Phospholipase Cβ2 Binds to and Inhibits Phospholipase Cδ1*

Phospholipase Cβ (PLCβ) isoforms, which are under the control of Gαq and Gβγ subunits, generate Ca2+ signals induced by a broad array of extracellular agonists, whereas PLCδ isoforms depend on a rise in cytosolic Ca2+ for their activation. Here we find that PLCβ2 binds strongly to PLCδ1 and inhibits its catalytic activity in vitro and in living cells. In vitro, this PLC complex can be disrupted by increasing concentrations of free Gβγ subunits. Such competition has consequences for signaling, because in HEK293 cells PLCβ2 suppresses elevated basal [Ca2+] and inositol phosphates levels and the sustained agonist-induced elevation of Ca2+ levels caused by PLCδ1. Also, expression of both PLCs results in a synergistic release of [Ca2+] upon stimulation in A10 cells. These results support a model in which PLCβ2 suppresses the basal catalytic activity of PLCδ1, which is relieved by binding of Gβγ subunits to PLCβ2 allowing for amplified calcium signals.

The binding of an agonist to its target G protein-coupled receptor stimulates heterotrimeric G proteins, which in turn can result in an increase in intracellular Ca 2ϩ through the activation of phospholipase C␤ (PLC␤) 1 (1,2). PLCs catalyze the hydrolysis of a minor lipid component, phosphatidylinositol 4,5-bisphosphate (PI(4,5)P 2 ), to release the second messengers diacylglycerol and inositol 1,4,5-trisphosphate (Ins(1,4,5)P 3 ). These messengers in turn activate protein kinase C and stimulate the release of Ca 2ϩ from internal stores. All PLC␤s (␤ 1 -␤ 4 ) are regulated by G␣ q subunits that are coupled to specific sets of G protein-coupled receptors. PLC␤ 2 and to a lesser extent PLC␤ 3 can also be stimulated by G␤␥ subunits (1,2). Because G␤␥ subunits have the potential to be released from all types of G␣-G␤␥ heterotrimers, PLC␤ 2 and -␤ 3 may be activated by a wider range of receptors.
In contrast to other mammalian PLC enzymes, the cellular regulation of PLC␦ enzymes is unclear. It is known that these enzymes are regulated by an increase in cellular Ca 2ϩ levels because PLC␦ is the only PLC family that is not active at basal Ca 2ϩ levels but is strongly activated when cytoplasmic Ca 2ϩ rises above basal levels (1,3). This behavior suggests that PLC␦s function to amplify, rather than initiate, calcium-mobilizing signals. At maximum Ca 2ϩ concentrations, the specific catalytic activity of purified PLC␦ 1 is typically 50 -100-fold greater than that of unstimulated PLC␤ or PLC␥. When reconstituted into permeabilized PC12 and HL60 cells, PLC␦ 1 , but not PLC␤ 1 or PLC␥ 1 , shows substantial activation by physiologic calcium levels (4). Overexpression of PLC␦ 1 in PC12 and Chinese hamster ovary cells enhances the increase in cellular Ca 2ϩ and soluble inositol phosphate levels produced by bradykinin (5) and thrombin (6). The Ca 2ϩ -amplification role of PLC␦ 1 has been clearly defined in keratinocytes derived from PLC␦ 1 -null mice in which the sustained elevation of cytoplasmic Ca 2ϩ that follows a PLC␥ 1 -stimulated rapid rise does not occur but can be reconstituted when PLC␦ 1 is introduced (7).
Several studies suggest that activation of PLC␦ 1 is under more complex control than the simple rise in cytoplasmic Ca 2ϩ . In frog oocytes expressing thrombin and platelet-derived growth factor receptors, microinjection of PLC␦ 1 antibody specifically inhibits thrombin but not platelet-derived growth factor-induced calcium mobilization (8). In Chinese hamster ovary cells, overexpression of PLC␦ 1 enhances the amount of inositol phosphates generated by ionomycin, but this increment is much smaller than the increase observed during thrombin stimulation (6). Similar results are obtained in bradykininstimulated PC12 cells expressing high levels of PLC␦ 1 (5). Here, raising calcium with high extracellular potassium, thapsigargin, or ionomycin induces a measurable increase in inositol trisphosphate, yet this increment is substantially less than that observed with a maximum dose of bradykinin. Thus, although these observations support a generalized amplification hypothesis, they suggest that receptor-generated signals other than calcium also contribute to PLC␦ 1 -dependent inositol phosphate generation.
Whereas protein regulators of most mammalian PLCs have been identified, those for PLC␦ 1 have not been well established. A novel form of RhoGAP associates strongly with PLC␦ 1 in cell lysates (9), stimulating its catalytic activity at low levels of calcium (0.1 M). There is compelling evidence that PLC␦ 1 is also regulated by an atypical GTP-binding protein, G h , or transglutaminase (10). G h is controlled by ␣ 1 -adrenergic receptor ␣ 1 b and ␣ 1 d in heart and liver (11)(12)(13), as well as oxytocin receptors in myometrium (14). PLC␦ 1 stimulates GDP/GTP exchange on TGII/G h (11,15), and reciprocally, G h allows PLC␦ 1 to be stimulated at lower Ca 2ϩ concentrations. Although these studies show that TGII/G h couples heptahelical receptors to PLC␦ 1 , the fraction of the total cellular response that this represents is unclear.
Other regulators of PLC␦ 1 have been suggested. Association of PLC␦ 1 to GAP43 results in a rise in cytoplasmic Ca 2ϩ , although it is uncertain whether this is because of increased membrane association of PLC␦ 1 (16). The activity of PLC␦ 4 is suppressed by a inactive PLC␦4-Alt3 variant (17), but it is unclear whether comparable regulators of other PLC␦ isozymes exist.
Most PLC␤ and PLC␦ isozymes are widely expressed with varying tissue distribution. PLC␤ 2 , which is the focus of this study, is expressed at high levels in cells of hematopoietic origin. Mice lacking this isozyme have abnormal chemokine signaling, possibly through G␤␥ released from G␣ i/o -coupled receptors (18). PLC␦ 1 is the most widely distributed PLC␦ isotype and is most strongly expressed in skeletal muscle, some layers of the skin, the spleen, testis, and lung (1,3,19). In the adult brain, PLC␦ 1 is found primarily in glial cells (20,21); however, like PLC␤ 2 , its presence is low in most neurons.
Previous work in our laboratories has focused on characterizing the regulation of PLC␦ 1 and PLC␤ 2 . Here we report that PLC␤ 2 inhibits PLC␦ and that this inhibition is relieved upon binding of G␤␥ to PLC␤ 2 . Our results suggest a novel mechanism in which G protein stimulation has the ability to amplify PLC␤ 2 -generated Ca 2ϩ signals through simultaneous activation of PLC␦ 1 under some cellular conditions.

MATERIALS AND METHODS
In Vitro Fluorescence Studies-G␤ 1 ␥ 2 subunits and PLC␤ 2 were expressed in Sf9 cells by using a baculovirus system (22). G␤ 1 ␥ 2 subunits were reconstituted into preformed lipid bilayers by simple addition (23). PLC␦ 1 was bacterially expressed and purified (24). Association measurements between membrane-bound proteins were carried out in large unilamellar vesicles composed of 67% anionic 1-palmitoyl-2-oleoyl-Lphosphatidylserine and 33% 1-palmitoyl-2-oleoyl-L-phosphatidylcholine, where both PLC enzymes are membrane-bound above 250 M total lipid (data not shown) (25,26). Protein-protein association was assessed using the fluorescence methods described previously (27). Briefly, proteins were covalently labeled with either acrylodan or coumarin (26,28). The labeled proteins were excited at 340 nm and scanned from 380 to 500 nm. The emission intensity was taken from the integrated area of the spectrum. We found that the emission intensity of labeled PLC␤ 2 showed a substantial and reproducible increase upon the addition of unlabeled PLC␦ 1 and gave a titration curve that showed the appropriate shift in midpoint when the initial concentration of acrylodan-PLC␤ 2 was changed, thereby reflecting protein-protein association. Proteinprotein associations were also assessed by fluorescence resonance energy transfer using coumarin-and DABCSYL-labeled proteins and were carried out similarly by using the methods and analysis described previously (28).
PLC Activity Measurements-PLC activity measurements were conducted by using purified proteins and detergent-mixed micelles substrates as described previously (26,29).
Cell Culture and Overexpression of PLCs-HEK293 cells and A10 cells were maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum and antibiotics. For A10 cells, the medium was supplemented with 1 mM sodium pyruvate at 37°C in 5% CO 2 . For fluorescence studies, HEK293 cells were cotransfected with BiFC-PLC␤ 2 /PLC␦ 1 vectors (15 g/10 6 cells in 60-mm dish) by calcium phosphate coprecipitation. HEK293 cells transfected with either BiFC-PLC␤ 2 /PLC␦ 1 vectors or vectors lacking the BiFC tags were used in the Ca 2ϩ and inositol phosphate studies as both produced identical results. A Western blot showing the levels of expression is given in Fig. 3. In the pertussis toxin (PTX)-treated cells, HEK293 cells were incubated with 100 ng/ml for 16 -20 h.
Imaging of Fluorescence in Living Cells-24 h after transfection, cells from 60-mm dishes were plated into Labtek chambered glass coverslips (Bio-Rad). Fluorescence emission from living cells was measured 48 -72 h after transfection on a Zeiss Axiovert microscope using 63ϫ 1.4 numerical aperature oil objective.
Emission spectra of BiFC complexes from a 100-mm dish of living cells were taken on an ISS spectrofluorometer (Champaign, IL). Cells expressing BiFC-PLC␤ 2 /PLC␦ 1 were placed in a 1-cm cuvette with stirring at an adjusted concentration of 1 ϫ 10 6 cells/ml. Spectra were taken at exc ϭ 500 nm and scanning from 535 to 600 nm. Cut-off filters (525 nm, Corion Optical) were placed before the monochromators to help reduce the amount of scattered light. Background samples were obtained from cultures transfected with only one of the BiFC constructs. Background spectra, which contributed 12-22% of the signal, were subtracted from the corresponding sample spectra.
For immunostaining, cells were grown on Labtek chambers, fixed in 4% formaldehyde solution in PBS, and permeabilized in 0.1% Triton X-100. Cells were blocked in PBS containing 5% goat serum, 1% BSA, and 50 mM glycine overnight. The monoclonal antibody against PLC␦ 1 (Upstate Biotechnology, Inc.) was used as the primary antibody. Polyclonal antibody against PLC␤ 2 (Santa Cruz Biotechnology) was also used as the primary antibody. Primary antibodies were diluted 1:200 in PBS containing 0.5% BSA. Cells were incubated in primary antibody at 37°C for 1 h. This was followed by three washes of 7 min each in PBS. Secondary antibodies were diluted at 1:2000 in PBS, 0.5% BSA. Fluorescein isothiocyanate-conjugated rabbit secondary antibody was to detect PLC␦ 1 antibody, and DSRed-conjugated anti-mouse secondary antibody was used to detect antibody against PLC␤ 2 . After 1 h of incubation at 37°C, the cells were washed by PBS (three times at 7 min each). Finally, PBS was added to the cells, and the specimens were viewed under the Zeiss Axiovert fluorescence microscope.
Preparation of BiFC-PLC␤ 2 /PLC␦ 1 Membranes-Cells expressing BiFC-PLC␤ 2 /PLC␦ 1 were harvested and centrifuged at 500 ϫ g for 5 min. The pellet was washed with PBS, resuspended in lysis buffer (10 mM Tris, pH 7.4, 1 mM MgCl 2 , 1 mM dithiothreitol, 1 mM phenylmethylsulfonyl fluoride, 10 g/ml pepstatin, and 10 g/ml leupeptin), and homogenized. The lysate was then centrifuged at 500 ϫ g for 5 min, and the supernatant was further centrifuged at 50,000 ϫ g for 30 min at 4°C. The pellet containing the cell membranes was resuspended, and the protein concentration was adjusted to 0.5 mg/ml. Both purified G␤␥ subunits and PLC␤ 2 were briefly dialyzed in 20 mM Hepes, 0.16 M KCl, 1 mM dithiothreitol, and 1 mM EGTA, pH 7.4, to remove detergent in the G␤␥ storage buffer.

Measurement of Cellular [Ca 2ϩ ] i -Cellular [Ca 2ϩ
] i was determined with the fluorescent calcium indicator dye fura-2/AM in an ISS spectrofluorometer. Briefly, cell monolayers were washed with Hanks' balanced salt solution (HBSS) (118 mM NaCl, 5 mM KCl, 1 mM CaCl 2 , 1 mM MgCl 2 , 5 mM glucose, 15 mM Hepes, 1% BSA, pH 7.4), and the cells were detached by a buffer stream. Suspended cells were labeled with 1 M fura-2/AM for 45 min at 37°C in HBSS with rotation. Thereafter, cells were washed twice and incubated in HBSS for another 20 min. After centrifugation, the cells were resuspended at a density of 1 ϫ 10 6 cells/ml and were measured in a continuously stirred cell suspension at room temperature. The ratio of fluorescence emitted at 340 and 380 nm (340:380 ratio) of cells was converted to Ca 2ϩ concentration by the method of Tsien and co-workers (30) using the relation: is the fluorescence at that wavelength in EDTA buffer, and F min is the fluorescence in the presence of detergent and excess calcium. R is the measured ratio, and R min and R max are the ratios corresponding to the EDTA and detergent/excess calcium conditions, respectively. In some experiments, [Ca 2ϩ ] i was determined in the absence of extracellular Ca 2ϩ by treatment cell with 1 mM EGTA.
Measurements of Inositol Phosphate Formation and Phosphoinositide Analysis in Intact Cells-Cells were prelabeled with myo-[ 3 H]inositol (1 Ci/ml) for 2 days in inositol-free medium. The cells were harvested, kept in suspension, and incubated for 10 min at 37°C with HBSS/LiCl (118 mM NaCl, 5 mM KCl, 1 mM CaCl 2 , 1 mM MgCl 2 , 5 mM D-glucose, and 15 mM Hepes, pH 7.4, supplemented with 10 mM LiCl), before challenging with agonist. For agonist stimulation, the cells were incubated with agonist (5 M carbachol) in the presence of LiCl for 30 min at 37°C. All reactions were stopped by removing the incubation medium and lysing the cells in 1 ml of ice-cold methanol. After the addition of 1 ml of chloroform and 0.5 ml of H 2 O, phase separation was performed by centrifugation at 2000 ϫ g for 10 min at 4°C. The aqueous upper phase was applied to AG 1-X8 anion exchange columns to isolate myo-[ 3 H]inositol phosphate formation. For determination of phosphoinositide levels, the lower phase was collected and evaporated by vacuum drying, and the lipids were resuspended in 50 l of chloroform and spotted onto LK5D linear-k silica gel TLC plates (Whatman). The plates were developed in chloroform, methanol, 2.5 M ammonium hydroxide (9:7:2, v/v). The areas corresponding to authentic PtdIns (R f ϭ 0.64), PtdIns(4)P (R f ϭ 0.45), and PtdIns(4,5)P 2 (R f ϭ 0.25) were scraped into vials, and the radioactivity was measured by liquid scintillation counting. The amount of disintegrations/min was normalized for protein content.

PLC␤ 2 and PLC␦ 1 Strongly Interact on Model Membrane
Surfaces-This study began with the unexpected finding that PLC␤ 2 and PLC␦ 1 form heteromeric complexes in solution and on membrane surfaces. We directly measured the formation of PLC␤ 2 ⅐PLC␦ 1 complexes by using fluorescence methods to quantify the association energy between proteins. Protein-protein association was monitored by the change in the fluorescence emission of the probe acrylodan covalently linked to PLC␤ 2 (see "Materials and Methods"). This probe undergoes an increase in emission intensity and energy upon the association of G␤␥ subunits with PLC␤ 2 and yields a dissociation constant identical to that obtained by using fluorescence resonance energy transfer (27). In Fig. 1A, we show the normalized increase in acrylodan-PLC␤ 2 fluorescence when PLC␦ 1 is added in solution. This increase was not observed when buffer or unlabeled PLC␤ 2 was substituted for PLC␦ 1 . These data can be fit to a bimolecular association constant that gives an apparent affinity of K app ϳ200 nM.
Both enzymes are capable of membrane binding, and it is expected that membrane binding would increase their affinity because of a reduction in dimensionality. PLC␦ 1 was added to acrylodan-PLC␤ 2 under conditions where both enzymes were completely membrane-bound (49). Under these conditions, a stronger apparent affinity of K app ϭ 9.2 Ϯ 4 nM was observed due to the concentrating effect of being bound to a membrane surface (see "Discussion"). The affinity of PLC␦ 1 for PLC␤ 2 on membranes was unchanged when the free Ca 2ϩ concentration was raised from 0 to 12 M using Ca 2ϩ /EGTA buffers. Similar studies monitoring the self-association of the two isolated proteins under membrane-bound conditions, which would promote associations, showed that PLC␤ 2 , which appears monomeric by chromatography under our purification conditions (31), does not self-associate at concentrations up to 300 nM. Similarly, PLC␦ 1 does not self-associate at concentrations up to 320 nM FIG. 1. PLC␤ 2 binds to PLC␦ 1 in vitro at a site that is mutually exclusive with G␤␥ subunits. Association of PLC␦ 1 to 50 nM acrylodan-PLC␤ 2 in solution (A) or on 300 M phosphatidylcholine/phosphatidylserine (1:2) lipid bilayers (B) as determined by the increase in acrylodan fluorescence that occurs upon protein association. Experimental data were corrected for dilution and background and fit to a bimolecular association curve to give an apparent K d value (23). Control studies substituted buffer or unlabeled PLC␤ 2 for PLC␦ 1 . B, identical data were obtained at 0 and 12 M free Ca 2ϩ (data not shown). C, decrease in the fluorescence of the energy transfer donor, coumarin-PLC␦ 1 , upon the addition of a nonfluorescent energy transfer acceptor, DABCSYL-PLC␤ 2 , due to transfer. Fluorescence energy transfer, measured by the decrease in intensity, is inhibited in the presence of 10 nM G␤␥. All data are an average of 3-5 measurements, and S.E. is shown.
(equivalent to the highest concentration achieved in the PLC binding measurements). Thus, the association between membrane-bound PLC␤ 2 and PLC␦ 1 is not explained by a tendency of the PLCs to form homomeric complexes. PLC␤ 2 Inhibits the Activity of PLC␦ 1 -We tested whether the association between PLC␤ 2 and PLC␦ 1 affected their enzymatic activities. These studies utilized the substrate [ 3 H]PI(4)P rather than PI(4,5)P 2 . The rationale for using [ 3 H]PI(4)P is that the reduced K m value of PLC␦ 1 toward this substrate as compared with PI(4,5)P 2 allowed us to carry out activity assays at enzyme concentrations above the apparent dissociation constant of PLC␤ 2 -PLC␦ 1 . We found that the catalytic activity of a mixture of PLC␤ 2 and PLC␦ 1 was much less than the additive values obtained from the isolated enzymes ( Fig. 2A), suggesting that the PLC␤ 2 -PLC␦ 1 association inhibits one or both of the enzymes. We thus began a series of studies to determine whether one of the enzymes is inhibited by the other. PLC␦ 1 binds to membranes primarily through a high affinity site for PI(4,5)P 2 /Ins(1,4,5)P 3 in its N-terminal pleckstrin homology (PH) domain in addition to the low affinity catalytic binding site for substrate/product (32). This high affinity site keeps PLC␦ 1 membrane-bound as long as substrate is present in the membrane and the amount of product in the aqueous phase is low. In contrast, the corresponding pleckstrin homology of PLC␤ 2 binds to membranes with little specificity (33). Addition of Ins(1,4,5)P 3 to PLC␦ 1 should greatly reduce its activity because it will inhibit the binding of the PH domain to substrate-containing membranes. Additionally, the presence of product will compete for substrate in the catalytic site, although very high levels of product are needed for inhibition by this latter process. Conversely, product inhibition of PLC␤ 2 will only occur at high levels of product. We measured the ability of Ins(1,4,5)P 3 to inhibit the PLC␤ 2 /PLC␦ 1 mixture, and we found that the activity of the mixtures was not significantly changed. This result suggests that the activity of PLC␦ 1 may be inhibited when bound to PLC␤ 2 .
To better isolate which enzyme is inhibited in the PLC␤ 2 ⅐PLC␦ 1 complex, we chemically treated PLC␦ 1 with diethyl pyrocarbonate (DEPC). This agent forms a covalent adduct with one or both of the catalytic His residues, thereby inactivating the histidine-dependent enzyme (34). Treatment of PLC␦ 1 resulted in complete loss of PLC␦ 1 activity but did not affect its ability to bind substrate membrane, and the apparent affinity with PLC␤ 2 on membrane surfaces fell within the error of its unmodified enzyme (K app ϳ10 nM on 200 M lipid). Most importantly, addition of DEPC-treated PLC␦ 1 to wild type PLC␤ 2 up to 20 nM did not change its rate of PI(4,5)P 2 hydrolysis (Fig. 2B).
We further tested whether PLC␤ 2 inhibits PLC␦ 1 activity by using a point mutant of PLC␤ 2 , H327N. This residue is necessary to stabilize the transition state charge of the phosphate group, and mutation of the corresponding residue on PLC␦ profoundly reduces activity (35,36). We find that the enzymatic activity is reduced 50 -100-fold, and as expected, its binding to membranes and G␤␥ subunits is within error to the wild type enzyme (K app ϳ180 mM in solution). At activating Ca 2ϩ levels, PLC␦ 1 activity decreased ϳ4-fold with increasing amounts of H327N-PLC␤ 2 , with an EC 50 similar to the K d for PLC␦ 1 association (Fig. 2B). These results demonstrate that PLC␤ 2 suppresses the catalytic activity of PLC␦ 1 at activating levels of Ca 2ϩ .
G␤␥ Subunits Disrupt PLC␤ 2 -PLC␦ 1 Association in Vitro-PLC␤ 2 and PLC␦ 1 both bind to G␤␥ subunits, although binding to the former enzyme is of higher affinity and results in enzyme activation (33). We tested whether G␤␥ subunits could interfere with PLC␤ 2 -PLC␦ 1 association by measuring complex formation on membranes in the absence or presence of G␤␥ subunits by fluorescence resonance energy transfer using the coumarin-DABCSYL donor-acceptor pair. Addition of DABC-SYL-PLC␤ 2 to coumarin-PLC␦ 1 resulted in a decrease in donor fluorescence as the two proteins associate (Fig. 1C), but this decrease was largely prevented by the addition of 10 nM G␤␥.
Association of PLC␤ 2 and PLC␦ 1 in Living Cells-To determine whether the two PLC subtypes associate in living cells, we initially used indirect immunofluorescence to view endogenous PLC in fixed rat aorta smooth muscle cells (A10) because these cells express both PLCs (see Ref. 37 and Supplemental Material). Images of these cells showed several pools of colocalized PLC␤ 2 and -␦ 1 that were significantly reduced when the cells were stimulated with acetylcholine before fixing and staining, although the relatively low signal limited interpretation of these images (data not shown). As an alternative approach, we used the technique of bimolecular fluorescence complementation (BiFC) (38). In this method, the N-terminal portion of YFP is linked to a target protein, whereas the Cterminal region encompassing the four missing ␤ strands is linked to a potential binding partner (39). Association between the proteins reconstitutes the fluorescent YFP fluorophore. BiFC-PLC␦ 1 /PLC␤ 2 plasmids were transfected in HEK293 cells, which exhibit relatively low levels of endogenous PLC␦ 1 and -␤ 2. We find that although a small amount of fluorescence is cytosolic, most of the reconstituted YFP fluorescence is confined to the plasma membrane (Fig. 3).
Epifluorescence images of HEK293 cells expressing BiFC-PLC␦ 1 /PLC␤ 2 before and after stimulation with acetylcholine are shown in Fig. 4. An intense yellow was observed, which rapidly decreased following stimulation. In contrast, no fluo- rescence was observed in cells transfected with only one of the BiFC-PLC plasmids or with the BiFC-PLC␤ 2 /Fos pair. Positive controls using BiFC-Fos/Jun showed bright fluorescence restricted to the nucleus (Fig. 4). Also, no fluorescence could be detected for the BiFC-PLC␤ 2 /Fos under conditions where a significant population of Fos was cytosolic (40), suggesting that the BiFC tags were not strongly promoting association between nonphysiologic protein partners.
In Fig. 4 we also show a representative epifluorescence image of a BiFC-PLC␦ 1 /PLC␤ 2 -transfected cell that was stimulated with the G protein-coupled receptor agonist acetylcholine. We find a significant decrease in total BiFC-PLC␦ 1 /PLC␤ 2 fluorescence after correcting for photobleaching in the first few minutes following agonist addition (Fig. 4, graph). At longer times (5-30 min), the intensities of all samples monitored were recovered to near the initial values. These studies demonstrate a dissociation of the BiFC PLC␦ 1 /PLC␤ 2 upon G protein stimulation.
To determine whether the loss in BiFC PLC␦ 1 /PLC␤ 2 fluorescence resulted from specific pools of BiFC PLC␦ 1 /PLC␤ 2 , we monitored changes of the cellular distribution of the proteins in HEK293 cells by confocal microscopy (e.g. Fig. 3). We found also that stimulation of the cells using 1 M carbachol diminished the overall fluorescence. Digitizing the images and comparing the changes in intensity from the plasma membrane population versus the cytosolic compartment suggested that the loss of YFP fluorescence was from the plasma membrane pool.
To sample the effect of G protein stimulation on a larger population of cells, we repeated the experiments with a suspension of cells in a spectrofluorometer. In Fig. 5A (closed circles), we show that addition of acetylcholine, but not the unrelated agonist insulin, causes a significant drop in the BiF-PLC␤ 2 /PLC␦ 1 fluorescence within the first 5 min. This response, which is similar to that seen for the attached cells, indicates a disruption of the BiF-PLC␤ 2 ⅐PLC␦ 1 complex in response to G protein activation. The extent of this drop varied from 18 to 47% of the initial fluorescence, which appeared to be related to differences in the expression level of the proteins (Fig. 5A). All samples recovered to near initial values at longer times.
Acetylcholine will release activated G␣ q and G␣ i subunits as well as G␤␥ subunits. To test which was involved in the changes of BiFC-PLC␦ 1 /PLC␤ 2 fluorescence, we pre-treated the cells with PTX to inhibit activation of G␣ i , because previous studies have shown HEK293 cells to be PTX-sensitive (41,42). PTX treatment, which prevents G␤␥ release from G␣ i , was BiFC-PLC␤ 2 /PLC␦ 1 fluorescence from transfected HEK293 cell membranes upon the addition of purified, unlabeled PLC␤ 2 or G␤␥ subunits, where the latter protein was dialyzed in a nondetergent buffer prior to use. The fraction of protein dissociated is presented. These values were calculated from the decrease in fluorescence relative to the initial value, which was normalized to 1.0 after subtraction of control samples containing equal amounts of membranes transfected with only the BiFC-PLC␤ 2 plasmid to give the fraction associated. These values were subtracted from 1.0 to give the fraction dissociated. Standard error is shown where n ϭ 6.
found to eliminate completely the decrease in fluorescence (Fig.  5A, open squares). These results imply that G␤␥ subunits released from G␣ i are required for the decrease in BiFC-PLC␦ 1 / PLC␤ 2 fluorescence.
To assess directly the ability of G␤␥ subunits to disrupt BiFC-PLC␤ 2 ⅐PLC␦ 1 complexes, membranes were prepared from HEK293 cells transfected with the BiFC-PLC␤ 2 /PLC␦ 1 plasmids. We then measured the reduction in YFP fluorescence as G␤␥ subunits or unlabeled PLC␤ 2 is added in vitro. As shown in Fig. 5B, we find that we can completely eliminate BiFC-PLC␤ 2 /PLC␦ 1 fluorescence by the addition of unlabeled PLC␤ 2 or G␤␥ subunits.
The Effect of Increased PLC␤ 2 and/or PLC␦ 1 on Intracellular Ca 2ϩ and Inositol Phosphate Levels-The physical association between the two proteins described above suggested that the activity of PLC␦ 1 would be suppressed by cotransfection with PLC␤ 2 . This prediction was first tested by monitoring cellular Ca 2ϩ levels in living HEK293 cells transfected with PLC␤ 2 or PLC␦ 1 expression plasmids or both by using the calcium indicator dye Fura-2. The results are presented in Fig. 6. Transfection of these cells with PLC␤ 2 does not significantly perturb either the basal or stimulated levels of Ca 2ϩ . In contrast, transfection with PLC␦ 1 results in a significant increase in both the basal and stimulated levels of Ca 2ϩ . However, cotransfecting PLC␦ 1 with PLC␤ 2 reduces these Ca 2ϩ levels to values much closer to those of control untransfected cells. These data support the idea that PLC␤ 2 binds to PLC␦ 1 and inhibits its PI(4,5)P 2 hydrolyzing activity.
In Fig. 6 we also present traces showing the recovery of elevated Ca 2ϩ after stimulation from a single series of studies. Pooling 9 -12 sets performed in triplicate allowed us to analyze the time dependence of the decrease in elevated cellular Ca 2ϩ following stimulation. We find that the recoveries of all but PLC␦ 1 -transfected cells could be fit to a single exponential decay to give 0.0192 Ϯ 0.003 nM [Ca 2ϩ ]/s (control cells), 0.0168 Ϯ 0.007 [Ca 2ϩ ]/s (PLC␤ 2 -transfected cells), and 0.0177 Ϯ 0.004 [Ca 2ϩ ]/s (PLC␤ 2 /PLC␦ 1 -transfected cells). However, PLC␦ 1 -transfected cells either did not recover (5 of 12 studies) or recovered too slowly to be fit to an exponential (7 of 12 studies). This persistence in Ca 2ϩ levels due to PLC␦ 1 most likely results from enhancement of PI(4,5)P 2 levels in the cells. It is interesting that coexpression of PLC␤ 2 with PLC␦ 1 reduces this persistent period to control levels. Recovery occurs during the time that the two enzymes are separated (see Fig.  5), suggesting that their coexpression entails feedback mechanisms in addition to or other than PLC␤ 2 regulation.
To verify that changes in cellular Ca 2ϩ shown in Fig. 6 reflected inositol phosphate generation, we measured changes in the total inositol phosphate production in the basal and stimulated states in cells transfected with PLC␤ 2 , PLC␦ 1 , or both. The results, presented in Fig. 7A, show that cells transfected with PLC␤ 2 display a small increase in inositol phosphate levels, whereas transfection with PLC␦ 1 alone resulted in much higher inositol phosphate production. Transfection with both enzymes returns the inositol phosphate levels to those seen for PLC␤ 2 alone. This same suppression of PLC␦ 1 - In the unstimulated state, PLC␤ 2 complexes with PLC␦ 1 on the plasma membrane surface to inhibit PLC␦ 1 catalytic activity and keep the levels of calcium ions in the cell low. Upon agonist binding to a G protein-coupled receptor, G␤␥ subunits are released from G␣ i subunits displacing PLC␦ 1 from PLC␤ 2 . The result is a rapid activation of both phospholipases, which in turn causes a synergistic release in Ins(1,4,5)P 3 and a sharp increase in intracellular calcium because of release from intracellular stores and the opening of Ca 2ϩ -activated ion channels on the plasma membrane. In the recovery phase, the high cellular Ca 2ϩ allows for a sustained PLC␦ 1 activation until its complete return to PLC␤ 2 and the basal state.
induced increase in inositol phosphate levels could be seen upon stimulation with 5 M carbachol. Thin layer chromatography of the inositol lipids showed that the basal levels of [ 3 H]PI(4,5)P 2 are maintained in cells transfected with PLC␤ 2 , whereas the levels of this lipid are elevated when the cells are transfected with PLC␦ 1 alone (Fig. 7B).
The in vitro results predict a synergistic release of Ca 2ϩ upon agonist stimulation that was not clearly seen in HEK293 cells. However, signaling in these cells may be limited at the level of receptor. To test this idea, we cotransfected G␤ 1 ␥ 2 with BiFC-PLC␦ 1 /PLC␤ 2 in HEK293 cells. We found that overexpression of G␤ 1 ␥ 2 reduced BiFC-PLC␦ 1 /PLC␤ 2 fluorescence from an initial normalized value of 1.00 Ϯ 0.35 to 0.30 Ϯ 0.01 (n ϭ 3) relative to control cells that were transfected with empty vector. Thus, the presence of excess G␤␥ results in a 70% reduction in BiFC-PLC␤ 2 ⅐PLC␦ 1 complexes.
We further tested whether enhanced Ca 2ϩ release by dissociation of PLC␤ 2 ⅐PLC␦ 1 complexes was limited at the receptor/G protein level in another series of studies utilizing rat vascular smooth muscle A10 cells. Western blot analysis has suggested that these cells have high endogenous levels of PLC␤ 2 and only trace amounts of PLC␦ 1 (37). We transfected these cells with small amounts of PLC␦ 1 and observed a 30% increase in Ca 2ϩ release (p ϭ 0.004) over mock-transfected cells with stimulation using 10 mM carbachol. However, transfection with higher amounts of PLC␦ 1 resulted in increased basal Ca 2ϩ and a corresponding rise in stimulated levels indicating that there is not enough available PLC␤ 2 to control the robust PLC␦ 1 activity. Therefore, it is possible to observe a synergistic release of Ca 2ϩ with PLC␦ 1 over a narrow range of expression. DISCUSSION The studies presented here demonstrate that PLC␤ 2 controls PLC␦ 1 activity. This control involves inhibition of PLC␦ 1 through its physical association with PLC␤ 2 at physiological concentrations of Ca 2ϩ . The association between these PLCs can be recapitulated in living cells as can the inhibition of increased cellular Ca 2ϩ and inositol phosphate levels due to overexpression of PLC␦ 1 . These findings connect the regulation of PLC␦ 1 activity with G protein-coupled receptor signaling.
By using purified proteins on model membranes, we find that PLC␤ 2 and PLC␦ 1 associate and that this association is not simply because of nonspecific aggregation of the proteins. These proteins associate with moderate affinity in solution and ϳ20fold more strongly when bound to lipid bilayers. Binding of proteins to a membrane surface confines them to a more restricted area and increases their effective concentration. We have previously treated the difference in dissociation constants between proteins in solution and bound to membranes, and we found the decrease in affinity between PLC␤ 2 and PLC␦ 1 to be appropriate (23). PLC␤ 2 binds strongly and fairly nonspecifically to membranes (26) in contrast to PLC␦ 1 , which only binds strongly to membranes that are highly negatively charged or if PI(4,5)P 2 is present (25). The strong affinity for the two proteins on membrane surfaces allows PLC␤ 2 to laterally associate with PLC␦ 1 to suppress its activity under nonstimulated conditions. It is noteworthy that the strength of PLC␤ 2 -PLC␦ 1 association on membrane surfaces is only ϳ5-fold weaker than that of PLC␤ 2 -G␤␥ (28) and may thus be competitive under many physiological conditions. The mechanism of inhibition, however, is not clear. Although PLC␦ 1 binds strongly and specifically to membranes containing PI(4,5)P 2 , no such specificity is observed for PLC␤ 2 , and so inhibition is not because of competition of the enzyme for substrate or membrane binding. Because the PH domains of both enzymes play a role in catalytic activity, and because the PH domains of both enzymes bind G␤␥ subunits, this region is a reasonable candidate for inhibi-tion (27,33). We note that we have preliminary evidence suggesting that the PH domain of PLC␦ 1 binds to PLC␤ 2 with an affinity in range of the whole enzyme. 2 We speculate that the PLC␤ 2 -PLC␦ 1 molecular association may be similar to the inhibitory association between PLC␦ 4 and the inactive PLC␦ 4 -Alt3, which also appears to involve the PH domain. Molecular models that dock the crystal structures of the PH and catalytic domains of PLC␦ 1 suggest a very close proximity of the PH domain to the catalytic site (27), leading to the idea that altered domain interactions could inhibit the conformational changes needed for effective catalysis. We note that while this study focused on PLC␤ 2 which binds strongly to G␤␥ subunits (28), it is likely that similar results would be obtained with the widely distributed PLC␤ 3 isozymes, which are also regulated by G␤␥ subunits (43), and we find the PLC␦ 1 association for PLC␤ 3 to be similar to PLC␤ 2 . 3 Cellular association between the two former proteins is currently under investigation. PLC␦ 1 is activated by increases in cellular Ca 2ϩ and inhibited by a high local concentration of Ins(1,4,5)P 3 that prevents substrate binding (29). Regulation of PLC␦ 1 by RhoGAP and transglutaminase (9,44) is through lowering the Ca 2ϩ concentration that induces PLC␦ 1 activation. We find here that PLC␤ 2 inhibition of PLC␦ 1 occurs even at high physiological levels of Ca 2ϩ , suggesting the mechanism of regulation is quite different from RhoGAP and transglutaminase. By maintaining PLC␦ 1 in an inactive state, PLC␤ 2 confers agonist-specific control on this Ca 2ϩ -stimulated enzyme. The inability of Ca 2ϩ to disrupt the PLC␤ 2 ⅐PLC␦ 1 complex suggests that only PLC␦ 1 populations not associated with PLC␤ 2 will be activated upon an increased in Ca 2ϩ through pathways other than those involving heterotrimeric G proteins. Of course, this regulation mechanism of PLC␦ 1 will only occur in cell lines that express both proteins. Although there are many reports identifying coexpression of these proteins by Western blot analysis in cultured cells lines, such as A10 and PC12 (37,45,46), we note that almost all of these studies employed the same commercial polyclonal PLC␤ 2 antibody, which may not have high specificity. Weak anti-PLC␦ 1 -PLC␦ 1 interaction precluded identifying these complexes by coimmunoprecipitation, and so we verified expression of mRNA of both PLC␤ 2 and PLC␦ 1 in heart and brain tissue and in A10 and PC12 cells by reverse transcription-PCR (Supplemental Material). These results suggest that some cell lines express both proteins. It is also noteworthy that preliminary studies show that more widely expressed PLC␤ 3 , which is also simulated by G␤␥ subunits, similarly inhibits PLC␦ 1 in vitro. 3 Most interestingly, we find that in living cells the PLC␤ 2 ⅐PLC␦ 1 complex is mainly localized on the plasma membrane, with a significant population diffusely distributed throughout the cytosol. Our studies suggest that G protein stimulation disrupts the plasma membrane-localized complexes as compared with the cytosolic. Whereas association between internal PLC␤ 2 -PLC␦ 1 is expected to suppress phosphatidylinositol lipid hydrolysis at these internal sites, it is unclear what causes dissociation of the cytoplasmic PLC␤ 2 ⅐PLC␦ 1 complex, although it is unlikely to be the rise in cytoplasmic free calcium concentration, because complex stability seems unperturbed by this divalent cation. The simplest mechanism could involve the generation of excess free G␤␥ subunits at the plasma membrane that would shift the dynamic equilibrium in favor of PLC␤ 2 plasma membrane association, which is supported by recent work (47). From this latter work, it is also possible that cytosolic PLC␤ 2 is regulated by RhoGTPases, which may also inhibit its association with cytoplasmic PLC␦ 1 . PLC␤ 2 -PLC␦ 1 association in living cells was monitored by BiFC. This technique has recently been established to detect in vivo protein-protein associations (38). Protein-protein association, which is measured as increased fluorescence, is technically easier to assess than fluorescence resonance energy transfer. Because the formation of the BiFC-YFP chromophore may stabilize the protein-protein association being viewed, it was important to show that complex formation was not driven by the natural tendency of BiFC-YFP fragments to associate. We found that BiFC-Fos⅐PLC␤ 2 complexes did not form under conditions where a significant population of Fos is cytosolic, suggesting that the BiFC-YFP tags do not drive the association of noninteracting protein partners. Moreover, we show that the fluorescence from transfected HEK293 cells due to BiFC-PLC␤ 2 /PLC␦ 1 could be extinguished by the addition of excess purified PLC␤ 2 or G␤␥ subunits. These two independent studies indicate that BiFC is an appropriate method for following PLC␤ 2 -PLC␦ 1 association in cells.
The agonists used in these studies, acetylcholine and its stable analog carbachol, are coupled to both G␣ q and G␣ i heterotrimers, which can both release G␤␥ subunits (48). G␣ q subunits directly bind to and activate PLC␤ 2 , and so it was possible that the disruption of the BiFC-PLC␤ 2 /PLC␦ 1 seen in cells could be caused by either G␣ q or G␤␥ subunits. We treated the cells with pertussis toxin that will modify G␣ i subunits, preventing their interaction with receptor and thus release of G␤␥ subunits (49). Treatment of pertussis toxin eliminated the decrease in BiFC-PLC␤ 2 /PLC␦ 1 fluorescence upon acetylcholine stimulation, suggesting that under these conditions disruption of the complex is due to release of G␤␥ from G␣ i/o subunits. The time scale observed for the decrease and the subsequent recovery of complex fluorescence matches the time of stimulation, sequestration, and recycling of acetylcholine receptors, which is consistent with the idea that G␤␥ disrupts PLC␤ 2 -PLC␦ 1 association (50).
Because the specific activity of PLC␦ 1 for PI(4,5)P 2 is stronger than the activity of PLC␤ 2 (51), the inhibition of PLC␦ 1 by PLC␤ 2 may be significant in terms of Ins(1,4,5)P 3 production and subsequent Ca 2ϩ release. The disruption of PLC␤ 2 /PLC␦ 1 association by G␤␥ subunits predicts that cellular stimulation by a G protein agonist will allow for a synergistic PLC response through G␤␥ activation of PLC␤ 2 and relief of PLC␦ 1 inhibition. The results in A10 cells show that this is the case at low PLC␦ 1 transfection levels. It is apparent that the cellular mechanisms that control the stoichiometry and accessibility of these enzymes are under tight control because PLC␤ 2 suppression of PLC␦ 1 activity is seen at higher PLC␦ 1 expression levels. Alternatively, in HEK293 cells synergistic Ca 2ϩ and inositol phosphate release was not observed. A simple explanation is that under the conditions of simultaneous overexpression of PLC␤ 2 and PLC␦ 1 , control of second messengers is at the receptor level, which limits the amount of G␤␥ subunit activation of PLC␤ 2 and in turn release of PLC␦ 1 . This idea is supported by results showing that overexpression of G␤ 1 ␥ 7 disrupts BiFC-PLC␤ 2 ⅐PLC␦ 1 complexes in HEK293 cells.
Taken together, our results suggest the model shown in Fig.  8 whereby PLC␤ 2 serves as a cellular regulator of PLC␦ 1 in which its control is linked to activation of heterotrimeric G proteins, specifically G␣ i/o -G␤␥. This link could also involve G␤␥-dependent amplification of the PLC␤ 2 -generated calcium signal. Release of PLC␦ 1 from PLC␤ 2 would also increase the free fraction of PLC␦ 1 available for nuclear uptake, where it may modulate gene expression.