Gβ5γ2 Is a Highly Selective Activator of Phospholipid-dependent Enzymes

In this study, Gβ specificity in the regulation of Gβγ-sensitive phosphoinositide 3-kinases (PI3Ks) and phospholipase Cβ (PLCβ) isozymes was examined. Recombinant mammalian Gβ1–3γ2 complexes purified from Sf9 membranes stimulated PI3Kγ lipid kinase activity with similar potency (10–30 nm) and efficacy, whereas transducin Gβγ was less potent. Functionally active Gβ5γ2 dimers were purified from Sf9 cell membranes following coexpression of Gβ5 and Gγ2-His. This preparation as well as Gβ1γ2-His supported pertussis toxin-mediated ADP-ribosylation of Gαi1. Gβ1γ2-His stimulated PI3Kγ lipid and protein kinase activities at nanomolar concentrations, whereas Gβ5γ2-His had no effect. Accordingly, Gβ1γ2-His, but not Gβ5γ2-His, significantly stimulated the lipid kinase activity of PI3Kβ in the presence or absence of tyrosine-phosphorylated peptides derived from the p85-binding domain of the platelet derived-growth factor receptor. Conversely, both preparations were able to stimulate PLCβ2 and PLCβ1. However, Gβ1γ2-His, but not Gβ5γ2-His, activated PLCβ3. Experimental evidence suggests that the mechanism of Gβ5-dependent effector selectivity may differ between PI3K and PLCβ. In conclusion, these data indicate that Gβ subunits are able to discriminate among effectors independently of Gα due to selective protein-protein interaction.

G-protein-coupled receptors are part of a major signal recognition system in mammalian cells that perceives a vast variety of external physical and chemical entities (1). These stimuli provoke cellular responses, thereby regulating almost any cellular function, including secretion, cell movement, growth, and differentiation. Accordingly, hundreds of ligands interact with Ͼ1000 of these heptahelical receptor proteins (2,3). In turn, ligand-occupied G-protein-coupled receptors couple to dozens of heterotrimeric G-protein combinations, which transduce the incoming signal into the cell (4,5). Both GTP-bound G␣ and G␤␥ modulate an ever increasing number of enzymes, transporters, or ion channels (6). On the G-protein level, signaling is controlled by the GTPase activity of G␣; stimulated by G-protein-coupled receptors; and fine-tuned by modulatory elements, including RGS (regulators of G-protein signaling) proteins (7). Hence, the guanine nucleotide-sensitive signal transduction machinery represents a highly complex network that also cross-talks to other systems such as growth factorlinked signaling systems. Different levels of finicky organization are required for a synchronized function. These include spatial and temporal expression of constituents as well as their specific and selective interaction. Accordingly, studies in living cells employing antisense RNA have demonstrated a remarkable specificity in the interaction between receptors and Gproteins (8). However, signaling specificity in whole cells appears to be significantly higher as compared with isolated and reconstituted components, which points to additional factors such as cell compartmentation or cell-specific protein expression patterns (9,10). Nevertheless, selective protein-protein interactions obviously represent the first step in signaling specificity.
G␣ heterogeneity correlates with specificity in G-proteineffector coupling. This is reflected by the subdivision of G␣ isoforms into four subfamilies that activate (G␣ s ) or inhibit (G␣ i/o ) adenylyl cyclases, stimulate phospholipase C␤ isozymes (G␣ q ), or regulate Rho proteins (G␣ 12/13 ). A corresponding structure-function relationship is missing for G␤␥ dimers despite the existence of 7 G␤ and 12 G␥ isoforms, which are grouped into two and three subfamilies, respectively, and probably form hundreds of distinct combinations. Only T D ␤␥ 1 (G␤ 1 ␥ 1 ) differs, being consistently less potent in its ability to regulate effectors as compared with most other G␤␥ combinations (6). This difference is thought to be due to the C 15 farnesyl moiety of G␥ 1 , whereas more potent isoforms are C 20 geranylgeranylated. Hence, the lipophilic modification is assumed as an important criterion determining the potency by which G␤␥ complexes modulate effectors. Furthermore, G␤ 1-4 isoforms show a high amino acid sequence identity from 79 to 90%, which may in part explain the observed low level of effector specificity. Nevertheless, major contact sites with effectors are located on the G␤ subunit, which represent the structural basis to establish selective protein-protein interactions (11).
G␤ 5 , the youngest member of the G␤ subunit family, appears to be an exception. Its amino acid sequence exhibits only 53% identity to other G␤ isoforms with an insertion of additional 13 amino acids at the N terminus (12). This region is thought to be important for dimerization with G␥ as well as effector coupling, recommending G␤ 5 as the candidate isoform to examine signaling specificity. In fact, a rather unique coupling pattern of G␤ 5 has been reported. It not only binds to G␥, but also dimerizes with RGS-6, -7, -9, and -11 proteins through interaction with a G␥-like domain (13)(14)(15)(16)(17). Reconstitution experiments indicated that G␤ 5 complexed to G␥ may selectively interact with G␣ q -coupled receptors through specific association of G␤ 5 with G␣ q proteins (18,19). Moreover, G␤␥ transfection studies have suggested differences in the regulation of downstream signaling pathways such as mitogen-activated protein kinase and c-Jun N-terminal kinase by G␤ 5 ␥ 2 as compared with G␤ 1 ␥ 2 (20). However, the molecular mechanisms for this unusual behavior remain unclear.
To examine the possibility of a so far unrecognized G␤ 5 -dependent specificity in effector regulation, we tested purified preparations of G␤ 5 complexed to G␥ 2 for their ability to regulate PI3K and PLC␤ isozymes. Heterodimeric G␤ 5 ␥ 2 was functionally active, as it supported pertussis toxin-catalyzed ADPribosylation of purified G␣ i1 . However, G␤ 5 ␥ 2 did not stimulate the lipid and protein kinase activities of PI3Ks. As a consequence, G␤ 5 ␥ 2 also failed to activate voltage-operated calcium channels in rat portal vein myocytes. Surprisingly, G␤ 5 ␥ 2 exhibited a remarkable specificity within different PLC␤ isozymes. It potently and efficiently stimulated PLC␤ 2 , and it stimulated PLC␤ 1 , but not PLC␤ 3 , with similar potency but less efficacy, suggesting that G␤ subunits are able to discriminate among highly related isozymes.

EXPERIMENTAL PROCEDURES
Recombinant PI3Ks and PLC␤ Isoforms-Construction of recombinant baculoviruses for expression of PI3K subunits and PLC␤ isoforms has been described (21)(22)(23)(24). PLC␤ 2 ⌬, a deletion mutant of human PLC␤ 2 lacking a C-terminal region necessary for stimulation by G␣ q , is indistinguishable from wild-type PLC␤ 2 in terms of its interaction with PI-4,5-P 2 , Ca 2ϩ , and G␤␥ dimers (22). The cDNA of human PLC␤ 3 cloned into the EcoRI site of a pBluescript vector was kindly provided by M. van Asseldonk (Department of Human Genetics, University Hospital Nijmegen, Nijmegen, The Netherlands) (25). The construct was digested with EcoRI and ligated into this site in the pVL1393 transfer vector. Recombinant baculoviruses were produced as described previously. Expression and isolation of PI3K and PLC␤ isoforms were carried out according to published protocols (22,24).
G␤␥ Complexes and Phosphotyrosyl Peptides-Purification of native G␤␥ complexes and T D ␤␥ from membranes isolated from bovine brains and retinal rod outer segments, respectively, have been described elsewhere (26,27). Baculoviruses encoding recombinant G␤ isoforms and G␥ 2 were generated and characterized as detailed previously (28,29). Membrane-associated recombinant G␤ x ␥ 2 complexes were purified by combined affinity and subunit exchange chromatography, followed by anion-exchange chromatography on Mono Q columns according to published protocols (30).
Purified proteins were quantified by Coomassie Blue staining following SDS-PAGE with bovine serum albumin as the standard. Integrity of recombinant G␤␥ dimers was analyzed by rechromatography and pertussis toxin-catalyzed ADP-ribosylation of G␣ i1 (27). The tyrosine-phosphorylated peptide used in this study, CGGY(P)MDMSKDES-VDY(P)VPMLDM, was based on that of the human platelet-derived growth factor receptor (31) and kindly donated by Dr. Andreas Steinmeyer (Schering AG, Berlin). A non-phosphorylated peptide served as a control and had no effect on PI3K enzymatic activity.
Gel Electrophoresis, Immunoblotting, and Antibodies-Generation and characterization of antisera against p101 and G␤ subunits (AS 398 and AS 422) were detailed elsewhere (28,32,33). The polyclonal anti-p110␤ and monoclonal anti-p85 antibodies were purchased from Santa Cruz Biotechnology (Heidelberg, Germany). Antisera specific for human PLC␤ 2 and PLC␤ 3 were a kind gift of Dr. P. J. Parker and were prepared as described elsewhere (34). Preparations containing PLC␤ 2 , PLC␤ 3 ; p110␤, p110␥, p101, or p85␣; G␤ and G␥ proteins were fractionated by SDS-PAGE and transferred to nitrocellulose membranes (Millipore, Eschborn, Germany). Visualization of specific antisera was performed using the ECL chemiluminescence system (Amersham Pharmacia Biotech, Freiburg, Germany) according to the manufacturer's instructions.
Lipid Vesicle Pull-down Assay-Experimental conditions for determination of G␤␥ and PI3K association on lipid vesicles were similar to measurement of the enzymatic activity of PI3K. In brief, 35 l of PI-4,5-P 2 -containing lipid vesicles (320 M phosphatidylethanolamine, 300 M phosphatidylserine, 140 M phosphatidylcholine, 30 M sphingomyelin, and 50 M PI-4,5-P 2 ) in buffer E (40 mM HEPES (pH 7.4), 120 mM NaCl, 1 mM EGTA, 7 mM MgCl 2 , 1 mM dithiothreitol, 1 mM ␤-glycerophosphate, and 0.1% bovine serum albumin) were mixed with different concentrations of G␤␥ complexes or vehicles only. After incubation on ice for 10 min, equal amounts of PI3K␤ (5 l) were added and thoroughly mixed. After an additional 10 min on ice, 10 l of reaction buffer (buffer E containing 40 M ATP) were added, followed by incubation at 30°C for 15 min. The mixture was put on ice and centrifuged for 2 min at 800 ϫ g at 4°C. The supernatant and pellet were separated and supplemented with electrophoresis sample buffer according to Laemmli (35). Thirty l of either solution were subjected to SDS-PAGE and transferred to nitrocellulose membranes. Semiquantitative analysis was realized by immunoblotting using specific antisera against p110␤ and G␤ subunits. As a control, PI3K␤ lipid kinase activities were measured in parallel experiments. The results confirmed sensitivity of PI3K␤ lipid kinase activity to G␤ 1 ␥ 2-His , but not to G␤ 5 ␥ 2-His .
Measurement of Enzymatic Activities-The lipid kinase activity of PI3K isoforms was determined as detailed previously (23,24). It was ensured that the effects of G␤␥ on PI3K activity were not affected by their vehicles due to suppression of enzymatic activity by detergents.
Protein Kinase Assay of PI3K␥-This was performed as described (24) with modifications. The assay volume was 25 l (2-3 Ci of [␥-32 P]ATP/tube) and usually contained 7 mM Mg 2ϩ . Lipid vesicles were devoid of PI-4,5-P 2 . The reaction was stopped with 25 l of ice-cold twice concentrated sample buffer according to Laemmli (35). Following separation on SDS-polyacrylamide gel, proteins were transferred to nitrocellulose membranes. Dried membranes were exposed to Fuji imaging plates, and autoradiographic signals were quantitated with a Fuji BAS 1500 imager (Raytest, Straubenhardt, Germany).
Phospholipase C␤ Activities-These were quantified as described previously (22) with minor modifications. In brief, 5 l of purified G␤␥ were supplemented with 10 l of soluble fraction of insect cells; infected with baculoviruses encoding PLC␤ isoforms (0.3-4.0 g of protein/ sample); and incubated for 5 or 30 min at 25°C in a volume of 60 l containing 50 mM HEPES (pH 7.2), 70 mM KCl, 3 mM EGTA, 2 mM dithiothreitol, 33 M [ 3 H]PI-4,5-P 2 (5 Ci/mol), and 536 M phosphatidylethanolamine. In control experiments, we ascertained that soluble fractions from wild-type baculovirus-infected insect cells exhibited no G␤␥-sensitive PLC activity. Crucial experiments were repeated using purified PLC␤ 2 ⌬ (22) and showed results identical to those obtained with soluble fractions of PLC␤ 2 ⌬ baculovirus-infected insect cells.
Measurement of Ca 2ϩ Currents-Isolated myocytes from rat portal vein were obtained by enzymatic dispersion as described previously (36). Cells were seeded at density of ϳ10 3 cells/mm 2 on glass slides in physiological solution and used the same day. Reconstitution experiments including voltage clamp and membrane current recordings were carried out exactly as described previously (32). Cell capacitance was determined in each cell tested, and current density was expressed as the maximal Ba 2ϩ current amplitude per capacitance unit (pA/picofarads). Experiments were performed at 30 Ϯ 1°C. The physiological solution used to record Ba 2ϩ currents contained 130 mM NaCl, 5.6 mM KCl, 1 mM MgCl 2 , 5 mM BaCl 2 , 11 mM glucose, and 10 mM HEPES (pH 7.4) with NaOH. The basic pipette solution contained 130 mM CsCl, 10 mM EGTA, 5 mM Na 2 ATP, 2 mM MgCl 2 , and 10 mM HEPES (pH 7.3) with CsOH. Bovine serum albumin (0.1%) was added in the pipette solution to increase protein infusion and had no effect by itself on the current charge density.
Presentation of Data-Averaged data are given as means Ϯ S.D. if not stated otherwise. Statistical significance was tested with Student's t test for paired or unpaired data.

Different Sensitivities of PI3K Isoforms to G␤␥ Complexes-
The starting point of this study was the intriguing observation that in COS cells, G␤ 5 , in contrast to G␤ 1 , failed to signal to the mitogen-activated protein kinase and c-Jun N-terminal kinase pathways following cotransfection with G␥ 2 (20). In this cell line, different PI3Ks have been identified as essential elements in the activation of mitogenic signaling cascades by G-proteincoupled receptors (37)(38)(39). Since we and others (23,24,49) have found that G␤␥ stimulates PI3K activity by direct interaction with PI3K␤ and PI3K␥, we now examined the coupling of different G␤␥ isoforms to these lipid kinases by employing purified preparations (Fig. 1). By doing so, we readily noticed significant differences in the sensitivity of heterodimeric PI3K␤ and PI3K␥ to bovine brain G␤␥ (G␤␥ bov ) complexes ( Fig.  2A). Whereas the lipid kinase activity of PI3K␥ was activated by G␤␥ bov with EC 50 values lower than 20 nM, the corresponding values for PI3K␤ were calculated to be at least 100 nM. In contrast, both enzymes showed a similar sensitivity to G␤␥ when stimulated in the absence of their particular non-catalytic subunit, i.e. p101 for p110␥ and p85 for p110␤ (Fig. 2B). Furthermore, p110␤ was activated by G␤␥ with identical EC 50 values regardless of which substrate was used, i.e. phosphatidylinositol or PI-4,5-P 2 (data not shown). As expected, the presence of the p85 adaptor did not affect G␤␥-induced stimulation of p110␤. In its heterodimeric state, PI3K␤ produced PI-3,4,5-P 3 much more efficiently when co-stimulated with phosphotyrosyl peptides resembling an intracellular p85-binding region of the platelet-derived growth factor receptor (data not shown, but see below). Taken together, these results confirm and extend our previous findings that the G␤␥ complex directly interacts with the catalytic subunit of different PI3Ks and that the non-catalytic p101 subunit of PI3K␥ sensitizes p110␥ for G␤␥ in a PI-4,5-P 2 -dependent fashion (24).
Stimulatory Activity of Defined G␤␥ Dimers on PI3K␥-Since G␤␥ bov purified from G i/o proteins represents a mixture of G␤ and G␥ isoforms, we expressed defined G␤␥ subsets in Sf9 cells. Earlier studies have revealed that the G␥ subunit affects the ability of the G␤␥ complex to modulate effectors (40,41). We therefore coexpressed G␤ isoforms with G␥ 2 , which was previously shown to associate efficiently with G␤ isoforms, and subsequently purified membrane-associated G-proteins from Sf9 cells by using the approach introduced by Kozasa and Gilman (30). Thereafter, purified recombinant G␤ 1 ␥ 2 , G␤ 2 ␥ 2 , and G␤ 3 ␥ 2 as well as native T D ␤␥ were tested for their ability to stimulate the lipid kinase activity of PI3K␥ (Fig. 3). The results demonstrated stimulatory activity of all four preparations tested, allowing several immediate conclusions. First, no marked differences in the potency between the three G␤ isoforms complexed to G␥ 2 were visible. Second, as noticed for other effector systems, T D ␤␥ was less potent (EC 50 Ͼ 500 nM) than G␤␥ combinations containing a G␥ predicted to be geranylgeranylated. Third, EC 50 values of ϳ30 nM were found for G␤ x ␥ 2 preparations, indicating that recombinant G␤␥ dimers exhibited potencies almost indistinguishable from that of native G␤␥ bov . It should be noted that despite some reports describing G␤ 3 ␥ 2 as an unfavorable combination (42,43), G␤ 3 ␥ 2 in this study was expressed as a G␣ i1-His ␤ 3 ␥ 2 heterotrimer in Sf9 cells and was isolated as a functional active heterodimer following exposure to aluminum fluoride. In contrast, we could not examine the shorter splice variant of G␤ 3 , i.e. G␤ 3(s) , since, in our hands, it did not form a complex with G␣ i1-His and G␥ 2 . Purified G␤ 3 ␥ 2 stimulated not only PI3K␥, but also PLC␤ 2 and PLC␤ 3 enzymatic activities in parallel with G␤␥ bov (Fig. 3 and  data not shown). These data corroborate a recent report that G␤ 1-3 complexed to G␥ 2-7 were equally effective in stimulating basal PLC activity following transfection of COS-1 cells (44) but do not support an earlier cotransfection study in COS-7 cells in which G␤ 3 ␥ 2 , unlike other G␤␥ dimers, was unable to stimulate inositol phosphate production and mitogen-activated protein kinase activity (45). For possible explanations of this discrepancy, it should be noted that the earlier study did not compare G␤ expression levels. Nevertheless, accumulating experimental evidence raises the possibility that G␤ 3 distinguishes among G␤␥-regulated effectors. G␤ 3 , unlike G␤ 1 and G␤ 2 , appeared to be almost incapable of inhibiting N-type calcium channels following microinjection of G␤␥ expression plasmids into superior cervical ganglion neurons of rats (46). Instead, expression of G␤ 3 led to a near doubling of the Ca 2ϩ current density. Furthermore, isoform specificity in the interaction of G␤ 3 with effectors was shown by demonstrating that the G-protein-coupled receptor kinase 2 carboxyl terminus bound G␤ 1 and G␤ 2 , but not G␤ 3 , whereas a G-protein-coupled receptor kinase 3 fusion protein bound all three G␤ isoforms (47).

Stimulatory Activity of G␤ 1 ␥ 2-His and G␤ 5 ␥ 2-His on PI3K
Isoforms-To study the effector coupling of G␤ 5 ␥ 2 , we applied a different strategy for its purification. In contrast to G␤ 3(s) , G␤ 5 formed a heterotrimer with co-infected G␣ i1-His and G␥ 2 in Sf9 cells since G␤ 5 ␥ 2 specifically dissociated from this complex upon exposure to aluminum fluoride. Unfortunately, G␤ 5 came off the final Mono Q column apart from G␥ 2 . 2 Anion-exchange chromatography was indispensable to remove Sf9 insect G␤␥, which stimulated PI3K␥ with similar potency and efficacy as G␤␥ bov . 3 Hence, we alternatively obtained G␤ 5 ␥ 2 dimers by using hexahistidine-tagged G␥ 2 . As a control, we purified G␤ 1 ␥ 2-His , which, like G␤ 5 ␥ 2-His , was able to support pertussis toxin-catalyzed ADP-ribosylation of purified bovine brain G␣ i1 (Fig. 4A). Furthermore, we confirmed the integrity of the G␤ 5 ␥ 2-His preparations under assay conditions (Figs. 4 (B and C) and 6C), ruling out dissociation of the dimer during the experiment. Next, we tested both G␤ 1 ␥ 2-His and G␤ 5 ␥ 2-His for their ability to stimulate PI3K␥ harboring two enzymatic entities. Since evidence is accumulating that PI3K␥ may signal through both lipid and protein kinase activities (48), we tested the effect of G␤␥ dimers on either enzymatic quality. Although G␤ 1 ␥ 2-His was slightly less potent than the untagged counterpart, it significantly stimulated the formation of PI-3,4,5-P 3 by PI3K␥ (EC 50 ϭ 30 nM) (Fig. 5A). In contrast, G␤ 5 ␥ 2-His was ineffective at all concentrations tested. A small enhancement of PI-3,4,5-P 3 formation at the highest G␤ 5 ␥ 2-His concentration was most likely due to minute contamination by Sf9 G␤␥. Autophosphorylation of the catalytic p110 subunit of PI3K␥ was stimulated by G␤ 1 ␥ 2-His with high potency and efficacy, whereas G␤ 5 ␥ 2-His was again ineffective (Fig. 5B). To check whether the inability of G␤ 5 ␥ 2-His to stimulate PI3K␥ was due to the failure of interaction, we examined the capability of G␤ 5 ␥ 2-His to interfere with G␤␥ bov -induced stimulation of PI3K␥ lipid kinase activity. G␤ 5 ␥ 2-His at concentrations up to 400 nM (Ͼ10-fold excess) had no inhibitory effect, suggesting that G␤ 5 ␥ 2-His did not compete with G␤␥ bov for binding to PI3K␥ (see below).
We also examined PI3K␤, which, unlike PI3K␥, is widely distributed in mammalian tissues. As described above, G␤␥ bov FIG. 2. Different sensitivities of G-protein-regulated class I PI3Ks to G␤␥. A, purified recombinant heterodimeric PI3K␤ (E) and PI3K␥ (q) were stimulated with increasing concentrations of bovine brain G␤␥ using PI-4,5-P 2 as a substrate. 32 P-Labeled lipid products were isolated and quantified as described under "Experimental Procedures." Data are means Ϯ S.D. of -fold stimulation. Basal activities were 0.047 Ϯ 0.003 and 0.138 Ϯ 0.014 mol⅐min Ϫ1 ⅐mol Ϫ1 for PI3K␤ and PI3K␥, respectively. B, the same experimental conditions were used as described for A, except that recombinant monomeric p110␤ (E) and p110␥ (q) were employed. The basal activities were found to be 1.05 Ϯ 0.14 and 0.25 Ϯ 0.03 mol⅐min Ϫ1 ⅐mol Ϫ1 for p110␤ and p110␥, respectively.

FIG. 3. Stimulation of PI3K␥ activity by distinct G␤␥ isoforms.
Recombinant PI3K␥ was assayed for its sensitivity to increasing concentrations of purified recombinant G␤ 1 ␥ 2 (q), G␤ 2 ␥ 2 (), G␤ 3 ␥ 2 (f), and native T D ␤␥ (E). The isolation and quantification of PI-3,4,5-P 3 were performed as described in "Experimental Procedures." Shown is one typical experiment out of three. and phosphotyrosyl peptides acted in a cooperative manner to activate PI3K␤, suggesting different binding sites for either stimulus (24,49). Whereas G␤ 1 ␥ 2-His stimulated PI-3,4,5-P 3 FIG. 4. A, effect of G␤ 5 ␥ 2 on pertussis toxin-catalyzed ADP-ribosylation of G␣ i1 . G␣ i1 (150 nM) purified from bovine brain was ADP-ribosylated in the absence or presence of 300 nM purified G␤ 1 ␥ 2-His or G␤ 5 ␥ 2-His using pertussis toxin. The reactions were stopped by addition of an equal volume of 2ϫ concentrated electrophoresis sample buffer and subjected to SDS-PAGE, followed by Western blotting. For visualization, dried nitrocellulose membranes were autoradiographed and analyzed using a PhosphorImager. Pertussis toxin-catalyzed 32 P incorporation into G␣ i1 was enhanced by G␤ 1 ␥ 2-His and G␤ 5 ␥ 2-His by 45-and 25-fold, respectively. B, binding of purified G␤ 5 ␥ 2-His to Ni 2ϩ -NTAagarose. Purified G␤ 5 ␥ 2-His preparations tested for stimulation of PI3K and PLC isoforms were checked for integrity by rebinding proteins to Ni 2ϩ -NTA-agarose. An aliquot of G␤ 5 ␥ 2-His (pool) was incubated with, washed (wash), and eluted from (imidazole) Ni 2ϩ -NTA-agarose as described under "Experimental Procedures." Comparable aliquots (15 l) of the loaded pool (1 ml) and eluted fractions (100 l) were subjected to SDS-PAGE and visualized by Coomassie Blue staining. Recovery of eluted G␤ 5 ␥ 2-His was calculated to be 70 -80% of the loaded pool. Similar results were obtained using G␤ 1 ␥ 2-His (data not shown). C, the retention of G␤ 5 on Ni 2ϩ -NTA-agarose is dependent on G␥ 2-His . Comparable amounts of G␤ 5 ␥ 2-His and G␤ 5 were bound to and eluted from Ni 2ϩ -NTAagarose as described for B. Eluted fractions were subjected to SDS-PAGE, followed by Western blotting to nitrocellulose membranes. Proteins were detected using G␤ 5 -specific antisera (AS 422).

FIG. 5. G␤ 5 ␥ 2-His does not stimulate PI3K␥ enzymatic activities.
A, lipid kinase activity. G␤ 5 ␥ 2-His (q) and G␤ 1 ␥ 2-His (E) were examined for their ability to stimulate the lipid kinase activity of PI3K␥ in a concentration-dependent fashion. 32 P-Labeled PI-3,4,5-P 3 was isolated, separated, and quantified as described under "Experimental Procedures." Shown are an autoradiograph of one representative experiment (top) and means Ϯ S.D. from three independent experiments (bottom). B, protein kinase activity. Purified recombinant PI3K␥ was assayed for autophosphorylation in response to increasing concentrations of either G␤ 5 ␥ 2-His (q) or G␤ 1 ␥ 2-His (E). An autoradiograph of one representative experiment is depicted on the top. Incorporation of 32 P into the catalytic p110␥ subunit is illustrated as -fold stimulation of basal activities (means Ϯ S.D.) in three independent experiments (bottom). formation in the absence (Fig. 6A) as well as presence (Fig. 6B) of phosphotyrosyl peptides, G␤ 5 ␥ 2-His did not stimulate lipid kinase activity under either condition. To further support our findings on the selective interaction of G␤␥ isoforms with PI3K␤, we examined the ability of G␤ 1 ␥ 2-His and G␤ 5 ␥ 2-His to recruit PI3K␤ to phospholipid vesicles (Fig. 6C). Notably, some PI3K␤ was associated with lipid vesicles even in the absence of G␤␥ (Fig. 6C, pellet). However, increasing concentrations of G␤ 1 ␥ 2-His significantly enhanced the association of PI3K␤ with liposomes, whereas G␤ 5 ␥ 2-His failed, although both G␤ 1 ␥ 2-His and G␤ 5 ␥ 2-His associated equally well with lipid vesicles. Taken together, our data strongly suggest that PI3Ks represent a G␤␥-regulated effector system completely spared by G␤ 5 . G␤ 1 ␥ 2 , but Not G␤ 5 ␥ 2 , Stimulates Calcium Channels in Smooth Muscle Cells-To study the physiological relevance of our observations, we examined smooth muscle cells from rat portal veins, which were previously shown to express both G␤ 1 and G␤ 5 (50). In this cell system, we recently found a novel receptor-induced G␤␥-mediated pathway for activation of Ltype voltage-operated calcium channels depending on PI3K FIG. 6. G␤ 5 ␥ 2-His does not activate PI3K␤. A, stimulation of lipid kinase activity. Purified heterodimeric PI3K␤ was assayed for its sensitivity to increasing concentrations of G␤ 1 ␥ 2-His (E) or G␤ 5 ␥ 2-His (q). The isolation, separation, and quantification of PI-3,4,5-P 3 were performed as described under "Experimental Procedures." The autoradiograph of one representative experiment is shown on the top. A corresponding concentration-response curve depicting means Ϯ S.D. is displayed on the bottom. B, G␤␥-mediated activation of Tyr(P) peptide-stimulated lipid kinase activity. The experiments were carried out as described for A, except that PI3K␤ was co-stimulated by increasing amounts of G␤ 1 ␥ 2-His (E) or G␤ 5 ␥ 2-His (q) together with 100 nM tyrosine-phosphorylated peptide derived from the platelet-derived growth factor receptor. Ordinates indicate either the -fold stimulation of activity based on sole stimulation with 100 nM tyrosine-phosphorylated peptide (left axis) or -fold stimulation based on the basal activity of PI3K␤ in the absence of the Tyr(P) peptide and G␤␥ (right axis). C, G␤ 1 ␥ 2-His (but not G␤ 5 ␥ 2-His ) enhances the vesicular association of PI3K␤. G␤␥ isoforms were tested for their ability to recruit PI3K␤ to lipid vesicles. For assay conditions, see "Experimental Procedures." Aliquots of pelleted vesicles and the supernatant were subjected to SDS-PAGE, followed by Western blotting. Proteins were visualized using specific antisera. Note that for semiquantitative estimation of the relative protein distribution to lipid and aqueous phase, pelleted vesicles were resuspended in the same volume as the supernatant prior to electrophoresis. stimulation (32). Using a combined electrophysiological and biochemical approach, we confirmed our earlier findings that stimulation of barium currents by infusion of purified G␤␥ bov complexes into smooth muscle cells is quantitatively blocked by wortmannin pretreatment of cells (Fig. 7A). To examine the G␤␥ isoform specificity of this pathway, we infused recombinant G␤␥ preparations at concentrations known to elicit barium currents. Steady-state barium currents were obtained within 2-4 min after establishment of the whole cell recording mode, as described previously (32). Whereas G␤ 1 ␥ 2-His significantly stimulated current density, G␤ 5 ␥ 2-His showed no effect (Fig. 7B). Furthermore, co-infusion of G␤ 1 ␥ 2-His together with PI3K␥ at concentrations that are ineffective when infused separately readily enhanced barium current density. However, similar experiments with G␤ 5 ␥ 2-His showed no stimulation of barium currents in portal vein myocytes (Fig. 7C). This led us to conclude that endogenous G␤ 5 present in these smooth muscle cells would not activate voltage-operated calcium channels offering signaling specificity encoded in the direct protein-protein interaction of G␤␥ and its effector.
We first ascertained that PLC␤ 2 ⌬ and PLC␤ 3 were both sensitive to nanomolar concentrations of G␤␥ bov (Fig. 8A). Furthermore, T D ␤␥ and G␤ 3 ␥ 2 stimulated both PLC isoforms (data not shown). As expected from experiments with turkey PLC (19), G␤ 5 ␥ 2-His and G␤ 1 ␥ 2-His stimulated human PLC␤ 2 ⌬ with similar efficacies (Fig. 8B). The G␤␥ concentrations required for stimulation were in the same range as those reported by other groups (59), suggesting that either G␤␥ preparation was correctly folded and processed. Surprisingly, although G␤ 5 ␥ 2-His activated PLC␤ 2 ⌬, it consistently failed to stimulate inositol formation by PLC␤ 3 at any concentration tested (Fig. 8C). This inability of G␤ 5 ␥ 2-His to stimulate PLC␤ 3 was not simply caused by the presence of the hexahistidine tag since G␤ 1 ␥ 2-His stimulated PLC␤ 3 and PLC␤ 2 ⌬ with a similar potency and efficacy. Moreover, under our experimental conditions, we found that nanomolar concentrations of G␤ 5 ␥ 2-His and G␤ 1 ␥ 2-His exhibited FIG. 7. Specificity of L-type Ca 2؉ channel activation at the level of G␤␥ dimers in vascular smooth muscle cells. A, wortmannin sensitivity of G␤␥-induced stimulation of L-type Ca 2ϩ channels in rat portal vein myocytes. Experiments were carried out as described under "Experimental Procedures." Traces represent typical Ba 2ϩ currents elicited by depolarization to ϩ10 mV from a holding potential of Ϫ40 mV under control conditions (left) and following infusion of purified bovine brain G␤␥ complexes (400 nM in the pipette solution) into cells similar stimulatory effects on PLC␤ 1 activity (data not shown). Therefore, these data unequivocally indicate a qualitatively different regulation of PLC␤ isoforms by G␤ 5 , suggesting the possibility that G␤ 5 exhibits a more restricted signaling specificity than other G␤ isoforms. Hence, two questions arise from our observations: what is the molecular basis for the observed specificity, and what functional impact may these findings have?
Molecular and Functional Implications of the Results-Much effort has been undertaken to define the structural elements of G␤ responsible for effector coupling (11, 60 -63). This led to the conclusion that, in fact, different G␤␥-regulated effectors as well as G␣ have multiple and overlapping binding sites on the G␤ surface. For instance, results from site-directed mutagenesis suggested that G␤ structures involved in activation of PLC␤ 2 should contain residues on the top surface and in the outer strands of blades 2, 6, and 7 of the G␤ propeller (62,63). More interestingly, distinct mutants of G␤ 1 showed a contrary ability to stimulate the closely related PLC␤ 2 and PLC␤ 3 isoforms. Unfortunately, the residues examined are highly conserved among all G␤ isoforms, which hampers more detailed reflections on structure-function relationship in the G␤ regulation of PLC␤ isoforms at this point.
Little, if any, is known about the PI3K-binding sites on G␤. However, based on our previous work (23), one can assume that they should overlap with G␣ or other effector-binding sites such as those for adenylyl cyclase or PLC␤. Accordingly, the predicted PI3K-interacting area of G␤ 1 should be significantly different in G␤ 5 , precluding interaction of G␤ 5 with this effector class. Assessing the impact of G␤-binding sites on the effector enzymes is even more complicated since structural features responsible for interaction of PLC␤ and PI3K isoforms with G␤␥ are either poorly defined or completely unknown (23,64,65).
As already outlined, little is known about the functional role of G␤ 5 . Nevertheless, G␤ 5 regulates a restricted number of effectors as compared with other G␤ isoforms due to the fact that it either activates effectors such as PLC␤ 2 upon direct binding or leaves out other effectors, e.g. class I PI3Ks, since it lacks interaction. In addition, effector binding without signal transfer may represent a third tempting possibility of effector modulation by G␤ 5 complexed to G␥. In this scenario, it functions as a competitor by antagonizing other G␤␥ complexes that are able to both bind and transfer signals. This assumption is supported by a recent report suggesting that the G␤ surface interacting with PLC␤ 2 can be resolved into two different entities, i.e. a general binding motif and a signal transfer region (66). Indeed, we obtained data indicating different mechanisms explaining why G␤ 5 ␥ 2-His failed to stimulate PI3Ks and PLC␤ 3 . G␤ 5 ␥ 2-His did not interact with PI3Ks, but, in contrast, did interact with PLC␤ 3 since it inhibited the stimulatory activity of G␤ 1 ␥ 2-His on this effector (Fig. 8D). Therefore, receptor-FIG. 8. Selective modulation of PLC␤ isoforms by G␤ 5 ␥ 2 . A, activation of PLC␤ 2 ⌬ (Ⅺ) and PLC␤ 3 (f) by G␤␥ bov complexes. Assays were carried out as described under "Experimental Procedures." Note different scaling of product formation for PLC␤ 2 (left axis) and PLC␤ 3 (right axis). B, stimulation of PLC␤ 2 enzymatic activity by distinct recombinant G␤␥ dimers. PLC␤ 2 ⌬ was assayed for its sensitivity to increasing concentrations of G␤ 1␥2-His (E) and G␤ 5 ␥ 2-His (q). Isolation and quantification of radioactive products were performed as described under "Experimental Procedures." Basal activity was 1.23 Ϯ 0.02 pmol/ min. C, stimulation of PLC␤ 3 enzymatic activity by recombinant G␤␥ dimers. Analogous experiments were carried out as described for B employing PLC␤ 3 . D, G␤ 5 ␥ 2-His was assayed for its ability to compete with G␤ 1 ␥ 2 for binding to PLC␤ 3 . Enzymatic activity was stimulated with G␤ 1 ␥ 2-His (40 nM) and inhibited with increasing concentrations of G␤ 5 ␥ 2-His (q). In contrast, G␤ 5 ␥ 2-His did not inhibit prestimulated (G␤␥ bov , 30 nM) PI3K␥ lipid kinase activity (E). Shown are the means Ϯ S.D. activated G␤ 5 ␥ 2 may affect G␤␥-dependent downstream signaling in three different ways: (i) by inducing signal transfer to effectors upon binding; (ii) by lacking binding and signal transfer; or (iii) by antagonizing G␤␥ signaling by binding to effectors, but without signal transfer.