Purification and Characterization of Gβγ-responsive Phosphoinositide 3-Kinases from Pig Platelet Cytosol*

A G-protein βγ subunit (Gβγ)-responsive phosphoinositide 3-kinase (PI 3-kinase) was purified approximately 5000-fold from pig platelet cytosol. The enzyme was purified by polyethylene glycol precipitation of the cytosol followed by column chromatography on Q-Sepharose fast flow, gel filtration, heparin-Sepharose, and hydroxyapatite. The major Gβγ-responsive PI 3-kinase is distinct from p85 containing PI 3-kinase as the activities can be distinguished chromatographically and immunologically and is related to p110γ as it cross-reacts with anti-p110γ-specific antibodies. The p110γ-related PI 3-kinase cannot be activated by G-protein αi/o subunits, and it has an apparent native molecular mass of 210 kDa. The p110γ-related PI 3-kinase phosphorylates phosphatidylinositol (PtdIns), phosphatidylinositol 4-phosphate (PtdIns4P), and phosphatidylinositol 4,5-bisphosphate (PtdIns(4,5)P2). The apparent K m values for ATP were found to be 25 μm with PtdIns, 44 μm with PtdIns4P, and 37 μm with PtdIns(4,5)P2 as the substrate. Gβγ subunits did not alter the K m of the enzyme for ATP; however,V max increased 2-fold with PtdIns as substrate, 3.5-fold with PtdIns4P, and 10-fold with PtdIns(4,5)P2. Under basal conditions the apparent K m values for lipid substrates were 64, 10, and 15 μm for PtdIns, PtdIns4P, and PtdIns(4,5)P2, respectively. In the presence of Gβγ subunits the dependence of PI 3-kinase activity on the concentrations of lipid substrates became complex with the highest level of stimulation occurring at high substrate concentration, suggesting that the binding of Gβγ and lipid substrate (particularly PtdIns(4,5)P2) may be mutually cooperative. Wortmannin and LY294002 inhibit the Gβγ-responsive PI 3-kinase activity with IC50 values of 10 nm and 2 μm, respectively. Unlike the p85 containing PI 3-kinase in platelets, the p110γ-related PI 3-kinase is not associated with a PtdIns(3,4,5)P3 specific 5-phosphatase. The p85-associated PI 3-kinase was not activated by Gβγ alone but could be synergistically activated by Gβγ and phosphotyrosyl platelet-derived growth factor receptor peptides. This may represent a form of coincidence detection through which the effects of tyrosine kinase and G-protein-linked receptors might be coordinated.

The first form of PI 3-kinase to be purified and cloned was identified as a heterodimer composed of a 110-kDa catalytic subunit with a tightly bound regulatory subunit of 85 kDa (11)(12)(13)(14). The p85 subunit possesses a number of regions with homology to recognized signaling proteins. These include a Bcr homology domain, a Src homology (SH) 3, and two SH2 domains and two proline-rich regions. Binding of the SH2 domains to phosphotyrosine residues within the sequence context, YMXM, which occurs in a wide range of activated growth factor receptors and adaptor proteins, causes the translocation and activation of the catalytic subunit (2,(15)(16). More recently, a number of distinct p85 and p110 subunit isoforms have been identified (11,14,17), but the functional significance of this heterogeneity is not yet clear (18).
Heterotrimeric G-protein-regulated forms of PI 3-kinase have also been identified following the observations that activation of G-protein-coupled receptors in neutrophils and platelets caused a rapid accumulation of PtdIns(3,4,5)P 3 (19 -21).
Stephens et al. (22) partially purified a G-protein ␤␥ subunit (G␤␥)-responsive PI 3-kinase from a myeloid cell line (U937). This enzyme is immunologically and biochemically distinct from a growth factor-regulated, p85-containing PI 3-kinase present in the same cells. Using degenerate oligonucleotide primers based on conserved regions of known PI kinases, Stoyanov et al. (23) isolated a p110 homologue, designated p110␥, from a U937 cell cDNA library. Recombinant p110␥ has a predicted molecular mass of 120 kDa and can be activated in vitro by both G␤␥ and the ␣ subunits of transducin and G i . p110␥ differs from the ␣ and ␤ isoforms mainly by its lack of the recognized p85 binding site which is replaced by a region which has been proposed to resemble a pleckstrin homology (PH) domain (23).
A G␤␥-responsive PI 3-kinase has also been reported to be present in platelet cytosol (24). Because this activity associated with a PDGF receptor phosphotyrosyl peptide and was immu-noprecipitated with a monoclonal antibody raised against the p85 subunit of PI 3-kinase, it was thought to possess a p85related subunit. However, Zhang et al. (25) reported a G␤␥stimulated PI 3-kinase in platelets that was not recognized by p85-directed antibodies. The latter study established that this platelet G␤␥-stimulated enzyme was immunologically related to p110␥.
The substrate specificities of identified PI 3-kinases vary substantially (11, 14, 22, 24, 26 -28). To define the molecular characteristics and properties of G␤␥-stimulated PI 3-kinases in platelets more clearly, we have now partially purified the major form of this enzyme from pig platelets. This enzyme is a p110␥-related PI 3-kinase that is distinct from p85-associated species and that phosphorylates all three potential phosphoinositide substrates with a marked preference for PtdIns(4,5)P 2 which is further enhanced by G␤␥.
As the results appeared to contradict previous detection of a G␤␥-responsive p85-containing PI 3-kinase in human platelets (24), we also studied the p85-containing PI 3-kinase in pig platelets. We found that this enzyme can be activated by G␤␥ in a manner that is largely dependent upon the presence of a tyrosine-phosphorylated PDGF receptor peptide.

Methods
Preparation of G␤␥-The major G-protein ␤␥ subunits were purified from cholate extracts of bovine brain membranes as described by Sternweis and Robishaw (39). The G␤␥ subunits were stored in 20 mM Tris, 1 mM EDTA, 0.1% Genaple C-100 and were more than 95% pure as determined by SDS-PAGE. The G␤␥ preparation was flash-frozen in 10-l aliquots and stored at Ϫ80°C until use.
Purification of the G␤␥-responsive PI 3-Kinase from Platelet Cytosol-Platelet cytosol was derived from 12 liters of freshly drawn pig blood. The detailed procedure for preparation of platelet cytosol was essentially the same as described (24). Platelets were sonicated in 120 ml of lysis buffer (10 mM HEPES, pH 7.4, 1 mM EGTA, 0.2 mM EDTA, 3 mM MgCl 2 , 10 mg/ml each of antipain and pepstatin, 1 mM each of DTT, sodium orthovanadate, phenylmethylsulfonyl fluoride, and benzamidine). The platelet lysate was centrifuged at 35,000 rpm for 1 h, and the resulting supernatant (platelet cytosol, 1049 mg of protein) was kept. G␤␥-responsive PI 3-kinase was precipitated by 5-15% PEG in buffer A (20 mM HEPES, pH 7.4, 0.2 mM EDTA, 3 mM MgCl 2 , 10 mg/ml each of antipain and pepstatin, 1 mM each of DTT, sodium orthovanadate, phenylmethylsulfonyl fluoride, and benzamidine). The pellet was resuspended in 36 ml of buffer A. The PEG sample (36 mg) was loaded onto a Q-Sepharose Fast Flow column (150 ϫ 15 mm), pre-equilibrated with buffer A, and eluted with a gradient of 0 -0.5 M NaCl (150 ml). G␤␥-responsive PI 3-kinase fractions (12 mg) were pooled and loaded onto a gel filtration column (Sepharose CL-4B, 100 ϫ 2.6 cm) preequilibrated with buffer B (buffer A plus 100 mM NaCl and 10% sucrose). The G␤␥-responsive PI 3-kinase activity fractions (1.35 mg protein) were pooled and loaded onto a heparin-Sepharose column (50 ϫ 10 mm) pre-equilibrated with buffer B. The column was washed with 30 ml of buffer B and eluted with a linear gradient of 100 -500 mM NaCl (60 ml). Fractions (0.36 mg) containing G␤␥-responsive PI 3-kinase were pooled and concentrated to 6 ml with Prepcentricon 30 (Amicon, Inc. Beverly, MA). Half of the sample was loaded onto a hydroxyapatite column (100 ϫ 10) pre-equilibrated with buffer C (20 mM K 2 HPO 4 , pH 7.0, 5 mM DTT, 0.1 mM each of phenylmethylsulfonyl fluoride and benzamidine); protein was eluted with 20 -750 mM K 2 HPO 4 during which the G␤␥-responsive PI 3-kinase was separated from the tyrosine kinase-regulated PI 3-kinase. Separation of the G␤␥-responsive PI 3-kinase from the tyrosine kinase-regulated PI 3-kinase can also be achieved by incubating half of the concentrated enzyme with 2 ml of protein G-Sepharose pre-coupled with anti-p85 antibodies overnight at 4°C with gentle agitation. The G␤␥-responsive PI 3-kinase prepared by either method did not contain any p85 protein nor any other detectable lipid kinase activity.
PI 3-Kinase Assay-Generally the enzyme activity was measured by adopting the following assay procedure. 10 l of platelet cytosol or column fractions were mixed with 30 l of lipid vesicles, which had been premixed with G␤␥ or their vehicle for 10 min on ice. 10 l of MgATP was added to start the reaction. The enzyme reaction was terminated after incubating at 37°C for 5 min by adding 200 l of 1 M HCl. To prepare lipid vesicles, equimolar amounts of PS and substrate lipid (PtdIns, PtdIns4P, or PtdIns(4,5)P 2 ) were dried onto a film under vacuum and probe-sonicated (3 ϫ 15 s with 1 min on ice between sonication, at setting 20 -30 on a Jencons Ultrasonic Processor) into kinase assay buffer (40 mM HEPES, pH 7.4, 1 mM EGTA, 1 mM DTT, 50 mM NaCl, 4 mM MgCl 2 ). The standard assay contained 100 M PtdIns(4,5)P 2 and PS, 10 -100 M ATP (10 M with purified fraction and 100 M with crude extract), 1 M G␤␥ or its vehicles and 10 Ci of [␥ 32 P]ATP. Lipid extraction and analysis was performed as described (24). The products of the PI 3-kinase reactions were identified by deacylation and separation of their glycerol derivatives by high performance liquid chromatography and compared with deacylated 3 H-labeled standards. Note that assays were done under first order conditions with respect to ATP as substrate (to optimize assay of radioactive product) ensuring that no more than 10% of ATP was consumed during a reaction.
For assays requiring activated G␣-proteins, G␣-proteins were incubated for 1 h on ice in the presence of 100 M GTP␥S and 5 mM MgCl 2 , mixed with PtdIns-containing lipid vesicles, and incubated again for 10 min on ice, before adding to the assay.

Partial Purification of G␤␥-responsive PI 3-Kinase from
Platelet Cytosol-A G␤␥-responsive PI 3-kinase was purified approximately 5000-fold, with an overall yield of 30%, from porcine platelet cytosol using PEG precipitation, Q-Sepharose, gel filtration, heparin-Sepharose, and hydroxyapatite. A typical purification is summarized in Table I. PEG precipitation resulted in a 30-fold enrichment with 100% recovery of G␤␥responsive activity. Elution of this sample from Q-Sepharose using a continuous salt gradient revealed two distinct peaks of G␤␥-responsive PI 3-kinase (Fig. 1A). The earlier eluting, minor peak of activity was inconsistently observed in a number of purifications from different batches of platelets and was not studied further. Analysis of fractions eluting from the Q-Sepharose column by Western blotting with an anti-p85 monoclonal antibody revealed that the second peak of G␤␥-responsive PI 3-kinase co-eluted with p85 immunoreactivity (Fig. 1B). p85 and G␤␥-responsive activity continued to co-migrate through gel filtration and heparin-Sepharose (data not shown). However, separation of p85 and the G␤␥-responsive PI 3-kinase was achieved through hydroxyapatite eluted with a linear gradient of K 2 HPO 4 /KH 2 PO 4 as shown in Fig. 2, A and B. Separation could also be achieved by immunodepletion of p85 using protein G-Sepharose that had been pre-coupled with anti-p85 antibod-ies (Fig. 3A). The G␤␥-responsive PI 3-kinase is related to p110␥ as it can be recognized by an anti-p110␥ anti-peptide antibody (Fig. 3B).
Characterization of G␤␥-responsive PI 3-Kinase-The native molecular mass of the G␤␥-responsive PI 3-kinase was determined using a calibrated Superose 12 size exclusion column. As shown in Fig. 4, the activity eluted at a volume indicating a size of approximately 210 kDa. The enzyme was not pure at that stage as revealed by silver-stained SDS-PAGE gel of active fractions (not shown). This indicated that the enzyme is a very minor protein in platelet cytosol.
The substrate specificity and kinetic characteristics of the enzyme were consistent between several different preparations. As the enzyme after hydroxyapatite cannot survive freezing and thawing, activity purified through heparin-Sepharose and immunodepleted to remove p85 was used for most experiments. Such preparations were purified approximately 2000fold with respect to platelet cytosol and contained no detectable PtdIns 4-kinase, PtdIns4P 5-kinase, or PLC (EC 3.1.4.3) activ-   The purification began with 1049 mg of platelet cytosol protein from 12 liters of blood. Columns were run as described under "Experimental Procedures." The G␤␥-stimulatable PI 3-kinase activity was defined as the activity in the presence of G␤␥ minus that in the absence of G␤␥ in the assay and is expressed as nmol of PtdInsP 3 formed per min.
Step Total   ities. PtdIns, PtdIns4P, and PtdIns(4,5)P 2 were all phosphorylated at the 3-position (see "Experimental Procedures"), and the phosphorylations of PtdIns and PtdIns4P were inhibited by the presence of PtdIns(4, 5)P 2 as described previously (29) indicating that a single enzyme activity accounts for the phosphorylation of all three substrates. Anti-p85 and anti-p110 antibodies were recently shown to co-immunoprecipitate with a PtdIns(3,4,5)P 3 -specific 5-phosphatase which was presumed to be present in a complex with this form of PI 3-kinase (30). Analysis of the partially purified G␤␥-sensitive PI 3-kinase using 32 P-labeled PtdIns(3,4,5)P 3 as substrate or by monitoring the ratio of PtdIns(3,4,5)P 3 /PtdIns(3,4)P 2 in PI 3-kinase assays failed to detect any 5-phosphatase in such preparations (data not shown).
K m values were determined under basal conditions for ATP and all three lipid substrates. Similar K m values for ATP of 25, 44, and 37 M were determined when PtdIns, PtdIns4P, or PtdIns(4,5)P 2 , respectively, were used as substrates (Fig. 5A and Fig. 6A). K m values for the lipid substrates were determined using 10 M ATP and maintaining a constant mole fraction of 1:1 (substrate/PS). 10 M ATP (i.e. somewhat less than the K m value) was used to estimate the sensitivity of the assays. However, since less than 5% ATP was consumed during incubation, the assays were linear under the conditions determined. The reactions approximated to Michaelis-Menten kinetics with K m values of 64, 10, and 15 M for PtdIns, PtdIns4P, and PtdIns(4,5)P 2 , respectively ( Fig. 5B and Fig. 6B). Similar values were obtained using either Lineweaver-Burk or Wolfe plots.
The effects of increasing concentrations of G␤␥ subunits were examined at an ATP concentration of 10 M with lipid substrate concentrations of 100 M. Interestingly the EC 50 values differed according to the lipid substrate as noted previously for the human platelet cytosol activity (24). When PtdIns(4,5)P 2 was the substrate, an EC 50 of approximately 300 nM was observed, but with PtdIns as the substrate this increased to about 600 nM (Fig. 7). The response to G␤␥ subunits was completely blocked in the presence of a 3-fold calculated molar excess of a preparation of GDP-liganded G␣ i/o and by 85% at an equimolar concentration of these proteins (Fig. 8), suggesting that the activation requires free G␤␥ and hence would require activated heterotrimeric G-proteins in vivo. The effects of 1 M G␤␥ were next examined over a range of substrate concentrations. The presence of G␤␥ did not significantly affect the K m for ATP whichever lipid substrate was employed. However, the effects of increasing lipid substrate concentrations on activity deviated from Michaelis-Menten kinetics in the presence of the activator. Thus, G␤␥ subunits were most effective at high concentrations of lipid especially when PtdIns(4,5)P 2 was used as the substrate. The apparent cooperativity with respect to increasing substrate concentration prevented the determination of K m values and allowed only estimation of V max values. Nevertheless, G␤␥ enhanced V max by approximately 2-, 3.5-, and 10-fold with PtdIns, PtdIns4P, and PtdIns(4,5)P 2 , respectively, as substrates (Fig. 5, A and B).
The fungal metabolite, wortmannin, is a potent inhibitor of several PI 3-kinase isoforms including the G␤␥-sensitive PI 3-kinase from U937 cells (22,26,27), whereas the quercetin analogue, LY294002, was developed as a PI 3-kinase inhibitor that lacked the chemical instability of wortmannin (37). Both of these compounds inhibited the platelet G␤␥-sensitive enzyme with IC 50 values of 10 nM and 2 M for wortmannin and LY294002, respectively (data not shown). The value for wortmannin is very similar to that reported by Stephens et al. (22) for the enzyme from U937 cells and that for LY294002 is almost identical to its IC 50 when assayed with RBL-2H3 cells (36).
Recombinant p110␥ was recently shown to respond to both G␤␥ and the GTP␥S-liganded ␣ subunits of G i and transducin. We repeated these observations using a p110␥-glutathione Stransferase fusion protein expressed in Sf9 cells and immobilized on glutathione-agarose (Fig. 9B). 1 nM G␣ i -GTP␥S, but not GTP␥S alone, induced a 50% increase in p110␥ activity, whereas 1 M G␤␥ enhanced activity 4-fold. By contrast, the platelet enzyme was insensitive to G␣ i but was more sensitive to G␤␥ being activated approximately 8-fold in the experiment shown in Fig. 9A.
The isolation from platelet cytosol of a G␤␥-responsive PI 3-kinase that is distinct from a p85-associated enzyme appeared to contradict our previous observation that G␤␥ activated a PI 3-kinase which associated with a biotinylated phosphotyrosyl peptide related to the p85 binding region of the PDGF receptor. We therefore investigated the possibility that G␤␥ could activate the p85-associated enzyme in a manner that depended on coincident association with the PDGF receptor peptides. This was indeed found to be the case. As shown in Fig.  10A, G␤␥ alone did not activate immunoprecipated p85-associated PI 3-kinase but greatly augmented the response to the phosphopeptide when assays were carried out under standard conditions. This effect was even more dramatic when substrate lipid was presented against a background of phosphatidylethanolamine rather than PS as described by Okada et al. (40). With PIP 2 /phosphatidylethanolamine vesicles, basal activity was found to be very low and was slightly activated by either G␤␥ or phosphopeptides. The combination of G␤␥ and phosphopeptides, however, resulted in a greater than 50-fold enhancement of PI 3-kinase activity (Fig. 10B). DISCUSSION A G␤␥-sensitive PI 3-kinase has been purified 5000-fold from pig platelet cytosol. A second form of G␤␥-responsive enzyme eluted as an early peak on Q-Sepharose, but the appearance of this peak and its magnitude relative to the major peak of activity were variable between platelet preparations. For these FIG. 4. Determination of the native molecular weight of G␤␥responsive PI 3-kinase. The G␤␥-responsive PI 3-kinase (after hydroxyapatite) was loaded onto a SP 12 column. Fractions (0.5 ml) were collected and assayed for PI(4,5)P 2 3-kinase activity in the presence and absence of 1 M G␤␥. The elution position of G␤␥-responsive PI 3-kinase is indicated. To calibrate the column, a range of molecular weight markers was loaded onto the column, and the elution volume of each of the known markers was noted. Void volume of the column was 7 ml (elution of blue dextran). Approximate molecular weight of G␤␥-responsive PI 3-kinase was calculated by plotting K av versus logM r , where K av where V e was the elution volume, V o was the void volume, and V t was the total column volume (24 ml). q, stimulated; E, basal.
reasons the early eluting peak was not analyzed further in this study; whether it represents a distinct species of PI 3-kinase, a processing variant, or a proteolytic fragment are not clear from these studies. The major G␤␥-sensitive enzyme co-eluted with p85 immunoreactivity through Q-Sepharose, gel filtration, and heparin-Sepharose but could be separated from the latter protein on hydroxyapatite. Moreover, separation of the major G␤␥responsive enzyme from p85-associated PI 3-kinase can also be achieved by p85 immunoprecipitation. One further line of evidence that excludes the association of the major G␤␥-sensitive PI 3-kinase with p85 was the lack of PtdIns(3,4,5)P 3 5-phosphatase activity which was previously shown to be complexed to p85 in human platelets (30). Nevertheless, since a polypeptide of approximately 120 kDa was observed on Western blots probed with a p110␥-specific anti-peptide antibody, and its native molecular weight was found to be in excess of 200 kDa, G␤␥-sensitive PI 3-kinase presumably exists either as a dimer or in complex with one or more additional polypeptides. This situation is similar to that reported for the partially purified enzyme from U937 cells (22).
In agreement with the previous report (24) on the G␤␥responsive PI 3-kinase in human platelet cytosol, the p85associated PI 3-kinase in pig platelet cytosol can also be activated by G␤␥ under certain assay conditions. Moreover, this G␤␥-responsive activity can be further synergistically augmented by phosphotyrosine PDGF receptor peptide. This may represent a form of coincidence detection through which the effects on cellular functions of tyrosine kinase and G-proteinlinked receptors might be coordinated. Similar findings have also been reported in human monocytic THB-1 cells (40).
Although the platelet enzyme that was sensitive to G␤␥ alone co-purified with a p110␥-immunoreactive component, it differed from the latter in its native molecular weight as noted above. It also differed in terms of its regulation, being insensitive to GTP␥S-liganded G␣ i/o and being stimulated to a greater degree by G␤␥ compared with recombinant p110␥. Whether these distinguishing features reflect the differences between the native and recombinant proteins, the catalytic subunit itself or the presence of an additional complexed polypeptide(s) cannot be discerned at present. The partially purified platelet enzyme phosphorylated Pt-dIns, PtdIns4P, and PtdIns(4,5)P 2 to give the corresponding 3-phosphorylated lipids as determined by co-chromatography of their deacylation products with authentic standards. The phosphorylation of PtdIns and PtdIns4P could be inhibited by an excess of PtdIns(4,5)P 2 suggesting that a single enzyme species was responsible for the observed phosphorylation of all three substrates (29). However, the efficiency with which these substrates were utilized varied substantially, with polyphosphoinositides exhibiting lower K m values than PtdIns and with V max values being greatest with PtdIns(4,5)P 2 as substrate. These features were exaggerated in the presence of G␤␥ subunits that enhanced PtdIns(4,5)P 2 phosphorylation more markedly than either PtdIns or PtdIns4P. Assuming these features are relevant to the situation in cell membranes, then this enzyme would be expected to synthesize mainly PtdIns(3,4,5)P 3 in vivo, consistent with the observed effects of thrombin on 3-phosphorylated inositol phospholipids in intact platelets (5).
In the absence of G␤␥ the PI 3-kinase activity obeyed Michaelis-Menten kinetics with respect to both ATP and lipid substrates. This was not the case in the presence of G␤␥ which induced sigmoidal kinetics for lipid substrates, especially in the presence of PtdIns(4,5)P 2 . This suggests that this form of PI 3-kinase might possess more than one PtdIns(4,5)P 2 binding site; by analogy with ␤ (31) and ␦ (32) isoforms of phospholipase C (PLC) the additional site(s) might be non-catalytic and function to associate PI 3-kinase at a substrate-bearing membrane and thus allow processive catalysis to occur. Because we only observed sigmoidal kinetics in the activated state, it is proposed that G␤␥ regulates the binding of substrate lipid at such a non-catalytic site. The results with different lipid substrates predict that the putative regulatory lipid site prefers PtdIns(4,5)P 2 over PtdIns4P which in turn is preferred over PtdIns. Interestingly this matches the expected rank order of binding of these lipids to some PH domains (33), a structural feature that has been proposed to occur in p110␥ but not other published forms of PI 3-kinase. A further interesting regulatory feature was the observation that the EC 50 for activation by G␤␥ also was affected by the nature of the lipid substrate suggesting that the binding of lipid (especially PtdIns(4,5)P 2 ) and G␤␥ are mutually cooperative. Such an observation is again reminiscent of the ligand binding properties of some PH domains and closely associated sequences C-terminal to the PH domain proper that appear to possess distinct binding sites for anionic lipids and G␤␥ subunits, respectively. However, it should be pointed out that without structural studies, the proposal that p110␥ possesses a PH domain remains speculative.
An alternative explanation for the observed cooperative kinetics for lipid substrates in the presence of G␤␥ might be that more G␤␥ is complexed with vesicles at the higher lipid concentrations. There are two lines of evidence suggesting that the membrane localization of G␤␥ is important for its function. Katz et al. (34) reported that transfection of COS-7 cells with cDNA for PLC␤2 and G-protein ␤1␥1 subunits caused an increase in PLC activity as evidenced by the accumulation of inositol phosphates. The use of a mutant cDNA encoding a ␥ subunit lacking the essential cysteine residue required for isoprenylation resulted in a shift of the ␤␥ complex to the cytosol and prevented the increase in cellular inositol phosphates. Furthermore, the purification of ␤␥ dimers from baculovirus transfected insect cells has shown that only C-terminally modified ␥ subunits confer PLC␤2-activating function on ␤␥ complexes (35), suggesting that the isoprenylation and carboxyl methylation of ␥ subunits may be important for both membrane location and functionality of the complex. However, the extent of membrane insertion of G␤␥ is unlikely to explain the marked differences between lipid substrates in terms of the cooperativity observed and the extent of activation of PI 3-kinase.
Much can be learned about the molecular diversity of protein families, such as the PI 3-kinases, using cloning strategies that exploit sequence relationships among the family members. Such approaches identified p110␥ and have also revealed a wider family that encompasses both inositol phospholipid and protein serine/threonine kinases. The initial observations that revealed the presence of G-protein-regulated forms of PI 3-kinase, however, were made using partially purified protein preparations from cell extracts. Since the major G␤␥-sensitive forms of PI 3-kinase present in both platelets and myeloid cells appear to be complexed to at least one other polypeptide, which is distinct from p85 and which confers unknown properties on the enzymes, further studies are required to define the molecular components of these native proteins. Such studies are also required to understand the structural basis for the regulatory mechanisms that we have described.