Phospholipase c β2 association with phospholipid interfaces assessed by fluorescence resonance energy transfer

Phospholipase C β2 (PLC β2) is activated by G protein βγ subunits and calcium. The enzyme is soluble and its substrate, phosphatidylinositol 4,5-bisphosphate (PIP2), is present in phospholipid membranes. A potential mechanism for regulation of this enzyme is through influencing the equilibrium association of the enzyme with membrane surfaces. In this paper we describe a fluorescence resonance energy transfer (FRET) method for measuring the association of PLC β2 with phospholipid bilayers. The method allows equilibrium measurements to be made under a variety of conditions, including those that support enzymatic activity and ability to be regulated by G proteins. Using this method it was found that PLC β2 bound to vesicles containing anionic lipids and demonstrated a selective and unique interaction with PIP2-containing vesicles. The FRET data were corroborated with a centrifugation based method for estimating the affinity of PLC β2 for vesicles. Apparently different modes of association of PLC β2 with vesicles of different composition can be distinguished based on alterations in resonance energy transfer efficiency. Association of PLC β2 with PIP2 vesicles requires an intact lipid bilayer, is blocked by neomycin, and is not affected by D-myo-inositol 1,4,5-trisphosphate (D-IP3). G protein βγ subunits do not alter the affinity of PLC β2 for lipid bilayers and at the PIP2 concentrations used to measure βγ-dependent stimulation of PLC activity, the majority of the PLC β2 is already associated with the vesicle surface. Furthermore, under conditions where βγ subunits strongly activate PLC activity, the extent of association with vesicles is unaffected by βγ subunits or calcium. These results indicate that activation of PLC β2 by G protein βγ subunits or Ca2+ in vitro does not involve translocation to the vesicle surface.

Phospholipase C ␤2 (PLC ␤2) is activated by G protein ␤␥ subunits and calcium. The enzyme is soluble and its substrate, phosphatidylinositol 4,5-bisphosphate (PIP 2 ), is present in phospholipid membranes. A potential mechanism for regulation of this enzyme is through influencing the equilibrium association of the enzyme with membrane surfaces. In this paper we describe a fluorescence resonance energy transfer (FRET) method for measuring the association of PLC ␤2 with phospholipid bilayers. The method allows equilibrium measurements to be made under a variety of conditions, including those that support enzymatic activity and ability to be regulated by G proteins. Using this method it was found that PLC ␤2 bound to vesicles containing anionic lipids and demonstrated a selective and unique interaction with PIP 2 -containing vesicles. The FRET data were corroborated with a centrifugation based method for estimating the affinity of PLC ␤2 for vesicles. Apparently different modes of association of PLC ␤2 with vesicles of different composition can be distinguished based on alterations in resonance energy transfer efficiency. Association of PLC ␤2 with PIP 2 vesicles requires an intact lipid bilayer, is blocked by neomycin, and is not affected by D-myo-inositol 1,4,5-trisphosphate (D-IP 3 ). G protein ␤␥ subunits do not alter the affinity of PLC ␤2 for lipid bilayers and at the PIP 2 concentrations used to measure ␤␥-dependent stimulation of PLC activity, the majority of the PLC ␤2 is already associated with the vesicle surface. Furthermore, under conditions where ␤␥ subunits strongly activate PLC activity, the extent of association with vesicles is unaffected by ␤␥ subunits or calcium. These results indicate that activation of PLC ␤2 by G protein ␤␥ subunits or Ca 2؉ in vitro does not involve translocation to the vesicle surface.
Enzymes of the phospholipase C ␤ (PLC ␤) 1 class hydrolyze phosphatidylinositol 4,5-bisphosphate (PIP 2 ) in response to ac-tivation of G protein-coupled receptors (1). Many receptor-coupled responses that are insensitive to treatment with the toxin from Bordatella pertussis are mediated by ␣ subunits of the G q family members (2)(3)(4), while responses inhibited by this toxin are thought to be via the ␤␥ subunits from G i proteins. Both of these G proteins can activate PLC ␤ through direct proteinprotein interactions as has been demonstrated in reconstitution assays with purified proteins (1,(5)(6)(7)(8) G protein ␤␥ subunits have been implicated as primary signal transducers in wide variety of signal transduction pathways. ␤␥ subunits have been shown to directly activate some isoforms of adenylyl cyclase and inhibit others (9), activate ␤-adrenergic receptor kinase (10), muscarinic receptor kinase (11), cardiac K ϩ channels (12), and phosphatidylinositol 3-kinase (13,14). ␤␥ subunits mediate cell cycle arrest via the mating pathway in the yeast Saccharomyces cerevisiae (15), but the effector involved has not been identified. Other work suggests that ␤␥ subunits can lead to activation of Ras and mitogen-activated protein kinase (MAP kinase) in COS-7 (16 -18) cells, implicating ␤␥ subunits in the regulation of cell growth. The list of potential effectors continues to grow, but the molecular mechanisms and interactions involved have yet to be understood.
Phospholipase C ␤ is a soluble enzyme whose subcellular localization has not been clearly elucidated. A large proportion of the enzyme is found associated with both the soluble and particulate fractions from cell and tissue lysates (8,19,20). In turkey erythrocytes 98% of the PLC ␤ is in the cytosol, while the PLC ␤ found in the particulate fraction is associated with the cytoskeleton, not the plasma membrane (21). Immunocytochemical studies demonstrate a large proportion of the PLC ␤ associated with the nucleus (22) but have failed to demonstrate a plasma membrane or cytosolic location. Thus, there is no conclusive direct evidence demonstrating PLC ␤ association with the plasma membrane. A potential mechanism for activation of the PLC ␤2 by ␤␥ subunits is through translocation of the PLC from the cytosol by binding to plasma membrane bound ␤␥ subunits. A translocation mechanism has been proposed for activation of cytosolic ␤-adrenergic receptor kinase by G protein ␤␥ subunits (10). Translocation is also one suggested mechanism for activation of cytosolic PLC ␥ by the epidermal growth factor receptor (23).
Reconstitution of PLC ␤2 and ␤3 with ␤␥ subunits in vitro in the presence of phospholipid vesicles containing the substrate results in activation of the phospholipases (8,24). A potential mechanism for this activation is through ␤␥ subunit-mediated association of PLC ␤2 with the vesicle surface. We have developed a method based on fluorescence resonance energy transfer for measurement of the association of fluorescein-labeled PLC ␤2 with vesicles. This method allows us to monitor association of PLC ␤2 with vesicles under conditions where ␤␥ stimulates PLC ␤2 activity. In this paper we have examined the properties of lipid vesicles and PLC ␤2 that contribute to association with lipid bilayer surfaces and examine the influence of G protein ␤␥ subunits on this association. The results presented here could help in understanding the role of ␤␥ subunit-dependent translocation in mediating PLC ␤ activation in vivo.

Methods
Plasmid Construction and Cloning of Recombinant Baculoviruses-To obtain the large quantities of PLC that are required for these experiments, recombinant PLC ␤2 was expressed using a baculovirus expression system. A 3.7-kilobase fragment extending from the Eco47III site at bp 265 to the XbaI site at bp 3992 in the PLC ␤2 cDNA was ligated (at the 5Ј-end of the cDNA fragment) to oligonucleotides encoding an EcoRI site (at the 5Ј-end) followed by coding sequence for an initiator methionine, an alanine, a histidine tag (6 histidines), an alanine, and the first 20 amino acids of PLC ␤2. These oligonucleotides ended in coding sequence for an Eco47III site. The PLC ␤2 cDNA fragment and the oligonucleotides were ligated into pBluescript KS at EcoRI and XbaI sites as above. The modified cDNA was excised from pBluescript KS with EcoRI and NotI, and this fragment was ligated into EcoRI and NotI sites in pVL1393.
Recombinant baculoviruses were generated by cotransfection (Lipofectin, Life Technologies, Inc.) of fall armyworm cells (Sf9 cells) with pVL1393 containing modified PLC ␤2 cDNA and with AcRP23-lacZ baculovirus DNA that had been linearized with Bsu36I. Positive clones were isolated by plaque assay and were identified by their ability to direct the expression of PLC ␤2 as determined by immunoblotting and by PLC activity assays (described below).
Sf9 Cell Culture and Purification of Phospholipase C-Sf9 cells were grown at 27°C in IPL-41 medium containing 10% fetal bovine serum, 0.1% pluronic acid, and 50 g/ml gentamycin. For large scale cultures (1 liter and above) the cells were switched into medium containing 1% fetal bovine serum and 1% lipid concentrate (Life Technologies, Inc.). Baculoviruses directing expression of recombinant His 6 PLC ␤2 were used to infect 1L of Sf9 cells at a density of 1.5 ϫ 10 6 cells/ml. Cells were incubated at 27°C and shaken at 125 rpm for 48 h. Cells were collected by centrifugation at 2500 rpm in a Beckman JA10 rotor for 20 min, suspended in 50 ml of phosphate-buffered saline, transferred to a 50-ml conical tube, and centrifuged again at 1000 ϫ g. The supernatant was removed, and the pellet was frozen in liquid N 2 and stored at Ϫ70°C until further processing could be performed.
Cells were lysed by thawing the frozen pellet in 25 ml of lysis buffer (50 mM NaHepes, pH 7.4, 0.1 mM EGTA, 0.1 mM EDTA, 0.1 mM DTT, 100 mM NaCl, and protease inhibitors (133 M phenylmethylsulfonyl fluoride, 21 g/ml N ␣ -p-tosyl-Llysine chloromethyl ketone, 21 g/ml tosylphenylalanyl chloromethyl ketone, 0.5 g/ml aprotonin, 0.2 g/ml leupeptin, 1 g/ml pepstatin A, 42 g/ml tosylarginine methyl ester, and 10 g/ml soybean trypsin inhibitor (SBTI)). The suspended cells were repeatedly (four times) frozen in liquid N 2 and thawed in a 37°C water bath. The volume of the lysate was adjusted to 45 ml, and 15 ml of 4 M NaCl was added. The resulting extract was centrifuged at 40,000 rpm in a Ti60 rotor for 45 min. The supernatant was diluted by addition of 240 ml of dilution buffer (10 mM NaHepes, pH 8.0, 10 mM ␤-mercaptoethanol, 0.1 mM EGTA, 0.1 mM EDTA, 0.5% polyoxyethylene-10lauryl ether (C 12 E 10 ), and protease inhibitors). The diluted extract was centrifuged at 40,000 rpm in a Ti45 rotor for 45 min. Deoxyribonuclease I (300 g) was added to the supernatant, which was loaded onto a 4-ml column of nickel-nitrilotriacetic acid resin (Qiagen) that had been equilibrated with dilution buffer. The column was washed with 80 ml of wash buffer (10 mM NaHepes, pH 8.0, 0.1 mM EDTA, 0.1 mM EGTA, 800 mM NaCl, 0.5% C 12 E 10 , 15 mM imidazole, plus protease inhibitors). The column was then washed with wash buffer lacking C 12 E 10 and SBTI and with 100 mM instead of 800 mM NaCl. Protein was eluted with six successive 4-ml applications of 10 mM NaHepes, pH 8.0, 0.1 mM EDTA, 0.1 mM EGTA, 50 mM NaCl, 125 mM imidazole, and protease inhibitors without SBTI. Fractions were analyzed by SDS-polyacrylamide gel electrophoresis followed by staining with Coomassie Brilliant Blue.
To concentrate the PLC ␤2 and to exchange the protein into a buffer compatible with FITC-labeling, the protein was bound to a 2-ml column of heparin-Sepharose CL-6B that had been equilibrated with 10 ml of equilibration buffer (20 mM NaHepes, pH 8.0, 0.1 mM EDTA, 0.1 mM EGTA, 100 mM NaCl, and protease inhibitors without SBTI). Purified PLC ␤2 from the nickel-nitrilotriacetic acid column was applied to the column followed by a wash with 10 ml of equilibration buffer. The protein was then eluted with successive 2-ml applications of 20 mM NaHepes, pH 8.0, 0.1 mM EDTA, 0.1 mM EGTA, 800 mM NaCl, and protease inhibitors without SBTI. Protein concentrations were determined by an Amido Black protein assay (26).
Expression and Purification of G Protein ␤␥ Subunits-Baculoviruses encoding ␤ 1 , ␥ 2 , and His 6 -tagged ␣ i1 were obtained from Alfred Gilman's laboratory. 1 liter of Sf9 cells at 1 ϫ 10 6 cells/ml were simultaneously infected with the three baculovirus constructs, and the ␤␥ subunits were purified according to the published procedure (27). Briefly, cells were lysed by repeated freeze thawing (four times) and the membranes recovered by centrifugation at 100,000 ϫ g for 1 h. The membranes were suspended and extracted with 1% cholate. Insoluble material was removed by centrifugation, and the detergent extract was applied to a 4-ml column of nickel-nitrilotriacetic acid-agarose. Under these conditions the ␣␤␥ heterotrimer bound to the resin via the His 6 tag on the ␣ subunit. ␤␥ subunits were selectively eluted by activation of the G protein with AlF 4 . The final preparation was concentrated and gel filtered into ␤␥ vehicle (50 mM NaHepes, pH 8.0, 1 mM EDTA, 1 mM DTT, 100 mM NaCl, 1% cholate). The yield was approximately 1 mg of ␤␥/liter of culture.
Fluorescent Labeling of Phospholipase C-0.5-1.5 mg of PLC (1 ml) in 20 mM NaHepes, pH 8.0, 0.1 mM EDTA, 0.1 mM EGTA, 800 mM NaCl, and protease inhibitors without SBTI was reacted on ice for 15 min with 20 l of 50 mM FITC freshly dissolved in dimethylformamide. The reaction was quenched by addition of 200 l of 1 M Tris-Cl, pH 8.0. The solution was left an additional 15 min on ice before proceeding. Fluorescently labeled PLC ␤2 was separated from free FITC by sequential chromatography on heparin-Sepharose and hydroxyapatite. A 1-ml column of heparin-Sepharose CL-6B (Pharmacia Biotech Inc.) was equilibrated with 6 ml of buffer A (50 mM NaHepes, pH 8.0, 200 mM NaCl, 1 mM DTT, 0.1 mM EDTA, 0.1 mM EGTA, and protease inhibitors). The FITC-PLC-Tris sample was diluted to 5 ml in a salt-free NaHepes buffer (50 mM NaHepes, pH 8.0, 0.1 mM EDTA, 0.1 mM EGTA plus protease inhibitors) and loaded on the heparin column. The column was washed with 50 ml of equilibration buffer, and labeled protein was eluted with five successive 1-ml volumes of a high salt buffer (50 mM NaHepes, pH 8.0, 800 mM NaCl, 1 mM DTT, 0.1 mM EDTA, 0.1 mM EGTA plus protease inhibitors). All washes and elutions were collected individually and kept on ice. At this stage the preparations contained between 20 and 40% of the total fluorescence as free FITC (see below for methods).
To remove the remaining free dye, fractions containing the highest protein concentration were processed by chromatography on hydroxyapatite. A 2-ml column of Macro-Prep ceramic hydroxyapatite (Bio-Rad) was equilibrated with (50 mM NaHepes, pH 8.0, 100 mM NaCl, 1 mM DTT, 0.1 mM EDTA, 0.1 mM EGTA, 100 mM potassium phosphate, pH 8.0, plus protease inhibitors). Pooled, FITC-labeled protein from the heparin column was loaded onto the column followed by a wash with equilibration buffer until the absorbance at 280 nm returned to base line. The protein was eluted with a linear gradient in equilibration buffer from 100 mM to 700 mM potassium phosphate in 20 ml. 1-ml fractions were collected and assayed for protein. Fractions containing the highest concentrations of protein were pooled and concentrated if necessary. After this step less than 1% of the total fluorescence was due to free FITC. The yield of protein starting with 1 mg of purified PLC ␤2 was generally about 10 -25%. Labeled protein was quantitated by Amido Black protein assay, and fractions having the highest protein concentration were stored at Ϫ80°C until later use.
Characterization of Labeled Protein-The stoichiometry of protein labeling was assessed by measurement of the absorbance of the modified protein at 495 nm to determine the number of moles of dye present based an extinction coefficient for FITC of 82,000 cm Ϫ1 mol Ϫ1 (Molecular Probes catalogue). This was divided by the molar amount of labeled PLC as determined by the Amido Black protein assay.
Assessment of relative amounts of bound and free dye was done by acetone precipitation and SDS-polyacrylamide gel electrophoresis (28,29). For acetone precipitation, 500 l of acetone was added to a 20-l FITC-labeled PLC ␤2 (F-PLC ␤2) sample (6 -20 g of protein), incubated on ice 20 min to facilitate protein precipitation, and centrifuged at 13,000 ϫ g for 15 min. The acetone supernatant containing free FITC was transferred to a new tube and dried under a gentle nitrogen stream. The dried acetone supernatant was resuspended in 20 l of elution buffer. The fluorescence of this sample was determined by dilution into 900 l of FRET buffer (50 mM NaHepes, pH 8.0, 67 mM KCl, 17 mM NaCl, 0.83 mM MgCl 2 , 3 mM EGTA, 0.17 mM EDTA, 1 mM DTT, 1 mg/ml BSA). The fluorescence was measured with 495 nm excitation and 520 emission wavelengths on a Perkin-Elmer spectrofluorimeter. The fluorescence of an equal volume of unprecipitated sample was also determined and the amount of free FITC based on the percentage of original fluorescence remaining in the dried acetone extract after acetone precipitation.
Labeled protein samples were also run on a 9% polyacrylamide SDS gel to separate bound from free dye. Free dye was visible in the dye front under ultraviolet light in samples that had not been subjected to hydroxyapatite chromatography. After the hydroxyapatite, no dye was visible in the ion front of the gel. Protein-associated dye migrated at the position of unlabeled PLC ␤2 (data not shown).
Determination of Lipid Vesicle Binding by Fluorescence Resonance Energy Transfer (FRET)-Experiments were carried out in 0.5-1 ml of FRET buffer (50 mM NaHepes, pH 8.0, 67 mM KCl, 17 mM NaCl, 0.83 mM MgCl 2 , 3 mM EGTA, 0.17 mM EDTA, 1 mM DTT, 1 mg/ml BSA). F-PLC ␤2 was added in 2-5 l and the fluorescence determined in a Perkin-Elmer LS-5B spectrofluorimeter at 30°C with an excitation wavelength of 495 nm and emission at 520 nm. Fluorescence quenching was found to be independent of protein concentration between 0.2 and 2 g of F-PLC ␤2. Lipid vesicles were prepared by drying the appropriate amount of lipid in chloroform under a stream of nitrogen. Sonication buffer (50 mM NaHepes, pH 8.0, 3 mM EGTA, 80 mM KCl, 1 mM DTT) was added, and the lipids were sonicated for 5 min in a bath sonicator. Vesicles were prepared with 500 M PIP 2 in various ratios with other lipids as described in the figure legends. In some experiments other anionic lipids were substituted for PIP 2 . In most of the experiments PE was used in a 4:1 ratio with PIP 2 . Whatever the lipid composition, PE-rhodamine was included at 1% of the total lipid. To measure association of F-PLC ␤2 with vesicles, rhodamine-containing lipid vesicles were added at the appropriate concentration after determination of F-PLC ␤2 fluorescence. Nonspecific fluorescence quenching was determined by addition of 0.1% C 12 E 10 prior to addition of lipids.
Calculations for energy transfer were done by division of fluorescence obtained after addition of lipid (F) by original fluorescence due to F-PLC ␤2 alone (F o ). The data are expressed as a percentage of the original fluorescence by the formula (100 Ϫ (F/F o ϫ 100). Where appropriate, data were normalized by subtraction of the amount of quenching in the presence of C 12 E 10 from the values obtained in the absence of detergent (as explained below).
Assay of F-PLC ␤2 Vesicle Binding by Centrifugation-F-PLC ␤2 was incubated with vesicles of various lipid composition as described in the figure legends. The incubation was performed in FRET buffer in a total volume of 150 l for 5 min at 25°C. 100 l was transferred to centrifuge tubes and centrifuged in a Beckman TL100.3 rotor at 80,000 ϫ g for 30 min at 25°C. The fluorescence of rhodamine in the vesicles was monitored before and after centrifugation at an excitation wavelength of 570 nm and emission wavelength of 590 nm to determine the efficiency of vesicle pelleting by centrifugation at all vesicle concentrations. In all of the experiments presented, 90% or greater of the vesicles were removed from the supernatant by centrifugation. The extent of F-PLC ␤2 binding was determined by measuring the amount of fluorescein fluorescence (measured at an excitation wavelength of 495 and emission wavelength of 520) in the supernatant of samples containing vesicles and comparing with samples centrifuged in the absence of vesicles (taken as 0% removed). In experiments where ␤␥ subunits were added in a volume of 2 l, an equal volume of ␤␥ vehicle was added to controls without ␤␥ subunits. Fluorescence was measured by adding 25 l of supernatant to 1 ml of NaHepes, pH 8.0, 0.1% C 12 E 10 .

RESULTS
Labeling of PLC ␤2 with FITC-Reaction of purified recombinant PLC ␤2 with FITC resulted in covalent attachment of the dye at a stoichiometric ratio of 0.8 -1.2 mol of dye/mol of protein. The residual free dye after labeling was 1% or less of the total fluorescence. Analysis by SDS-polyacrylamide gel electrophoresis indicated that a single protein was labeled at ϳ140 kDa, corresponding to the size of the intact PLC ␤2 protein.
The activity of the F-PLC ␤2 was similar to the activity of the unlabeled protein, and both the modified and unmodified proteins were activated by addition of G protein ␤␥ subunits (Fig.  1). This activation was blocked by the addition of recombinant myristoylated ␣ i1 (data not shown). The dependence of activity on ␤␥ concentration and the specific activities in the presence of ␤␥ subunits for the labeled and unlabeled proteins were very similar. This indicated that covalent labeling of PLC ␤2 with FITC did not significantly alter the functional properties of the enzyme.
Binding of PLC ␤2 to Vesicles as Assessed by FRET-The results in Fig. 2A demonstrate that there was increased fluorescence quenching as vesicles containing PE, PIP 2 , and PErho (4:1:0.05) are added, indicating association of F-PLC ␤2 with the vesicles. The association of F-PLC ␤2 with the vesicles, as measured by fluorescence quenching, increased as vesicle concentration increased and approached saturation with maximum FRET efficiencies ranging from 15 to 25%. FRET efficiency is defined as the maximum percent quenching observed at saturation. The majority of the quenching was disrupted by addition of 0.1% C 12 E 10 , a nonionic detergent. This detergent had no effect on the fluorescence of F-PLC ␤2 in the absence of added lipids.
The interaction between F-PLC ␤2and PIP 2 -containing vesicles had an apparent K d of approximately 2 M (Fig. 2B) and approached saturation at concentrations above 10 M PIP 2 as determined by fitting the data to a standard rectangular hyperbola. This indicates that the majority of the F-PLC ␤2 is bound to vesicles at concentrations of PIP 2 higher that 10 M. There was some variability with respect to the measured K d (1-5 M) and maximum FRET efficiency. The source of this variation is unclear, but within an experimental series, results were very reproducible. The data presented are representative of many replicate experiments with multiple preparations of F-PLC ␤2 and lipids. When anionic lipids (PS, PI, or PA) were substituted for PIP 2 , lower levels of FRET were detected at saturation but binding occurred with an apparent affinity similar to that seen for PIP 2 vesicles (Fig. 2C). Very little FRET was observed with PC vesicles, suggesting that negatively charged lipids are required in the bilayer to observe significant vesicle association.
In Table I the results from several treatments are shown that are designed as controls to demonstrate that the fluorescence quenching is due to F-PLC ␤2-phospholipid bilayer interactions and not other trivial causes of fluorescence quenching. First it is shown that quenching is strictly dependent on the presence of rhodamine in the vesicle. This indicates that potential light scattering effects caused by phospholipid vesicles are not influencing the measurements and that binding of F-PLC ␤2 to the lipid vesicles is not resulting in autoquenching of fluorescein fluorescence. Addition of detergent prior to addition of the vesicles completely prevents the quenching, suggesting that integrity of the vesicles is required to observe quenching. In the presence of FITC coupled to Tris, rhodaminecontaining vesicles do not cause fluorescence quenching.
Also measured was sensitized rhodamine fluorescence. Here, F-PLC ␤2 was added to PIP 2 :PE:PE-rho vesicles (10 M PIP 2 ), and fluorescence was monitored with an excitation wavelength of 495 nm and emission at 590 nm. Fluorescence from F-PLC ␤2 was negligible at 590 nm. In the presence of both F-PLC ␤2 and vesicles, the emission at 590 was significantly greater than the sum of the emissions from F-PLC ␤2 and the rhodamine containing vesicles when measured separately. This increase was prevented by addition of 0.1% C 12 E 10 prior to addition of the F-PLC ␤2 (data not shown). At the wavelengths used to measure sensitized rhodamine emission, there is considerable base-line fluorescence contributed from rhodamine that is partially excited at 495 nm. Thus the changes in fluorescence over this base line are difficult to measure accurately. For this reason, all the data collected are as quenching of fluorescein fluorescence at 520 nm.
These data, taken together, indicate that the inhibition of fluorescence emission from F-PLC ␤2 by lipid addition was due to association of F-PLC ␤2 with vesicles resulting in FRET, not to other trivial causes of fluorescence quenching. Fluorescence quenching that did occur in the presence of C 12 E 10 was considered to be nonspecific and probably arose from dilution effects, alterations in fluorescence associated with sample manipulation, and inner filter effects caused by increasing absorbance of the solution as high concentrations of PE-rho were added. All of the data presented (unless otherwise indicated) are specific quenching obtained after subtracting the quenching that occurred in parallel experiments done in the presence of C 12 E 10 . This correction also accounted for dilution effects that occurred as lipids were added to the reaction.
To demonstrate that inclusion of PIP 2 in the vesicles was important for FRET, experiments were performed determine if blocking the PIP 2 head group on the vesicle surface could interfere with FRET. We hypothesized that if PLC ␤2 associates with PIP 2 , pretreatment with a sufficient quantity of unlabeled PLC would compete for the interaction of F-PLC ␤2 with the vesicles. Preincubation of PIP 2 -containing vesicles (1 M PIP 2 ) with 1 M unlabeled PLC ␤2 prevents association of F-PLC ␤2 with those vesicles as measured by FRET (Table I). This treatment inhibits, but does not prevent, quenching at 10 M PIP 2 . We interpret this to indicate that the unlabeled PLC ␤2 is competing for PIP 2 binding sites and thereby preventing association of F-PLC ␤2 with the vesicles. When the concentration of PIP 2 exceeds that of the unlabeled PLC ␤2, more binding sites are available for F-PLC ␤2 association and thus some FRET is observed. An alternative explanation is that the PLC ␤2 is hydrolyzing all of the PIP 2 before the F-PLC ␤2 can associate; however, these experiments were done in the absence of Ca 2ϩ and in the presence of 3 mM EGTA, a condition where PLC ␤2 is inactive. In support of the hypothesis that competition for PIP 2 binding sites prevents FRET is data demonstrating that neomycin, which specifically interacts with PIP 2 , inhibits quenching at 100 M and completely blocks quenching at 1 mM (Table I). Neomycin (1 mM) also blocks F-PLC ␤2 binding to vesicles when measured using centrifuga-  tion to separate bound from unbound F-PLC ␤2 (data not shown). These data all indicate that binding to phospholipid vesicles and association with PIP 2 are required for high efficiency resonance energy transfer.
Since FRET preferentially detects interactions between F-PLC ␤2 and PIP 2 -containing vesicles, we tested whether D-IP 3 could disrupt FRET. It has been reported previously that D-IP 3 blocks binding of PLC ␦ to PIP 2 -containing vesicles by up to 90% at low M concentrations of D-IP 3 (30). This is thought to occur through disruption of a specific interaction between a pleckstrin homology (PH) domain at the N terminus of PLC ␦ and PIP 2 (31). In our experiments, 1 mM D-IP 3 has no effect on the association of PLC ␤2 with PIP 2 -containing vesicles at any PIP 2 concentration tested (10 nM to 30 M data not shown).
Vesicle Association of F-PLC ␤2 Assessed by Centrifugation-To corroborate the data obtained by FRET we examined the association with of F-PLC ␤2 with vesicles by centrifugation to separate F-PLC ␤2 bound to vesicles from free F-PLC ␤2 (Fig. 3). Using this method we estimate that F-PLC ␤2 associates PIP 2 :PE:PE-rho vesicles with an apparent K d of 0.6 M and the binding saturates above 10 M. If PE-rho is omitted from the vesicles the apparent K d is 1.3 M. These data are similar to the data obtained by FRET and indicate that the presence of PE-rhodamine in the vesicles does not significantly affect the apparent affinity of the F-PLC-vesicle interaction. Interestingly there is also interaction between F-PLC and PS containing vesicles with an apparent K d of 1.3 M that is saturable above 10 M PIP 2 . When compared with the results obtained by FRET, these data suggest that the nature of the interaction between F-PLC ␤2 and vesicles containing PIP 2 and vesicles containing PS is different, since the affinities measured by centrifugation and FRET are similar yet there are large differences in FRET efficiency. It could be that the specific interaction with PIP 2 orients the F-PLC such that the fluorescein label is positioned for better resonance energy transfer. Alternatively, the PIP 2 and PS could orient the PErho differently. It should be noted less of the F-PLC ␤2 is removed from the supernatant by PS vesicles at saturation. This is not due to a difference in the ability to centrifuge PS-containing vesicles versus PIP 2 -containing vesicles, because in both cases, greater than 90% of the vesicles were pelleted by centrifugation. The reason for this difference in binding at saturation is unclear, but it may contribute to some of the observed differences in FRET efficiency.
Requirement for a Phospholipid Bilayer for Efficient FRET-The dramatic inhibition of FRET following addition of detergent could be the result of conversion of vesicles to micelles thereby disrupting the interaction between PLC and PIP 2 . This suggests that a lipid bilayer is required for association of PLC ␤2 with lipids. An alternate explanation is that the detergent is simply diluting the mole fraction of PE-rho on the surface of the vesicle, thereby reducing the efficiency of FRET. To distinguish between these possibilities, the concentration dependence of detergent effects on FRET were examined with vesicles containing either 1% PE-rho or 10% PE-rho. The results from this experiment are shown in Fig. 4. C 12 E 10 inhibits FRET with an IC50 of ϳ100 M for vesicles containing 1% PE-rho and ϳ200 M for vesicles containing 10% PE-rho. If the inhibition were truly a dilution phenomenon, then there should be a 10-fold difference in the IC 50 for C 12 E 10 .
The sharp dependence on concentration for inhibition of FRET (Fig. 4) is suggestive of a cooperative process that probably corresponds to the formation of micelles. The critical micellar concentration (CMC) for C 12 E 10 is approximately 100 M, which corresponds to the IC 50 for inhibition of FRET. If octyl glucoside is used for the same experiment, inhibition is again independent of PE-rho mole fraction and occurs between 10 and 30 mM octyl glucoside, which corresponds to the CMC for octyl glucoside (data not shown). These data suggests that it is the formation of micelles that disrupts FRET, not simply dilution on the vesicle surface.
Another possible explanation for the observations is that significant levels of detergent do not partition in to the phospholipid vesicles below the CMC and only when micelles are formed is significant dilution on the surface observed. In the experiments shown in Fig. 4, 50 M total lipid is present at each detergent concentration (10 M PIP 2 , 40 M PE). At the IC 50 values, the lipids are diluted 3-fold in the case of 1% PE-rho vesicles or 5-fold in the case of 10% PE-rho vesicles (assuming all the C 12 E 10 partitions into the micelles). Thus, at the respective IC 50 values, the 10% PE-rho vesicles have a 6-fold higher effective concentration of PE-rho on the surface of the micelle than 1% PE-rho vesicles. This indicates that dilution is not responsible for the inhibition and the disruption of FRET must be due to the physical conversion of vesicles to micelles.
Effects of ␤␥ Subunits and Ca 2ϩ on Affinity of PLC ␤2 for Vesicles-We measured the effects of G protein ␤␥ subunits on the affinity and final extent of binding of F-PLC ␤2 to substrate vesicles using FRET. This binding is assessed under conditions that are known to support the activation of PLC ␤2 by ␤␥ subunits (8). The results presented in Fig. 5 demonstrate association of F-PLC ␤2 with PE:PIP 2 :PE-rho (4:1:0.05) vesicles (Fig. 5A) or PE:PS:PE-rho (4:1:0.05) vesicles (Fig. 5B) as a function of lipid concentration in the absence or presence of ␤␥ subunits (300 nM). Neither ␤␥ subunits nor Ca 2ϩ has a significant effect on the interaction of F-PLC ␤2 with either PIP 2 -or PS-containing vesicles. Binding to PIP 2 -containing vesicles was slightly inhibited in the presence of Ca 2ϩ , which may have been due to substrate hydrolysis during the course of the measurements. To support the results from FRET, the effects of ␤␥ were examined in an experiment where vesicles were centrifuged to separate bound and free F-PLC ␤2 at various vesicle concentrations. The results in Fig. 6 show that ␤␥ subunits do not alter the binding of F-PLC ␤2 to these vesicles in agreement with the data obtained by FRET.
Significant substrate hydrolysis occurs at low PIP 2 concentrations in the presence of both ␤␥ and Ca 2ϩ , making affinity measurements difficult. For this reason we tested binding at a fixed, relatively high concentration (50 M) of PIP 2 in the presence of both ␤␥ subunits and Ca 2ϩ and measured the fluorescence immediately after addition of lipids. This concentration of PIP 2 is the same as that used in our enzyme assays and the conditions exactly mimic conditions known to support significant PLC ␤2 activation by ␤␥ subunits. Since the interaction between F-PLC ␤2 and vesicles, as assessed by FRET, approaches saturation above 10 M (a hypothesis supported by the centrifugation experiments in Figs. 3 and 6), it suggests that the majority of the F-PLC ␤2 is bound to vesicles at 50 M and therefore that it would not be possible for ␤␥ subunits to activate the PLC by translocation. The results presented in Fig.  7A demonstrate that ␤␥ and Ca 2ϩ together do not significantly alter the association of PLC ␤2 with the vesicle surface. In Fig.  7B is shown an experiment where activation of PLC ␤2 was measured under identical conditions to the FRET assay, demonstrating a strong activation of PLC ␤2 by ␤␥ subunits. DISCUSSION Soluble enzymes whose lipid substrates are localized to membrane surfaces can potentially be regulated by controlling the equilibrium association of the enzyme with the surface of the lipid bilayer. The paradigm for this type of enzyme is secretory phospholipase A 2 (reviewed in Ref. 34). A number of G protein ␤␥ subunit-regulated effectors are also likely to be regulated by the equilibrium association of the enzyme with the cell membrane, including PLC ␤2 and ␤3, ␤-adrenergic receptor kinase (10), Brutons tyrosine kinase, and phosphatidylinositol 3-kinase (13). To determine if altering equilibrium vesicle association of effectors by ␤␥ subunits is involved in their activation mechanism, it is critical that measurements of vesicle binding are performed under conditions where regulation of enzymatic activity by G protein ␤␥ subunits is known to occur. The goal of the experiments presented here was to monitor the association of PLC ␤2 with phospholipid bilayers under conditions that are known to support the activation of PLC ␤2 by ␤␥ subunits. We chose to use fluorescence energy transfer to monitor vesicle association, because it allows equilibrium measurements to be made under virtually any reaction condition. Thus, it can be determined whether factors known to influence PIP 2 hydrolysis under certain conditions do so by influencing PLC ␤2-membrane bilayer association under that same condition.
The results presented here demonstrate that resonance energy transfer can be used to monitor the association of PLC ␤2 with phospholipid vesicles. Data obtained by centrifugation analysis strongly support and add to the results obtained by FRET. In these assays several factors can affect FRET: 1) association of F-PLC ␤2 with the bilayer; 2) the orientation of the protein and fluorophore on the vesicle surface; 3) the concentration of PE-rho on the vesicle surface. Since the PE-rho mole fraction on the vesicle surface is constant in all of our experiments, this factor is not influencing the data presented here. Changes in affinity for the vesicle surface will be reflected in changes in the concentration dependence for binding, while orientation factors will influence the maximum transfer effi- ciency that occurs at saturation. Changes in both of these parameters can be observed in the data presented. For example, while a greater resonance energy transfer is observed for the interaction of F-PLC ␤2 with PIP 2 :PE vesicles relative to PE vesicles containing other anionic lipids, the binding affinities for these vesicles are similar. We do, however, detect a specific interaction with PIP 2 that is probably the result of positioning of F-PLC ␤2 on the membrane surface such that a relatively high resonance energy transfer efficiency is observed.
Vesicle association can be disrupted by low concentrations of C 12 E 10 in a manner that is insensitive to the PE-rho mole fraction. That detergent disrupts the interaction suggests that association is not solely based on the presence of PIP 2 or other anionic lipids. The phospholipid head groups are still present in the mixed micelles that result after detergent addition, yet no F-PLC ␤2 binding to these micelles is detected. This suggests that the PIP 2 must be presented in the context of a phospholipid bilayer to permit a high affinity association of PLC ␤2. PIP 2 -dependent association of PLC isoforms with vesicles has been reported previously. PLC ␤ and ␦ isozymes were shown to bind to PIP 2 containing vesicles in a centrifugationbased assay (35,36). In these experiments it was reported that vesicle binding was critically dependent on the presence of PIP 2 , which could not be duplicated by the addition of PS. We observe a specific and unique interaction of PLC with PIP 2 detected as an increase in FRET efficiency, but in our hands PIP 2 does not alter the affinity of F-PLC ␤2 for vesicles. Several factors distinguish the experiments here from those presented previously. In the experiments by the other investigators, PIP 2 or PS were included in the vesicles at a small mole fraction (3%) of PC vesicles. It is possible that PLC ␤2 does not bind well to PS in the context of PC lipid or that the low mole fraction of PS did not allow association because of insufficient charge on the vesicle. This could explain our apparent lack of an absolute requirement for PIP 2 for lipid association. The conditions used for their experiments were unlikely to support enzyme activity and certainly would not support activation by ␤␥ subunits. Thus, while specific conditions may be found where PIP 2 is required for PLC-vesicle association, our conditions are particularly relevant to measurable enzyme activity and regulation by G proteins.
The binding of PLC ␦ to PIP 2 containing vesicles is thought to occur via a pleckstrin homology (PH) domain at the N terminus. Binding to vesicles is blocked by addition of 15-60 M D-IP 3 , but not L-IP 3 (30). It has been shown recently that the isolated PH domain from PLC ␦ binds D-IP 3 with high affinity (1.7 M) (31). PLC ␤ isoforms have been proposed to contain PH domains at their N termini (37). In our experiments 1 mM D-IP 3 has no effect on PLC ␤2 binding to vesicles, measured either as a change in affinity or FRET efficiency (data not shown). Additionally the binding is mimicked by inclusion of other anionic lipids as has been discussed. This suggests that either PLC ␤2 PH domain is not required for vesicle association or that the PH domain on this protein is relatively nonselective. It is possible, however, that binding of the PH domain to PIP 2 is required for the orientation of the PLC on the membrane surface that yields high efficiency FRET and that IP 3 does not interact with the PH domain on PLC ␤ isoforms. A role for putative PH domains in PLC ␤ isozymes remains to be demonstrated.
Inclusion of ␤␥ subunits in the vesicle binding assays has no effect on the affinity of F-PLC ␤2 for PIP 2 vesicles or on the orientation of the PLC on the vesicle surface. In our in vitro assays we normally monitor ␤␥-mediated increases in PLC ␤ activity using 50 M PIP 2 in the presence of 0.001-1.0 M Ca 2ϩ . The fact that binding of PLC ␤2 to vesicles occurs at low lipid concentrations, and that this binding approaches saturation at concentrations below those that have been used to assay ␤␥stimulated PLC ␤2 activity, indicate that under standard assay conditions (50 M PIP 2 ), most of the PLC is bound to the vesicle surface. Enhancing this interaction, therefore, is not a viable mechanism for activation of this enzyme by ␤␥ subunits, Ca 2ϩ , or any other factors at this PIP 2 concentration. This is further emphasized by the direct demonstration that neither ␤␥ nor Ca 2ϩ influences the extent of PLC ␤2 binding to these vesicles, while under identical conditions, ␤␥ significantly increases the activation of PLC ␤2 enzymatic activity (Fig. 7). Also of significance is that neither ␤␥ or Ca 2ϩ alters the FRET efficiency on the surface of PIP 2 vesicles, suggesting that ␤␥ subunits are not significantly altering the orientation of the PLC on the vesicle surface and the specific interaction with PIP 2 that we observe.
Demonstration that ␤␥ subunits do not translocate PLC ␤2 in vitro does not preclude a role for this process in vivo. These results show that translocation is not required to observe enzyme activation by ␤␥ subunits and that other kinetic mechanisms such as altering K m or V max must be involved in the activation of the enzyme. If one considers that PS, PIs, and PA are restricted to the inner surface of the plasma membrane, there is considerable anionic character to this surface. This suggests that PLC ␤ isoforms could bind to the plasma membrane surface by binding to PIP 2 and/or other anionic lipids to bring the enzyme in proximity to the G protein and the substrate.