L to a similar extent, whereas the inhibition by PhLP S was only partial. IC 50 values for PhLP L and PhLP S (18.7 Ϯ 4.5 and 160 Ϯ 44.6 nM, respectively) were within the previously published range (4) but were not different between PhLP L and phospho-PhLP L (16.5 Ϯ 5.2 nM). Therefore, a decrease in the G␤␥ binding affinity of CK2-phosphorylated wild-type PhLP L did not seem to account for its reduced capacity to inhibit G␤␥-stimulated inositol phosphate generation in intact cells.

G␤␥-binding by PhLP L Is Dependent on Trp-66 in Its N
Terminus-To further analyze the G␤␥ regulatory role of PhLP L in intact cells, we mapped the G␤␥-interacting sites. In the homologous retinal protein phosducin, the exchange of a tryptophan to a valine (Trp-29 to Val) within a conserved amino acid sequence of the N terminus leads to reduced G␤␥ binding (7). The resolution of the crystal structure of the phosducin⅐G␤␥ complex shows that this conserved region forms the first ␣-helical region of phosducin and PhLP L (8) (Fig. 1C). In order to investigate the functional role of this homologous region in PhLP L , we mutated Trp-66 in PhLP L (which is anal- , and glutathione S-transferase (GST; 9.59 g/ml) were incubated with lysate from HEK 293 cells transfected with G␤ 1 and G␥ 2 cDNA. After incubation for 30 min at 37°C, the recombinant proteins were precipitated with Ni 2ϩ -NTA-agarose, separated by SDS-PAGE, and transferred to polyvinylidene difluoride membranes. The phosphorylation state of PhLP L was controlled by the mobility shift of the Coomassie-stained band (12). Western blot detection was performed with a G␤-specific antibody (IB: G␤). Shown is a blot representative of three independent experiments and the corresponding Coomassie stain of the gel. B, inhibition of G␤␥-stimulated ADP-ribosylation of G␣ o by PhLP. After prephosphorylating PhLP L by CK2 as in A, the effect of increasing amounts of PhLP L , phospho-PhLP L , and PhLP S on the pertussis toxin-dependent (100 ng) and G␤␥-dependent (6 nM) [ 32 P]ADP-ribose transfer onto purified G␣ o (4 nM) was analyzed as described under "Experimental Procedures." As control, CK2 was added. Shown are the results of 3-7 independent experiments. Maximal inhibition Ϯ S.E. was as follows: for PhLP L , 90.2 Ϯ 1.8% (n ϭ 7); for phospho-PhLP L , 98.6 Ϯ 0.3% (n ϭ 6); for PhLP S , 22.6 Ϯ 5.2% (n ϭ 4). The IC 50 values for PhLP L and phospho-PhLP L were 18.7 Ϯ 4.5 and 16.5 Ϯ 5.2 nM, respectively, and 160 Ϯ 44.6 nM for PhLP S . C, alignment of the conserved G␤␥-binding motif from rat PhLP L (rat PhLP), rat phosducin (rat Phd), and PhLP from Drosophila melanogaster (dro PhLP). This region forms the ␣-helix 1 of the N terminus of these proteins (8). The essential tryptophan (W) marked in boldface type is at position 66 in rat PhLP L and at position 29 in rat phosducin (upper panel). In the lower panel, a representative Western blot of binding experiments is shown, where the different PhLP L mutants expressed in HEK 293 cells were captured with G␤ 1 ⅐His 6 -G␥ 2 , which was then precipitated with Ni 2ϩ -NTA-agarose (AP: His 6 ). The precipitate was analyzed in Western blots with the PhLP-CT antibody and in parallel with a G␥ 2 -antibody to confirm equal binding of the His 6 -tagged G␤␥ complex to the Ni 2ϩ -NTA-agarose. Five percent of each lysate (5% load) were analyzed as loading control with the PhLP-CT antibody and showed equal expression of the indicated proteins. D, HEK 293 cells were transiently transfected with cDNAs of PhLP L , PhLP L W66V, or control vector (8 g/10-cm cell culture dish) along with cDNAs of G␤ 1 , G␥ 2 , and PLC␤ 2 (3 g/10-cm dish). Inositol phosphate levels were determined 42 h later. Data represent mean Ϯ S.E. of five independent experiments (***, p Ͻ 0.001 versus control; छछ, p Ͻ 0.01 versus PhLP L W66V). ogous to Trp-29 in phosducin) to valine (PhLP L W66V). We further created a double mutant where, in addition to the Trp-66 to Val exchange, the CK2-dependent phosphorylation sites (Ser-18, Thr-19, and Ser-20) were replaced by alanines (PhLP L AV). To prove that the mutation of Trp-66 was sufficient to reduce binding of PhLP L to G␤␥ subunits, we used a His 6 -tagged-G␥ 2 cDNA (His 6 -G␥ 2 ) and performed a co-precipitation experiment with HEK 293 cell lysates. For this experiment, we expressed all PhLP isoforms transiently in HEK 293 cells (Fig. 1C) and separately expressed G␤ 1 plus His 6 -G␥ 2. We then lysed the cells with a detergent-containing buffer and incubated equal amounts of the G␤ 1 ⅐His 6 -G␥ 2 -containing lysate with cell lysate containing each of the PhLP isoforms. Subsequently, G␤ 1 ⅐His 6 -G␥ 2 complexes were precipitated with Ni 2ϩ -NTA-agarose. Then the amount of co-precipitated PhLP was analyzed by SDS-PAGE and Western blotting. As depicted in Fig. 1C, there was strong binding of cellular PhLP L to the G␤ 1 ⅐His 6 -G␥ 2 complex, whereas either mutation at the Trp-66 site (as in PhLP L W66V and PhLP L AV) or the loss of the N terminus (as in PhLP S ) led to a marked decrease in binding to the G␤ 1 ⅐His 6 -G␥ 2 complex. Thus, direct binding of PhLP L to G␤␥ subunits was clearly dependent on Trp-66. In addition, G␤␥-stimulated inositol phosphate generation was not inhibited when Trp-66 was mutated (Fig. 1D).
The PhLP L N Terminus and PhLP L Exhibit Similar G␤␥ Regulatory Function in Cells-Because of the marked effect of the N-terminal W66V mutation, we hypothesized that the N terminus of PhLP L alone should be able to inhibit inositol phosphate generation to the same extent as the wild-type PhLP L . We constructed several N-terminally deleted mutants of PhLP ( Fig. 2A) and tested their ability to inhibit G␤␥-mediated inositol phosphate generation (Fig. 2B). Expression of the diverse constructs was comparable (Fig. 2B, lower panel). We found that the N-terminal constructs (containing at least the first 83 amino acids of PhLP L ) inhibited G␤␥-stimulated signaling to a similar extent as did the wild-type full-length PhLP L in intact cells (by about 40 -50%).
PhLP S and Phosphorylation-deficient PhLP L Down-regulate G␤␥ Protein-Looking at the effect of PhLP S and of both phosphorylation-deficient mutants PhLP L A18 -20 and PhLP L AV (Fig. 3A), we found that not only PhLP S but also the two mutants were able to inhibit the inositol phosphate signal more effectively than wild-type PhLP L . This was seen although direct binding of G␤␥ to PhLP S and PhLP L AV (which bears the Trp-66 to Val

FIG. 4. Modification of the G␤ and G␥ N termini stabilizes their expression but does not rescue their function.
A, quantitative analysis of the effects of PhLP L (L) and PhLP S (S) on the protein content of G␤ 1 , AGT-G␤ 1 , G␥ 2 , and AGT-G␥ 2 . Four or five immunoblots (IB of indicated protein) were quantified after densitometric scanning and normalized to the density of the respective control condition (ctr). Whereas PhLP S significantly reduced the protein levels in wild-type G␤ 1 and G␥ 2 (***, p Ͻ 0.001 versus ctr), the modification of the N terminus with AGT completely prevented down-regulation of the modified G-protein subunit (but did not prevent down-regulation of the unmodified partner in the same cells). B, inositol phosphate assays performed with different combinations of wild-type and N-terminally modified G-protein subunits (as indicated). Compared with PLC␤ 2 transfection alone (basal), the co-transfection of either combination of G␤ and G␥ (and the indicated AGT fusion constructs) stimulated the formation of inositol phosphates (ctr). Identically as with wild-type G␤␥ (G␤ 1 G␥ 2 ), PhLP L (L) and PhLP S (S) exhibited the same pattern of inhibition for all G␤␥ combinations. Shown are results from representative experiments (of 3-5) performed in triplicate. C, effect of PhLP on association between AGT-G␤ 1 and AGT-G␥ 2 . AGT-G␤ 1 and AGT-G␥ 2 were cotransfected along with empty vector (control) or PhLP L or PhLP S cDNA as in B (right panel). Precipitation was performed using a G␤ antibody (IP: G␤), and the amount of co-precipitated AGT-G␥ 2 was determined by Western blotting with a G␥ 2 -specific antibody (IB: G␥ 2 ). In the presence of PhLP S , G␤ 1 and G␥ 2 did not seem to form a functional dimer. mutation in addition to the A18 -20 mutation) was less than that of wild-type PhLP L (Fig. 1C). We therefore concluded that a different type of regulation on G␤␥ subunits must exist in cells, which is distinct from direct G␤␥ binding. In line with this conclusion, the protein content of both G-protein subunits, G␤ and G␥, were dramatically decreased in HEK 293 cells co-transfected with PhLP L A18 -20, PhLP L AV, or PhLP S (Fig. 3B). As a control, the protein level of PLC␤ 2 was unchanged (Fig. 3B, lower panel). The decrease in the amount of G␥ 2 by PhLP S was partially restored by a 4-h incubation with the specific proteasome inhibitor lactacystine, suggesting the involvement of proteasomal degradation in the PhLP S -induced G␥ decrease (Fig. 3C).  145-301). C, co-immunoprecipitation of endogenous TCP-1␣ from HEK 293 cells with PhLP S and its constructs. HEK cells were transiently transfected with the indicated cDNAs. Forty-two h later, cells were lysed, and PhLP was precipitated with the PhLP-CT antibody (IP: PhLP-CT). Samples were separated by SDS-PAGE and immunoblotting against endogenous TCP-1␣ was performed (IB: TCP-1␣). One and two percent of the control lysate was used to compare TCP-1␣ expression (1 and 2% loading; IgG-HC denotes IgG heavy chain). Expression of the PhLP S constructs was controlled with the PhLP-CT antibody (IB: PhLP-CT). D, down-regulation of transfected G␥ 2 by RNA interference with endogenous TCP-1␣. HEK 293 cells in a 12-well plate (50% confluent) were simultaneously transfected with cDNAs for G␤ 1 , G␥ 2 , and PLC␤ 2 (0.3 g each) and either empty vector (control, 1 g), PhLP S -cDNA (1 g), or the siRNAs (1 g or a total of 80 pmol, respectively), as indicated. Forty-two h later, cells were lysed in Laemmli buffer (75 l), subjected to SDS-PAGE on Tris/Tricine gels (10%), and subsequently transferred to polyvinylidene difluoride membranes. Shown are immunoblots (IB) from one representative experiment (of five) demonstrating that siRNAs directed against TCP-1␣ (siTCP-1␣) led to a reduction of TCP-1␣ and G␥ 2 protein levels. The experiment was controlled by siRNA against GFP (siGFP). E, effect of siTCP-1␣ and control siRNA (siGFP) on the protein level of endogenous G␤ 1 . HEK 293 cells were transfected with siRNA (80 pmol/well of a 12-well plate) and 42 h later were subjected to SDS-PAGE and immunoblotting against the indicated proteins. Blots are from a representative experiment of six independent transfections and show that endogenous G␤ 1 is reduced upon inhibition of TCP-1␣.
The Fate of G␤␥ Subunits-To analyze whether the PhLP Smediated decrease in G␤␥ protein levels accounted for the apparent inhibition of G␤␥-mediated inositol phosphate signaling, we attempted to prevent G␤␥ degradation by stabilization of G␤ and/or G␥ subunits. Recently, it was demonstrated that the G␥ 2 subunit is a substrate for ubiquitinylation and degradation via the N-end rule pathway in bovine brain (27). That means that posttranslational modification of the N terminus leads to susceptibility to proteasome-dependent degradation. It was also demonstrated that mutation of the N terminus could protect the G␥ subunit from degradation. We therefore stabilized the G␤ and G␥ subunits by N-terminal fusion to the dye-labeling protein O 6 -alkylguanine-DNA-alkyltransferase, AGT (15). These modifications did not prevent functional interaction with PLC␤ 2 (see below). In line with previous obser-vations, this modification stabilized G␤ as well as G␥ subunits (27) and prevented G␤ and G␥ from disappearance in the presence of PhLP S (Fig. 4A). However, expression of the AGTstabilized N-terminal constructs of G␤ and G␥ either with the wild-type partner or in combination did not change the pattern of functional inhibition seen with PhLP S (Fig. 4B). These experiments also illustrate that the AGT-modified G␤ and G␥ subunits were functional, since their transfection increased the inositol phosphate generation, although less than the wild-type forms (compare control (ctr) to basal in each panel). Taken together, PhLP S led to down-regulation of G␤␥ (most probably by the proteasome pathway and via the N-end rule) in cells. But G␤␥ down-regulation did not seem to account for PhLP Smediated inhibition of G␤␥-stimulated signaling in cells. To further analyze the mechanism of G␤␥ dysfunction in the pres-FIG. 6. Effect of PhLP S on function and protein level of endogenous G␤␥ subunits. A, quantitative analysis of the human G␤ 1 mRNA level after transfection of empty vector (ctr) or PhLP S -cDNA in HEK 293-TSA cells. In parallel, inhibition of inositol phosphates by PhLP S was controlled as in B (not shown). Left, reverse transcription-PCR (RT-PCR) of human G␤ 1 (endogenous) and rat PhLP (transfected) was performed on RNA preparations as described under "Experimental Procedures" (Ϫ denotes (a) RNA preparation treated without reverse transcriptase and (b) PCR without template; ϩ denotes human G␤ 1 cDNA). Right panel, analysis by quantitative reverse transcription-PCR. Results were normalized according to a dilution standard. B, inhibition of inositol phosphate generation. HEK 293-TSA cells were transiently transfected (PhLP S cDNA or empty vector (ctr)) along with PLC␤ 2 to determine inositol phosphates. Data are mean Ϯ S.E. of eight independent experiments performed in triplicates (***, p Ͻ 0.001 versus control). C, down-regulation of endogenous G␤. In parallel to B, Western analysis of endogenous G␤ subunits was performed. D, inhibition of G␤␥-dependent GIRK-currents. HEK 293 cells stably transfected with G-protein-activated inwardly rectifying K ϩ channel 1/4 (GIRK) were transfected with ␣ 2A -adrenergic receptor and PhLP S cDNA, and whole cell currents were obtained by patch clamping. Basal GIRK currents and agonist (1 M norepinephrine)-induced GIRK currents were determined by using Ba 2ϩ blocking of the channel. Basal GIRK currents were reduced by 50.1 Ϯ 9.1% (n ϭ 15) by PhLP S and agonist-induced GIRK currents were reduced by 34.9 Ϯ 11.3% (n ϭ 15) by PhLP S (means Ϯ S.E.; *, p Ͻ 0.05 versus control). ence of PhLP S we analyzed the interaction of G␤ and G␥ by co-precipitation. In contrast to PhLP L , expression of PhLP S led to a strongly decreased interaction of G␤ with G␥ as determined by the failure of precipitated AGT-G␤ 1 to co-precipitate AGT-G␥ 2 in the presence of PhLP S (Fig. 4C). Since efficient interaction of G␤ and G␥ is indicative of the proper folding of the G␤ subunit (28), the lack of G␤-G␥ interaction is suggestive of G␤ misfolding induced by PhLP S .
Structural Determinants in PhLP S for G␤␥ Down-regulation-We further analyzed the mechanism of PhLP S on G␤␥ and therefore consecutively truncated the N terminus of PhLP S to investigate the structural requirements for the G␤␥-downregulating effect of PhLP S in intact cells (Fig. 5A). The Cterminal constructs PhLP 119 -301 and PhLP 132-301 were able to inhibit inositol phosphate generation to the same extent as was PhLP S (Fig. 5B). Further truncating the N terminus (PhLP 145-301) decreased G␤␥-regulating effects. We therefore concluded that the ability of PhLP S to regulate G␤␥ was dependent on the function of ␣-helix 3 together with the Cterminal half of PhLP. Recently, it was reported that PhLP L can bind and inhibit the cytosolic chaperonin complex (CCT; a hexadacamer consisting of eight different subunits) via binding to one of its subunits named TCP-1␣ (14). It was demonstrated that PhLP L inhibited the folding of luciferase and the G tprotein ␣ subunits, but a role in G␤␥ folding has not been addressed so far. To analyze whether TCP-1␣ does also interact with PhLP S , we performed co-immunoprecipitations of TCP-1␣ with PhLP S . Immunoprecipitation of PhLP S and the C-terminal constructs revealed that TCP-1␣ interacted with the same constructs that caused inositol phosphate inhibition (Fig. 5C) and G␤␥ down-regulation (data not shown). In order to investigate whether TCP-1␣ plays a role in the folding of G␤␥, we performed additional knock-down experiments with the help of siRNA technology (for a review see Ref. 29). Fig. 5D shows a representative experiment where the specific knock-down of endogenous TCP-1␣ in HEK-TSA cells led to down-regulation of transfected G␥ 2 subunits. Cells transfected with siRNA (siTCP-1␣ or, for control, siGFP) were unaffected in terms of expression of other transfected (PLC␤ 2 ) or endogenous proteins (␤-actin). We therefore conclude that CCT inhibition either by PhLP S or by siRNA-induced protein reduction of the subunit TCP-1␣ had similar consequences on the levels of functional G␤␥-complexes in intact cells. Analysis of endogenous G␤ 1 (Fig.  5E) showed that knock-down of TCP-1␣ reduced the expression not only of co-transfected but also of endogenous members of the G␤␥ complex.
Effects on Endogenous G␤␥ Subunits-To show that the effect of PhLP S on G␤␥ (suggested to occur by induction of G␤␥misfolding) was also effective with endogenously expressed G␤␥, we investigated function and down-regulation of endogenous G␤ subunits in HEK 293 cells. In order to investigate potential effects of PhLP S on mRNA synthesis, total RNA was collected from HEK 293 cells, which were transfected and treated as for inositol phosphate determination. RNA was reverse-transcribed, and quantitative PCR of the human G␤ 1 cDNA was performed. No differences in G␤ 1 cDNA levels between control and PhLP S -transfected cells were detected (Fig.  6A), indicating that PhLP S did not exert an inhibition of G␤ mRNA synthesis. We then determined PLC␤ 2 -mediated inositol phosphate generation in HEK 293 cells and found that PhLP S was effective in inhibiting inositol phosphate generation mediated by endogenously expressed G␤␥ subunits (Fig. 6B). We measured the endogenous G␤ content by Western blotting and found that PhLP S decreased G␤ (Fig. 6C). To further analyze the functional role of PhLP S on G␤␥ signaling, we measured ␣ 2A -adrenergic receptor-mediated signaling. Exper-iments were performed in HEK 293 cells stably expressing the G-protein-activated inwardly rectifying potassium channel composed of subunits 1 and 4 (GIRK1/4). Basal and agonistinduced (1 M norepinephrine) currents of GIRK channels have been shown to depend on G␤␥ function (30 -32). Both signals were significantly inhibited by PhLP S (Fig. 6D). This indicates that PhLP S effectively inhibits signals by endogenous G␤␥ complexes stimulated by a G-protein-coupled receptor.

DISCUSSION
In this work, we show that inhibition of G␤␥-mediated signaling by PhLP occurs through at least two different mechanisms: (a) through direct binding by a high affinity N-terminal G␤␥ binding site (PhLP L ; Figs. 1, 2, and 7) or (b) through inhibition of, most likely, G␤␥ folding and subsequent downregulation of G␤ and G␥ protein levels (Figs. 3, 4, and 7). The down-regulation (as a putative indicator of misfolded G␤␥) was a specific phenomenon seen with PhLP S and the phosphorylation-deficient mutants of PhLP L (bearing the A18 -20 mutation). This finding would imply that regulation of PhLP (either by dephosphorylation of the constitutively phosphorylated PhLP L or by alternative splicing to produce PhLP S ) can switch the G␤␥-regulatory function of PhLP from direct binding toward inhibition of G␤␥-folding (Fig. 7). We further show that the diminished protein levels of G␤ and G␥ were sensitive to modification of the N terminus. This finding is in agreement with a recent report that the G␥ subunit is subject to regulation by the proteasome via the N-end rule pathway (27). In line with this, we observed a partial reversal of the down-regulation by PhLP S in intact cells by the proteasome-specific inhibitor lactacystine (Fig. 3C). We also show that endogenous G␤ was effectively down-regulated by PhLP S and that this resulted in inhibition of PLC␤ and receptor-activated GIRK activity.
Recently, it was reported that PhLP L selectively inhibited the function of CCT, which is involved in the folding of actin, tubulin, and G␣ subunits but has been suggested to participate, more importantly, in folding of WD40 repeat proteins like the G-protein ␤ subunit (33). However, a role in G␤␥ subunit folding remained to be elucidated (14). We now demonstrate FIG. 7. Proposed mechanism of G␤␥ subunit regulation by PhLP: Two faces of a regulator. The constitutively phosphorylated long isoform PhLP L binds to G␤␥ and inhibits its effects on a G␤␥-dependent effector such as PLC␤ 2 (1). A second type of G␤␥ regulation renders newly synthesized G␤ subunits nonfunctional, leading to the degradation of G␤␥ subunits. This effect appears to be mediated by an interaction with the TCP-1␣ subunit of CCT (2), which is involved in the folding of WD40 repeat proteins (like the G␤ subunit). Switching between the two mechanisms is exerted either by alternative splicing (since PhLP S only weakly binds to G␤␥) or by CK2-dependent phosphorylation of PhLP (since P-PhLP L does not inhibit CCT). that the siRNA-induced gene silencing of the CCT subunit TCP-1␣ resulted in a reduction of both endogenous G␤ 1 and co-transfected G␤␥ subunits. Since G␤␥ down-regulation was the prominent feature seen with PhLP S , this observation suggests that PhLP S -dependent G␤␥-down-regulation occurred via inhibition of TCP-1␣-mediated G␤␥ folding. To add further evidence to this contention, we show that the G␤␥-down-regulating domains of PhLP S and the TCP-1␣-binding domains are identical (Fig. 5).
Given the fact that PhLP L was constitutively phosphorylated by CK2 in HEK cells and diverse tissue types (12), and combined with the observation that phosphorylated PhLP L was not able to down-regulate G␤␥, inhibiting the function of CCT could be the crucial step in regulating G␤␥ function, when PhLP is dephosphorylated or alternatively spliced. Thus, PhLP can inhibit G␤␥-mediated effects via two entirely different mechanisms: direct G␤␥ binding and impairment of G␤␥ folding. The switch between these two mechanisms is regulated either at the level of PhLP synthesis by alternative splicing or at the post-translational level by CK2-dependent phosphorylation (Fig. 7). Such a switch between two entirely different types of regulation is, to our knowledge, without precedent and illustrates that PhLP is a unique and complex regulator of Gprotein-mediated signaling.