The Small G Protein Rac1 Activates Phospholipase Cδ1 through Phospholipase Cβ2*

Rac1, which is associated with cytoskeletal pathways, can activate phospholipase Cβ2 (PLCβ2) to increase intracellular Ca2+ levels. This increased Ca2+ can in turn activate the very robust PLCδ1 to synergize Ca2+ signals. We have previously found that PLCβ2 will bind to and inhibit PLCδ1 in solution by an unknown mechanism and that PLCβ2·PLCδ1 complexes can be disrupted by Gβγ subunits. However, because the major populations of PLCβ2 and PLCδ1 are cytosolic, their regulation by Gβγ subunits is not clear. Here, we have found that the pleckstrin homology (PH) domains of PLCβ2 and PLCβ3 are the regions that result in PLCδ1 binding and inhibition. In cells, PLCβ2·PLCδ1 form complexes as seen by Förster resonance energy transfer and co-immunoprecipitation, and microinjection of PHβ2 dissociates the complex. Using PHβ2 as a tool to assess the contribution of PLCβ inhibition of PLCδ1 to Ca2+ release, we found that, although PHβ2 only results in a 25% inhibition of PLCδ1 in solution, in cells the presence of PHβ2 appears to eliminates Ca2+ release suggesting a large threshold effect. We found that the small plasma membrane population of PLCβ2·PLCδ1 is disrupted by activation of heterotrimeric G proteins, and that the major cytosolic population of the complexes are disrupted by Rac1 activation. Thus, the activity of PLCδ1 is controlled by the amount of bound PLCβ2 that changes with displacement of the enzyme by heterotrimeric or small G proteins. Through PLCβ2, PLCδ1 activation is linked to surface receptors as well as signals that mediate cytoskeletal pathways.

Mammalian phospholipase Cs (PLCs) 2 are critical signaling enzymes whose activity results in an increase in intracellular Ca 2ϩ through the hydrolysis of PI(4,5)P 2 . There are now nine known types of mammalian PLCs that vary in their tissue distribution and their cellular regulation (for review see Refs. 1, 2). Most PLCs have established protein regulators, and many appear to have multiple mechanisms of regulation that, in principle, may connect different signaling pathways. It is possible that, under some circumstances, different PLC isozymes coordinate to elicit a specific cellular response. As described below, we have recently found that PLC␤ and PLC␦ couple to synergize Ca 2ϩ responses by surface receptors.
The PLC␤ family is activated by heterotrimeric G proteins, which respond to extracellular agonists such as hormones and neurotransmitters through transmembrane G protein-coupled receptors (see Ref. 3). All four known PLC␤ (␤1-4) are strongly activated by G␣ q , and PLC␤2-3 are also activated by G␤␥ subunits. Additionally, PLC␤2 is activated by members of the Rho family of small G proteins; most strongly by Rac1 (for review see Ref. 4). Rac1 is activated by the phosphoinositide 3-kinase pathway to initiate events leading to cytoskeletal rearrangements (5,6).
Although PLC␤ has well characterized protein regulators, protein regulators of PLC␦ are less established. Unlike other PLC families, PLC␦ enzymes are inactive at basal calcium concentrations but become active upon the rise in intracellular Ca 2ϩ brought about by the activity of other PLCs. Two of the three reported PLC␦ regulators, RhoA and transglutaminase, lower the amount of Ca 2ϩ required for activation (7,8). Recently, our laboratory discovered a third PLC␦ regulator, PLC␤2 and -␤3 (9). These enzymes were found to bind to PLC␦1 and inhibit its activity. Association between PLC␤2-3 and PLC␦1 can be disrupted by the addition of G␤␥ subunits in solution, and preliminary studies in cultured cells suggest that the same association may occur in cells. These results suggested a model in which, upon release of G␤␥ subunits after stimulation, PLC␤ becomes activated by G␤␥ subunits, and concomitantly activates PLC␦ through a combination of increased level of Ca 2ϩ and loss of inhibition by PLC␤2 and -␤3.
To better understand the linkage between PLC␦ and cell surface receptors, we have found that the pleckstrin homology (PH) domain of PLC␤2 and -␤3 is responsible for binding and inhibition of PLC␦1 and have used this as a reagent to study the importance of PLC␤ inhibition of PLC␦1 in cells. We first found that a significant population of PLC␤2 and PLC␦1 associate in cells. Stimulation by carbachol reduces the association of the plasma membrane-localized population, whereas microinjection of activated Rac1 reduces the association in the cytosolic population. Microinjection of the PH domain of PLC␤2 (PH␤2) resulted in a decrease in FRET between PLC␤2 and PLC␦1 suggesting that the interaction between these two proteins in cells is achieved through the PH domain. Moreover, microinjection of PH␤2 resulted in a greatly reduced Ca 2ϩ release after stimulation with carbachol suggesting that the extent of a Ca 2ϩ response is directly related to the level of unbound PLC␦1.
These studies link PLC␦1 regulation to surface receptors and cytoskeletal movement through PLC␤2.

MATERIALS AND METHODS
In Vitro Studies-His 6 -PLC␤2 was expressed in Sf9 cells using a baculovirus system with minor modifications (see Ref. 10 for details about the expression and purification). His 6 proteins PLC␦1, PH␤2, PH␤3, ⌬PH-PLC1, and Rac1 were expressed in Escherichia coli and purified on a Ni 2ϩ column using previously reported methods (see Refs. 9,[11][12][13][14]. Expression and purity was assessed by Western blot analysis using commercial antibodies purchased from Santa Cruz Biochemicals, and by SDS-PAGE electrophoresis. PH␤2 C18S was prepared by the Molecular Cloning Facility at Stony Brook University. PLC activity was assessed by measuring the hydrolysis of [ 3 H]PI(4,5)P 2 dispersed on sonicated phosphatidylserine:phosphatidylethanolamine:PI(4,5)P 2 (2:1:0.5, v/v) membranes as described previously (15). Note that the assays are carried out at a total lipid concentration of 2 mM giving a large excess of PI(4,5)P 2 substrate as compared with the nanomolar amount of protein. For the inhibition studies, we used the protein concentrations above the dissociation constant.
Protein association was measured by labeling one of the proteins by the addition of a 4:1 probe:protein molar ratio of the thiol-reactive probe 7-diethylamino-3-(4Ј-maleimidylphenyl)-4-methylcoumarin (CPM) in the absence of reducing agents (for details see Ref. 16). After labeling, a small amount of protein ranging in concentration from 1 to 20 nM was placed in a microcuvette and reconstituted on large unilamellar vesicles composed of PI(4,5)P 2 , palmitoyl oleoyl phosphatidylethanolamine and palmitoyl oleoyl phosphatidyl-serine prepared by extrusion through a 100-Å membrane according to the manufacturer's instructions (Avanti Biochemicals, Alabaster, AL). Protein association was assessed by plotting the increase in integrated intensity of CPM as the second protein was incrementally added and fitting the curve to a biomolecular association (see Ref. 17). Measurements were taken on an ISS PC1 spectrofluorometer using (ex) ϭ 360 nm and scanning the fluorescence emission from (em) ϭ 390 -590 nm.
Cell Culture and Overexpression of PLCs-HEK293 cells were maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum, 50 units/ml penicillin, and 50 g/ml streptomycin sulfate at 37°C in 5% CO 2 . For fluorescence studies, HEK293 cells were transfected with eYFP-PLC␤2 and eCFP-PLC␦1 vectors or eGFP-PLC␦1 vector (5-10 g/10 6 cells in 60-mm dish) by calcium phosphate coprecipitation (9). The protein expression of endogenous PLC␦1 was knocked down using small interference RNA PLC␦1 (Dharmacon Inc.) according to the manufacturer's instructions along with the negative control purchased from the manufacturer. Western blot analysis showed this procedure produces ϳ80% knock down.
After addition of 20 mg of protein A-Sepharose beads, the mixture was gently rotated for 4 h at 4°C. Beads were washed three times with lysis buffer, and bound proteins were eluted with sample buffer for 5 min at 95°C. Precipitated proteins were loaded onto 8% polyacrylamide gel. For purified protein coimmunoprecipitation instead of cell lysate the same amount PLC␤3 and PLC␦1 was added. After SDS-PAGE and transfer to polyvinylidene difluoride membranes, protein were detected by immunoblotting with anti-PLC␤3 (Santa Cruz Biotechnology, Santa Cruz, CA) antibody.
Imaging of Fluorescence in Living Cells-24 h after transfection, cells from 60-mm dishes were plated onto glass bottom culture dishes (MatTek). Images of fluorescent cells were collected 48 -72 h after transfection on an Olympus Fluoview1000 confocal microscope equipped with a 40 ϫ 1.4 numerical aperture oil immersion objective. Analysis of the FRET images was performed using standard routine incorporated into software provided for the Olympus microscope. The FRET analysis used in this report differs slightly from our previous analysis (e.g. Refs. 18,19) where the efficiency, E, is calculated using Equation 1, where ⑀ is the fluorescence through the CFP channel, and nFRET is equal to the image obtained through the FRET channel minus the CFP and YFP bleedthrough (a and b, respectively), and in Equation 2, r⌿ is the ratio of the detector response of the CFP and YFP channels, and rQ is the ratio of the quantum yields of the CFP and YFP.
Microinjection Samples and Conditions-Transfected cells were grown in MatTek dishes for 48 h to achieve 50 -60% confluence. Prior to microinjecting, the medium was changed to phenol-free Leibovitz L-15 medium. PH␤2 (50 nM) was microinjected with solution containing 0.2% deoxycholic acid, 20 mM Hepes, 150 mM NaCl, pH 7.2, and trace amounts of the red dye Cy5 (Invitrogen). The control solution did not contain PH␤2. Prior to microinjections, PH␤2 S18C was labeled with Alexa546, which served as a acceptor for GFP-PLC␦1.
Rac1 was activated using the procedure described in a previous study (13). Briefly, the sample was dialyzed against 20 mM Hepes, 150 mM NaCl, 1 mM dithiothreitol, pH 7.2, for 3 ϫ 15 min at 4°C, and then incubated with reaction buffer containing 50 mM HEPES (pH 8.0), 1 mM dithiothreitol, 2 mM EDTA, 0.1% Lubrol, and 1 M GTP␥S at 30°C for 15 min. The reaction buffer was used as the control solution for microinjections.
Microinjections were performed on an Axiovert 200M from Zeiss using InjectMan NI2 with a FemtoJet pump from Eppendorf. Samples were microinjected into cytoplasm. We typically set the injection pressure P i ϭ 30 hPa and kept the compensation pressure (P c ) at 15 hPa. The injection time (t) was 0.4 s. Typically, we injected ϳ10 -25 cells within a 10-to 20-min period. We examined the microinjected cells under the phase microscope (Axiovert 200M from Zeiss with a 40ϫ phase 2 objective) to select viable cells. Cells were then transferred to the Olympus Fluoview1000 for viewing, which was carried out within 2 h after microinjection.
Measurement of Cellular [Ca 2ϩ ] i -For single cell Ca 2ϩ measurements, cells were labeled with 1 M Calcium Orange or Calcium Green (Invitrogen) for 45 min at room temperature, washed 3 times with Leibovitz L-15 medium, kept at room temperature in the dark without dye for 30 min, and then washed again three times with Leibovitz L-15 medium. The additional washes are required to remove dye that is nonspecifically bound to the plasma membrane. Cells were imaged on a Zeiss LSM510 laser scanning confocal microscope. Calcium Green was excited at 488 nm with an argon ion laser, Calcium Orange was excited with a 543 nm HeNe laser line, and the emission spectrum was recorded using a long pass 560 nm filter. Images were analyzed using Zeiss software. Images were collected every 20 s over a 5-min period. Data were analyzed by comparing the change in intensity of a selected cell or group of cells over the time period. This selection insures that dark pixels are not considered. Usually, five or six cells or cell groups were selected in each experiment.

RESULTS
The PH Domain of PLC␤2/3 Is Responsible for PLC␦ Inhibition-Before studying the regulation of PLC␦1 by PLC␤2 in cells, we set out to identify the region of PLC␤2 that is responsible for PLC␦1 inhibition with the goal of developing a reagent to inhibit association of the enzymes. Mammalian PLCs have a modular structure that are generally composed of an N-terminal PH domain followed by two EF hands, a catalytic domain, and a C2 domain (Fig. 1A). PLC␤ enzymes have a distinct C-terminal region that is required for activation by G␣ q subunits (see Refs. 2,20). Several years ago, Nagano and coworkers demonstrated that the PH domain of an inactive variant of PLC␦4 could inhibit PLC␦4 activity (21). Therefore, we wondered whether the PH domain of PLC␤2 could be involved in PLC␦1 inhibition. Additionally, we have found that G␤␥, which makes a primary binding contact with the PH domain of PLC␤2 (11), disrupts PLC␤2⅐PLC␦1 association, also suggesting that association may occur through this region.
We measured the ability of the PH domains of PLC␤2 and -␤3 to inhibit PLC␦1 and its isolated catalytic domain (Table 1). To compare the PH domains to the full-length enzymes, we inactivated PLC␤2 and PLC␤3 by covalently modifying the catalytic His residue with diethyl pyrocarbonate as in previous studies (9). Activity measurements were carried out at high PI(4,5)P 2 levels to avoid competition between the two PLCs for substrate. Similar to our earlier work (9), we found that the full-length PLC␤ enzymes are able to inhibit PLC␦ by 40% (e.g. Fig. 1B). A similar inhibition was observed when we substituted the catalytic domain of PLC␦ for the full-length enzyme. These data suggest that PLC␦ inhibition is caused by contacts between PLC␤2 and the catalytic domain of PLC␦1 but not other PLC␦1 domains.
We then measured the inhibition of PLC␦1 by the PH domains of PLC␤2 and -␤3. Although the amount of PLC␦1 inhibition by the PH domains was reduced as compared with the full-length enzyme (i.e. 25-30% see Table 1), it was still significant. We note that, unlike the full-length enzyme, the PH domains were unable to inhibit the isolated catalytic domain of PLC␦1. These results fit well with the idea that the PH domain This experiment shows the decrease in the activity of 10 nM ⌬PH-PLC␦1 with the addition of full-length PLC␤2 where the substrate, PI(4,5)P 2 , was dispersed in 20 mM sonicated vesicles composed of phosphatidylcholine:phosphatidylethanolamine:phosphatidylserine with 5% PIP(4,5) 2 . C, a graph illustrating the binding of PH␤3 to full-length CPM⅐PLC␦1 and CPM-⌬PH-PLC␦1, where association was assessed by the normalized change in fluorescence intensity, which averaged 18% for CPM⅐PLC␦1 and 11% for CPM-⌬PH-PLC␦1. For both sets of data, n ϭ 3 and standard deviation is shown. A compilation of these data can be found in Table 1.

TABLE 1 Binding and inhibition of PLC␦1 proteins by PLC␤ proteins
Binding studies were carried out by adding unlabeled protein (listed leftmost), to CPM-labeled protein partner. CPM-labeled protein was prebound to 200 M phosphatidylserine:phosphatidylcholine (2:1, v/v). Activity measurements were conducted as described under "Materials and Methods." Mean values and standard deviations are given. Activity measurements were carried out using 2 nM PLC␦1 and 5 nM ⌬PH-PLC␦1 and adding excess inactive (i.e. DEPC-treated) PLC␤2/␤3 or PH␤2/␤3 so that all of the proteins should be associated. The initial k cat of PLC␦1 under our experimental conditions was 140 Ϯ 12 s Ϫ1 and 47 Ϯ 10 s Ϫ1 for ⌬PH-PLC␦1.
To determine whether differences in PLC␦1 inhibition by full-length PLC␤2/␤3 and its PH domains are due to differences in their binding strength, we measured the affinity of the enzymes and their constructs using fluorescence spectroscopy. Specifically, we labeled one of the protein constructs with an environmentally sensitive fluorophore, CPM, and measured the changes in its fluorescence as the unlabeled binding partner was incrementally added (see Ref. 17). An example of a titration is shown in Fig. 1C, and the results are tabulated in Table 1. We found that the PH domains of PH␤s and PH␦ did not appear to interact while the other constructs all showed a strong and similar affinity whose apparent dissociation constants (K d ) ranged from 1.5 to 15 nM. The similarities in affinities suggest that binding occurs between the PH domains of PLC␤2/3 and the catalytic domain of PLC␦1. We note that, in general, PLC␤3 and its PH domain bound to PLC␦1 and ⌬PH-PLC␦1 with a slightly stronger affinity than PLC␤2.
Rac1 Disrupts PLC␤2/␤3 and PLC␦1 Association-The structure of a large portion of PLC␤2, including the PH and catalytic domains, bound to activated Rac1 has been solved (22). This structure shows that Rac1 associates with the PH domain of PLC␤2. Thus, if PLC␤2 associates with PLC␦1 through its PH domain, then addition of Rac1 would be expected to disrupt that association. We measured the binding between PLC␤2 and PLC␦1, and between PH␤2 and PLC␦1 in the absence and presence of Rac1(GDP) and Rac1(GTP␥S). We found that, even in the deactivated form, Rac1 inhibited association between the enzymes (Fig. 2 left).
In a separate study, we added Dabcyl-Cl-labeled PLC␦1 to a solution of CPM⅐PLC␤2. Dabcyl is a non-fluorescent FRET acceptor, and a loss in CPM fluorescence has been observed due to transfer of CPM emission to this non-fluorescent probe (see Ref. 17). We added Dab-PLC␦1 to CPM⅐PLC␤2 at a concentra-tion so that 50% of the proteins were associated. We then measured the ability of deactivated and activated Rac1 to disrupt the PLC␤2⅐PLC␦1 complex. We found that activated Rac1 was better able to dissociate the two PLCs (Fig. 2, right). These results support the idea that the PH domain of PLC␤2 is responsible for interaction with PLC␦1.
Association between PLC␤2 and PLC␦1 in Cells-We studied the association between PLC␤2 and PLC␦1 in HEK293 cells by co-transfecting the cells with fluorescence-tagged chimers. It has been previously shown that GFP tags on the C terminus of PLC␦1 do not affect its cellular localization (22). In Fig. 3A we compare the localization of untagged and eYFP-tagged PLC␤2 expressed in HEK293 cells where the untagged protein was detected by immunofluorescence. We also present images of endogenous PLC␤2, which is expressed at low levels in these cells. These results suggest that the eYFP tag does not affect cellular localization of PLC␤2. Western blot analysis of untagged and eYFP-PLC␤2 from the soluble and membrane fractions of HEK293 cells gave identical results (data not shown).
Expression of eCFP-PLC␦1 in HEK293 cells showed the enzyme to be widely dispersed with a distinct plasma membrane localization as well as populations in the cytosol and nucleus (Fig. 3, top). This distribution has been previously observed (23,24). Similar to previous reports (23), stimulation of the cells with carbachol caused translocation of a population of the enzyme to edges of the cell. In contrast, eYFP-PLC␤2 was almost entirely cytosolic without a distinct plasma membrane component and no nuclear localization (Fig. 3, bottom). Unlike PLC␦1, stimulation of the cells with carbachol did not significantly change localization of overexpressed or endogenous PLC␤2 (Fig. 3, top and middle panels).
We have previously found that PLC␤2 and PLC␦1 associate in cells using biomolecular fluorescence complementation (see Ref. 9). In this method, two portions of eYFP are attached to potential protein partners, and association of these partners reconstitutes the eYFP generating a fluorescent signal. However, this method only detects associated proteins, and nonassociated proteins are optically silent. Therefore, we used FRET, which detects the fluorescence of the individual proteins as well as the amount of energy transfer from the associated species (25). This method allows us to identify regions of the cell where both the unassociated and associated enzymes are localized.
We transfected HEK293 cells with eYFP-PLC␤2 and CFP-PLC␦1, viewed the localization of the fluorescent proteins, and measured the degree of FRET in the localized areas (Fig. 4 A), where all of the FRET data presented are corrected for bleedthrough and the small amount of photobleaching (Ͻ5%) that is sometimes observed (see "Materials and Methods"). Keeping in mind that the eYFP label is on the N terminus of PLC␤2 and the eCFP label is on the C terminus of PLC␦1, and that association is through the PH␤2, then a crude estimate of the distance between the two probes is Ͻϳ100 Å when the enzymes are associated. The R o , or the distance in which 50% of light is lost to FRET, is ϳ50 Å for this pair of fluorophores (26). Thus, the presence of FRET observed should be directly related to the physical association between the two enzymes. Using a positive control where eCFP and eYFP are attached through a dodecapeptide flexible linker, and a negative control consisting of free eCFP and free eYFP, we found that ϳ20% of the enzymes in the cytosol were associated as estimated by FRET (Fig. 4C). We supported this result by co-immunoprecipitation studies of endogenous PLC␦1 and PLC␤3, which is expressed at a higher level than PLC␤2 in these cells allowing for more accurate detection (Fig. 4B). These studies support the idea that PLC␤ and PLC␦ associate in cells.
Changes in PLC␤2-PLC␦1 Association with Stimulation-We determined whether stimulation would affect the association of the two enzymes. We first monitored the amount of FRET between eYFP-PLC␤2 and CFP-PLC␦1 when we activated muscarinic acid receptors by the addition of carbachol. Although carbachol stimulation did not significantly change the amount of FRET in the cytosol, we found that it reduced the amount of FRET from the small plasma membrane fraction in a time-dependent manner supporting the observation that released G␤␥ subunits can dissociate the complexes (Fig. 5). We interpret the residual FRET as due to the inability to generate enough G␤␥ subunits to completely disrupt the complexes.
We then tested whether another activator of PLC␤2, the small GTPase, Rac1, disrupts the association between the two PLCs. To this end, we monitored the amount of FRET between eYFP-PLC␤2 and CFP-PLC␦1 when activated Rac1 was microinjected into the cells. We identified microinjected cells by including trace amounts of the red dye Cy5, which will not interfere with the absorption or fluorescence of the eCFP or eYFP probes, into the microinjection solution (Fig. 6, top). We found that microinjection of activated Rac1 abolishes FRET between the two PLCs (Fig. 6, bottom). Identical studies that substitute buffer or deactivated Rac1 did not affect the FRET (Fig. 6, bottom).
Association of PH␤2 to PLC␦1 Diminishes the Ca 2ϩ Response to Stimulation-Because we found that PH␤2 is the main region of interaction between PLC␤2 and PLC␦1, we used this domain as a reagent to determine whether the associated enzymes in the cytosol contribute to Ca 2ϩ release with carbachol stimulation. To carry out these studies, we introduced a Cys residue into PH␤2 (C18A) on an external site, which allowed us to specifically label the domain with Alexa546. We then monitored FRET from GFP-PLC␦1 donors to microinjected Alexa546-PH␤ acceptors (Fig. 7). As with the full-length proteins, FRET was localized in the cytosol (Fig. 7A). The FRET value obtained for this combination is within error of the value seen for the full-length proteins supporting the idea
Because PH␤2 binds to PLC␦1 and inhibits its activity, we tested whether this association could quench Ca 2ϩ signals generated with carbachol stimulation. We microinjected unlabeled PH␤2 into untransfected HEK293 cells treated with a Ca 2ϩ indicator and measured the amount of Ca 2ϩ released upon carbachol stimulation at 20 s. Our control studies showed a lack of Ca 2ϩ response with epidermal growth factor stimulation (see Ref. 27). We note that PH␤2 does not specifically bind to PI(4,5)P 2 or Ins(1,4,5)P 3 and should not interfere with substrate or product binding of PLC␦1 (11). We found a statistically significant reduction in the amount of released Ca 2ϩ in the microinjected cells as compared with non-injected cells, and to cells that were microinjected only with microinjection buffer (Fig. 8, A and  B). This result correlates well with our observation that PLC␦1 small interference RNA treatment, which results in ϳ80% reducing in the level of protein, abolishes a significant calcium response as detected by calcium green (Fig. 8C). These results suggest that the amount of Ca 2ϩ released is directly related to the amount of free PLC␦1.
The lack of Ca 2ϩ response with microinjected PH␤2 could be a result of it binding to its G␤␥ subunits, which would prevent PLC␤2 activation and leave the basal level of Ca 2ϩ below the threshold needed for PLC␦1 activity. Although this possibility is unlikely given the cytosolic localization of PH␤2 as opposed the plasma membrane-localized G␤␥, we microinjected Alexa546-PH␤2 into cells expressing eGFP-G␤␥ subunits. We could not detect FRET between these two proteins, but this did not exclude the possibility that microinjected PH␤2 binds to G␤␥ subunits.

DISCUSSION
The regulation of calcium signals in cells involves a complex series of events that are mediated by PLC activity. PLC␦1 is activated by the rise in intracellular Ca 2ϩ generated by other PLCs in response to signaling events, such as G protein activation and receptor tyrosine kinase activation, and can be thought of as a Ca 2ϩ signal amplifier. The PLC␦ effectors, RhoA and transglutaminase, reduce the levels of Ca 2ϩ needed to activate PLC␦ (7,8). We have previously uncovered a novel regulator of PLC␦1, namely PLC␤2 (9). PLC␤2 inhibits PLC␦1, and this inhibition is reversed when G␤␥ displaces PLC␤2 from PLC␦1. This mechanism of regulation directly connects PLC␦1 activity to surface receptors. In this study, we connect PLC␦1 activity with Rac1 activation of PLC␤2.
Our experiments show that the PH domains of PLC␤2 and -␤3 are responsible for binding and inhibition of PLC␦1. This novel function of the PH␤ domains adds to the two established roles that PH␤ plays:

Rac1 Activates PLC␦1 through PLC␤2
AUGUST 6, 2010 • VOLUME 285 • NUMBER 32 anchoring the host enzyme to the surface of membranes and docking the host enzyme to G␤␥ and to RhoGTPase proteins (11,28). Here, we found that the PH domain of PLC␦1 does not participate in the association with PLC␤2, suggesting that its functional role is solely to bind PI(4,5)P 2 and Ins(1,4,5)P 3 (12) thereby promoting membrane association and dissociation, respectively, of the host enzyme (see below). Although we followed the association of full-length PLCs and their PH domains bound to a model membrane, we have previously shown that these proteins will associate in the absence of membranes but with a reduced affinity (9,29).
It is well established that activation of PLC␦1 occurs through specific binding to PI(4,5)P 2 on the membrane surface (12). Not only does the PH domain anchor the enzyme to the membrane surface allowing it to access substrate, but this interaction has been suggested to reorient the catalytic site with respect to the membrane surface to allow more effective product release. Activation of PLC␤2 is thought to occur by a similar mechanism in which the binding of G␤␥ to the PH domain mediates changes in membrane orientation of the catalytic domain (see Refs. 16,30). 3 Interestingly, the PH domains of PLC␤2 and PLC␦1 can be swapped to produce chimeric enzymes that are activated by either PI(4,5)P 2 or G␤␥ subunits (31,32). Because activation occurs through interdomain movement between the PH and catalytic domains, then it is possible that PH␤2 binds to an auxiliary site on the PLC␦1 catalytic domain that contacts its own PH domain and the catalytic domains upon enzyme activation.
Our previous studies indicated that PLC␤2 almost completely inhibited PLC␦1 activity toward PI (9). Here we observed a 40% inhibition under assay conditions that employ a large excess of PI(4,5)P 2. This difference in inhibition can be traced to the weak binding of PI by PH␦1, which does not allow enzyme activation through interdomain movement, versus excess PI(4,5)P 2 where interdomain PLC␦1 movement is operative. The observation that the PH␤2/3 constructs cannot inhibit ⌬PH-PLC␦1, which cannot be activated by its own PH domain, supports the idea that PLC␤2 and -␤3 are inhibiting the interdomain movement needed for full PLC␦1 activity.
Our in vitro studies suggest that PLC␤3 interacts with PLC␦1 with a somewhat stronger affinity than PLC␤2. PLC␤2 is expressed at low levels in many cell lines but is highly expressed in cells of hematopoietic origin (33). In contrast, tissue expression of PLC␤3 is more ubiquitous and follows PLC␦1 expression more closely (34,35). We focused our studies on PLC␤2, because it is not well expressed in HEK293 cells to minimize the detrimental effects of its overexpression. The regulation of PLC␦1 though PLC␤2 and -␤3 is expected to have a range of importance that depends on the tissue type.
We found that PLC␤2 and PLC␦1 are mainly cytosolic in HEK293 cells despite evidence that the majority of PI(4,5)P 2 , their major substrate, is found on the plasma membrane. We note that the cells were in media containing serum until viewing thereby maintaining basal level activity. The localization pattern we observed here for PLC␦1 is similar to that observed by Rebecchi and coworkers (23). These authors found both active PLC␦1 and a catalytically inactive mutant localized to the cytosol with significant concentration in membrane ruffles, and in the nucleus where it plays a critical role in cell cycle progression (24). Our images suggest that PLC␤2 does not localize to the nucleus.
We previously found that the majority of PLC␤1 is localized on the plasma membrane of PC12 and HEK293 cells where it is complexed with G␣ q (18). In sharp contrast, very little PLC␤2 localizes to the plasma membrane. Instead, PLC␤2 is found predominantly in the cytosol. Previous studies of GFP-PLC␤2 expressed in COS cells show that it is also cytosolic, but a small portion localizes to the plasma membrane when co-expressed with membrane-bound Rac2(12V) (36). Surprisingly, we found that PLC␤2 remains in the cytosol even when the cells are stimulated with carbachol to activate G␣ q and G␤␥. Although the lack of movement of PLC␤2 to the membrane may be due to a number of experimental factors, such as expression level, cell type, etc., it could suggest that PLC␤2 is activated by G protein subunits that internalize in early endosomes after receptor activation. Although we and others found that G␣ q and G␤␥ remain on the plasma membrane after stimulation (37,38), G␤␥ subunits released by G␣ s have been shown to internalize (39).
A second and more plausible explanation for the cytosolic localization of PLC␤2 is that the enzyme is functionally important in this compartment. Our studies suggest that a significant fraction of cytosolic PLC␦1 is associated with PLC␤2. Specifically, we found that the amount of FRET between the two PLCs is ϳ20% of the positive control and approximately half the amount seen when similar FRET studies are carried out using eCFP/eYFP-tagged G proteins subunits. Because the labels on the proteins are at opposite termini, we cannot exclude the possibility that most of the proteins are associated, but the fluorophores are too far apart to engage in FRET. We can only be certain that at least 20% of PLC␦1 population is associated to PLC␤2 suggesting that the remaining populations are either free or have other interacting partners. Support for this association comes from co-immunoprecipitation studies.
We also found that PLC␤2⅐PLC␦1 association is disrupted by activated Rac1 (for review see Ref. 4). Rac1 integrates diverse signals that can involve integrins, stress factors, cytokines, as well as growth, and neurotropic agents, and its activation is associated with lamellipodia formation. It is plausible that the cytosolic PLCs are distributed on intracellular membranes and associated with actin structures via Rac1. Thus, cytosolic Rac1 has the potential to be a more important cellular regulator of PLC␤2 than G protein subunits. It is possible then that the cytosolic PLCs regulate the amount of PI(4,5)P 2 available for cytoskeletal proteins such as cofilin and actin.
PLC␤2 is most sensitive to Rac1 but can be activated by other members of this family (see Ref. 4). In contrast, PLC␦1 is inhibited by RhoA (40) suggesting that its regulation by PLC␤2 might be synergistic with release of RhoA. It is interesting to note that the PLC⑀ enzymes are regulated by RasGTPases (41), and their potential regulation of PLC␦1 has not been determined.
Previously, we used bimolecular fluorescence complementation to show that PLC␤2 and PLC␦1 can associate in cells (9). Although this method only detects associated species, we note that the localization of the bimolecular fluorescence complementation fluorescence is identical to the localization of FRET, which should be the case. We had observed a reduction in bimolecular fluorescence complementation when we focused on the bottom plasma membrane upon carbachol stimulation. In parallel, we find a decrease in the FRET from the plasma membrane fraction with stimulation. Based on our previous studies, we interpret this decrease as being due to the displacement of plasma-membrane localized PLC␤2 from PLC␦1 by G␤␥ subunits released during stimulation.
We used the PH domain of PLC␤2 as a reagent that should allow us to isolate the importance of PLC␤ inhibition of PLC␦1 activity in cells. This reagent remained in the cytosol after microinjection where it bound to PLC␦1. (We note that microinjection of this regent also appeared to reduce eCFP-PLC␦1/eYFP-PLC␤2 FRET, but the sampling population for this study is small, because these experiments were technically challenging.) The association between PH␤2 and PLC␦1 does not change with carbachol stimulation as expected from the observation that only the plasma membrane population of the associated PLCs is affected by stimulation. Interestingly, we found that association of PH␤2 and PLC␦1 in the cytosol inhibits the release of Ca 2ϩ (Fig. 8). This inhibition is not due to binding of PH␤2 to substrate, because we have found the PH␤2 does not bind specifically to PI(4,5)P 2 (11). Inhibition of Ca 2ϩ release may be due to the inability of G␤␥ to activate PLC␤2, because the microinjected PH␤2 bound to the G protein. A lack of PLC␤2 activation in turn may not allow the cellular Ca 2ϩ concentration to rise high enough to activate PLC␦1. We tested this possibility by measuring FIGURE 8. A, example of the change in Calcium Green intensity of a HEK293 cells 20 s after stimulation. B, the response of untransfected HEK293 cells that were either not microinjected (n ϭ 12) or microinjected only with tracer dye (Cy5) in buffer (n ϭ 5) (these data were indistinguishable), as compared with cells that were microinjected with unlabeled PH␤2 and Cy5 tracer (n ϭ 13). C, small interference RNA-PLC␦1 treatment of HEK293 cells results in an approximate 80% reduction of endogenous PLC␦1 as seen by Western blotting, and elimination of a significant response of Calcium Green intensity to 1 M carbachol where cells were imaged 10 s after stimulation.
FRET between PH␤2 and G␤␥ subunits, but no FRET could be detected. Thus, we conclude that PH␤2 is quenching PLC␦1 activity to a low enough level that second messengers cannot be generated. Thus, the amount of Ca 2ϩ generated may be proportional to the amount of unbound PLC␦1.
It is notable that PH␤2 significantly reduced the amount of released Ca 2ϩ , even though our in vitro assays only show an inhibition of ϳ25% for PLC␦1 rather than 40% for the fulllength enzyme. This seeming discrepancy may be due to one or a combination of possible reasons. The first is technical. As shown in Fig. 8, there is considerable spread in the single cell calcium measurements, and so the effect of microinjected PH␤2 on calcium release may be in the lower range of the data. Also, the integrity of PH␤2 might have diminished in the labeling and microinjection steps. It is also possible that the degree of inhibition varies with the membrane surface. As noted, PLC␦1 is activated by specific binding of the PH domain to PI(4,5)P 2 , and our inhibition assays are carried out using excess PI(4,5)P 2 . As discussed above, PH␤2 appears to inhibit PI(4,5)P 2 -induced interdomain movement that allows for PLC␦1 activation. Therefore, a large excess of PI(4,5)P 2 would effectively compete with PH␤2 inhibition. In contrast, in a cellular setting where the relative concentration of PI(4,5)P 2 is very low, then PH␤2 inhibitory interactions could dominate.
The most likely reason for the apparent enhanced inhibition of PLC␦1 by PH␤2 in cells stems from the requirement of PLC␦1 for calcium. After a lag concentration, PLC␦1 follows an exponential activation curve with Ca 2ϩ concentration, and inhibition of a significant population PLC␦1 may not generate enough Ca 2ϩ to result in activation of the entire cellular PLC␦1 population. Thus, the microinjected PH␤2 can effectively keep Ca 2ϩ levels below the threshold needed to activate PLC␦1.
The results here suggest that PLC␤2 may function to keep PLC␦1 inhibited in the basal state. The strong affinity and the degree of FRET suggest association between the two enzymes, which in turn serves to keep a significant population of PLC␦1 inhibited. Because PLC␦1 has very low activity at basal Ca 2ϩ , then the combination of a loss of PLC␤2 inhibition and the increased Ca 2ϩ provide large scale PLC␦1 activation.