The Pleckstrin Homology Domains of Phospholipases C- (cid:1) and - (cid:2) Confer Activation through a Common Site*

Mammalian inositol-specific phospholipase C- (cid:1) 2 (PLC (cid:1) 2 ) and PLC (cid:2) 1 differ in their cellular activators. PLC (cid:1) 2 can be activated by G (cid:1)(cid:3) subunits, whereas PLC (cid:2) 1 can be activated by phosphatidylinositol 4,5 bisphosphate (PI(4,5)P 2 ). For both proteins, the N-terminal pleckstrin homology (PH) domain appears to mediate activation. Here, we have constructed a chimera in which we placed the N-terminal PH domain of PLC (cid:2) 1 into remaining C-terminal regions of PLC (cid:1) 2 . The PH (cid:2) PLC (cid:1) chimera showed PI(4,5)P 2 -dependent membrane binding similar to PLC (cid:2) 1 and a G (cid:1)(cid:3) interaction energy close to that of PLC (cid:2) 1 . Like PLC (cid:2) 1 , the chimera was activated by PI(4,5)P 2 through the PH domain but not by G (cid:1)(cid:3) . Because these and previous results indicate a common site of contact between the PH and catalytic domains in these two enzymes, we computationally docked the known structures of the PH and catalytic domains of PLC (cid:2) 1 . A synthetic peptide whose sequence matches a potential interaction site between the two domains inhibited the basal activity of PLC (cid:1) 2 , PLC (cid:2)

long, 400-residue tail on their C termini, and deletion of this tail abolishes activation by G␣ but not G␤␥ (4 -6). The neighboring C2 domain plays a role in the specific binding of PLC␤ 2 to activated G␣ subunits (7).
Binding and subsequent activation of PLC␤ 2 by G␤␥ subunits is regulated by the N terminus of the protein which contains a pleckstrin homology (PH) domain (8,9). The association of the N terminus to G␤␥ subunits does not appear to be highly specific because the pleckstrin homology domain of a related protein, PLC␦, that is not activated by G proteins binds to G␤␥ with only a 4-fold weaker affinity than that of PLC␤ 2 . These relatively close interaction energies between the two types of PLC can be compared with specific binding of the PLC␤ 2 C2 domain to activated G␣ q , where a similar association to PLC␦ 1 C2 cannot be detected (7).
Because PLCs attach to membranes to access their substrate, characterization of their membrane binding properties has been carried out. It has been found that unlike PLC␦ 1 , which only binds strongly to membranes if its substrate PI(4,5)P 2 is present (10,11), PLC␤ 2 binds strongly and fairly nonspecifically to lipid membranes (12,13). However, membrane binding of both types of proteins appears to be mediated by the N-terminal PH domain (10, 14 -18). Measurements of the membrane association of the isolated PH domains of PLC␤ 2 and PLC␦ 1 show that this region is responsible for both the affinity and specificity of membrane interaction (8).
To better understand the role of the N-terminal PH domains, we have previously constructed a chimera in which the PH domain of PLC␤ 2 was swapped into the catalytic portion of PLC␦ 1 to yield the chimeric protein, PH␤ 2 -PLC␦ 1 (9). Although this protein had the same membrane and G␤␥ interaction properties as wild type PLC␤ 2 , it was also activated by G␤␥, showing that the insertion of the PH of PLC␤ 2 into PLC␦ 1 confers G protein activation. Thus, the PH domain of PLC␤ 2 directly confers G␤␥ activation to the catalytic core. Analogously, it has been shown that the specific binding of PI(4,5)P 2 to the PH domain of PLC␦ 1 resulted in activation of the catalytic core (17). Taken together, these studies imply that the PH domain of PLC␤ 2 or PLC␦ 1 contacts a common site in the catalytic region that transmits activation when G␤␥ or PI(4,5)P 2 binds, respectively. One possible site may be a highly conserved site in the catalytic region that has been implicated in mediating G␤␥ activation of PLC␤ 2 (19).
In this study, we have constructed a chimeric protein consisting of the PH domain of PLC␦ 1 and the remaining regions of PLC␤ 2 , called PH␦PLC␤. We also constructed a chimera that includes the PH of PLC␦ 1 , a short (ϳ20 residue) linker sequence, and the first helix on the neighboring EF hand (PH L ␦PLC␤). The purpose of this latter construct was to test whether altering this linker and EF hand would affect the interaction between the PH domain and the catalytic core. We find that both proteins have the same basal activity as wild type PLC␤ 2 and displayed PI(4,5)P 2 -specific binding and acti-vation as PLC␦ 1 . As expected, the chimeric proteins are activated by PI(4,5)P 2 but not G␤␥, suggesting that activation of the catalytic cores of PLC␤ 2 and PLC␦ 1 are mediated by a common site in their PH domains. By using a peptide that we predicted should inhibit the interaction between the PH and catalytic domains based on theoretical models, we found that we could inhibit both the basal activity of both enzymes and eliminate both PI(4,5)P 2 and G␤␥ activation of the PLC catalytic cores in a concentration-dependent manner. Our results thus give a molecular basis for the common activation of these two enzymes.

MATERIALS AND METHODS
Protein Preparation-Synthetic peptides were prepared and purchased from the American Peptide Company (Sunnyvale, CA) and used as received. PLC␤ 2 and G␤␥ subunits were expressed in Sf9 cells using a baculovirus system and purified as previously described (see Ref. 12). PLC␦ 1 was expressed in bacteria as before (11).
Construction of Chimeras-Human PLC ␦ 1 was expressed in Escherichia coli and recombinant human PLC␤ 2 in Sf9 cells. Two chimeras were prepared. The first, PH␦PLC␤, replaced residues 1-134 of PLC␤2 with residues 1-117 of PLC ␦1. The second, PH L ␦PLC␤, included a 20-residue linker region and the first EF hand of PLC␦ 1 . The coding sequence of PH and PH-L domain of PLC ␦ 1 was amplified by PCR using corresponding PLC␦ 1 cDNA. (The primers were 5Ј-ATGGACTCGGGC-CGGGAC 3Ј and 3Ј-GCCTGAGTGGTGGACTAGTTTGTGCAGCCCCA-GC-5Ј or 3Ј-GCAGCTCCTGAA GCTGCTAGCTTGTCCTTGTTTTTGT-CAGC-5Ј.) The chimeras were constructed by introducing SpeI and NheI restriction enzyme sites into a pVL1392 vector containing the PLC␤ 2 insert. These were used along with NotI to remove the PH domain (NotI-NheI) and the PH-linker-EF portion (NotI-SpeI) of PLC-␤ 2 . The corresponding PH domain and PH-linker-EF region were amplified by polymerase chain reaction using PLC␦ 1 as a template. These were then ligated into the PLC␤ 2 vector within the NotI and NheI/SpeI sites to give the corresponding chimeric proteins. The chimera enzymes were expressed in Sf9 cells and purified using the protocol for PLC␤ 2 . The purified proteins were identified by Western blotting using a monoclonal antibody to the N terminus of PLC␦ 1 and a monoclonal antibody to the C-terminal region of PLC␤ 2 .
Hydrolysis of PI or PI(4,5)P 2 in Phospholipid Vesicles-Hydrolysis of [ 3 H]PI was carried out using small, unilamellar vesicles composed of PI:PC:PS (1:5:5) with increasing amounts of PI(4,5)P 2 as described in the text. We note that the total amount of PI in each assay (100 M) greatly exceeded the highest PI(4,5)P 2 concentration used for activation (5 M). To assay the PI-PLC activity, purified enzymes (0.45-15 nM) were incubated in an assay volume of 60 l containing 50 mM Hepes, pH 7.2, 160 mM KCl, 3 mM EGTA, 3 mM dithiothreitol, 1.5 mM CaCl 2 . The samples were incubated for 5 min at 37°C using [ 3 H]PI and 3 min using [ 3 H]PI(4,5)P 2 . The reaction was terminated by adding 0.2 ml of cold 10% trichloroacetic acid, followed by 0.1 ml of 1% bovine serum albumin. The aqueous and organic phases were separated by centrifugation, and 0.3 ml of the upper aqueous phase was counted by liquid scintillation. Using [ 3 H]PI as a substrate, we found that the specific activity of PLC␦ 1 is 51 times higher than PLC␤ 2 and 17 times higher than the chimeras.
Fluorescent Probes-All of the probes were purchased from Molecular Probes, Inc. Concentrated stocks of the probes were made in N,Ndimethylformamide and stored at Ϫ20°C under nitrogen in bottles wrapped with aluminum foil.
The proteins were labeled with the amine-reactive probe coumarin succinyl ester by initially raising the pH of the protein solution to 8.0 and adding a small aliquot of probe dissolved in N,N-dimethylformamide at a 4:1 probe:protein molar ratio. Unreacted probe was removed by extensive dialysis in 20 mM Hepes, 0.16 M KCl, 1 mM EGTA, and 1 mM dithiothreitol, pH 7.0, for either PLC proteins or G␤␥. G␣ q subunits labeled in the presence of GTP were dialyzed against 150 mM Hepes, 40 mM ␤-mercaptoethanol, 100 mM (NH 4 ) 2 SO 4 , 150 mM MgSO 4 , and 100 mM EDTA. 100 M GTP␥S was added immediately after dialysis. This labeling procedure consistent gives a protein:probe labeling ratio of ϳ1:1.
Fluorescence Measurements and Analysis-Fluorescence measurements were performed on an ISS spectrofluorometer (Champaign, IL) using 3-mm cuvettes. Coumarin-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. In the lipid titration curves, the background spectra of lipid alone was subtracted from each spectra along the titration curve. All of the spectra were corrected for the 10 -12% dilution that occurred during the titration.
Membrane binding was determined by following the change in fluorescence of a dilute solution (50 -100 nM) of labeled protein as freshly extruded large, unilamellar vesicles were added. The results were fit to a binding isotherm after correcting for dilution and background scattering. The reported partition coefficient refers to the lipid concentration at which amount of bound versus unbound protein (K p ϭ [P b ]/[P un ]) is identical.
We found that the emission intensity of coumarin-labeled G␤␥ showed a substantial and reproducible increase upon the addition of unlabeled PLC␤ 2 and gave values of apparent K d identical to those previously obtained using fluorescence resonance energy transfer (3,20). Most importantly, the titration curves showed the appropriate shift in midpoint when the initial concentration of coumarin-labeled G␤␥ was changed, showing that the changes in coumarin fluorescence reflect protein-protein associations. Therefore, in these studies we determined lateral association by the change in coumarin intensity rather than fluorescence resonance energy transfer so as to minimize the handling of these proteins.
We analyzed the titration curves assuming that all proteins can form 1:1 complexes. Because the protein association curves are conducted on membrane surfaces, the resulting bimolecular association curves yield apparent K d values that are dependent on the total available membrane surface area. We have previously translated these apparent values of K d to the dissociation constant that would be observed if the proteins were not membrane-bound (20). However, because all of the studies here were done at an identical lipid concentration, this analysis is not required. We have only reported the experimental values of the apparent K d to be used for comparative purposes.
Molecular Modeling of the PH and Catalytic Domains of PLC␤ 2 -Models of PH domain-XY domain and complexes of rat PLC␦ 1 were constructed using a protein docking program, Global Range Molecular Matching or GRAMM (21). We also docked models (obtained from 3D Jigsaw) of PLC␤ 2 PH domain to its catalytic region using high and low resolution methods.

Membrane Binding Properties of the PH␦PLC␤ Chimera-
Although the isolated PH domain of PLC␤ 2 binds strongly to lipid membranes (12,13,22), the contribution of other enzyme domains in membrane association is not clear (see Ref. 6), and so the chimeras in Fig. 1 may allow us to determine the contribution of these other PLC␤ 2 domains in membrane association. The PH domains of PLC␤ 2 and PLC␦ 1 have different binding characteristics. The PH domain of PLC␦ 1 binds strongly (i.e. K p ϭ ϳ50 -200 M) to membranes containing PI(4,5)P 2 because of specific interactions, and also to highly negatively charged lipids because of electrostatic attraction by the positive lobe of the protein (14,15). In contrast, the PH domain of PLC␤ 2 binds strongly to membranes with little specificity for charge or chemical structure (8,14). Thus, if the PH domain solely directs membrane association of the whole enzyme, we would expect the PH␦-PLC␤ chimeras to show identical characteristics to PLC␦ 1 .
We have previously found that the emission intensity of coumarin-labeled PLC␤ 2 and the isolated PH domains of PLC␤ 2 and PLC␦ 1 increase upon membrane binding (8). We thus measured the binding of the proteins to large, unilamellar vesicles of varying membrane compositions by the increase in fluorescence of the probe coumarin covalently attached to the proteins (see methods). In Fig. 2a we show the association of PLC␤ 2 , PLC␦ 1 , and the PH␦PLC␤ chimera to POPC bilayers. In accord with previous studies using both spectroscopic and sedimentation techniques, PLC␦ 1 does not bind to POPC bilayers in this concentration range, whereas PLC␤ 2 binds strongly (12)(13)(14)22). Unlike PLC␦ 1 , the PH␦PLC␤ chimera bound to POPC bilayers with an affinity strong enough to be measured by fluorescence, but 20-fold weaker than PLC␤ 2 . These results support the idea that regions other than the PH domain of PLC␤ 2 contribute to membrane binding (see Ref. 6).
When 5% PI(4,5)P 2 is incorporated into POPC bilayers, the membrane binding affinities of PLC␤ 2 and PLC␦ 1 become comparable in accord with previous work (12,14). In Fig. 2b we show that the presence of PI(4,5)P 2 increases the binding affinity of the PH␦PLC␤ chimera 20-fold, bringing its value (K p ϭ 20.6 Ϯ 2.8 M) close to that of PLC␤ 2 (K p ϭ 28.4 Ϯ 11.8 M) and PLC␦ 1 (K p ϭ 19.6 Ϯ 10 M). Thus, the characteristic PI(4,5)P 2 binding specificity of the PLC␦ 1 PH domain remains intact in the chimera. This increase in membrane binding affinity of PLC␦ 1 is not entirely because of electrostatic effects because the chimera and PLC␦ 1 bound much more weakly to membranes containing 30% negatively charged lipids (i.e. PC:PS (2:1)) (data not shown).
Binding of PLC Enzymes to G Protein Subunits-We have previously characterized the lateral association between PLC␤ 2 and G protein subunits (3) and between the isolated PH domains of PLC␤ 2 and PLC␦ 1 to G␤␥ subunits on membrane surfaces (8). We have found that even though PLC␦ 1 is not activated by G␤␥ subunits, it will bind to these subunits with a 4-fold weaker affinity than the G␤␥-activable PLC␤ 2 . The comparable G␤␥ affinities between the PH domain of PLC␤ 2 and the whole enzyme, and similarly between the PH domain of PLC␦ 1 and the whole enzyme, suggest that G␤␥ association is mediated entirely through the PH domain (8). If this is the case, then the G␤␥ binding affinity of the PH␦PLC␤ chimera should be closer to PLC␦ 1 than to PLC␤ 2 .
We measured the lateral association of the PLC proteins to G␤␥ subunits on membrane surfaces by labeling G␤␥ subunits with coumarin, reconstituting C-G␤␥ into lipid bilayers and adding the PLC proteins under conditions where the proteins would be entirely membrane-bound. Control studies substituted buffer for the titrating protein solution. Under our labeling conditions, the fluorescence intensity of the coumarin probe attached to G␤␥ is sensitive to PLC association, giving association constants identical to previous values obtained using fluorescence resonance energy transfer (3,8,20). Also, the midpoint of the titration curves shift appropriately over an 8-fold initial amount of G␤␥ to give identical values of K d , which is a key indicator of a protein-protein association process. The observation that the coumarin label changes upon association with PLC␤ 2 implies that the labeling site may be  3). b, increase in binding of the PH␦PLC␤ chimera (ƒ), relative to PLC␤ 2 (q) and PLC␦ (f) at 50 nM when 5% PI(4,5)P 2 is incorporated into the membranes (n ϭ 3).

FIG. 1. Schematic diagram of the chimeras made in this study.
The length of the EF hands of PLC␤ 2 is poorly defined. Therefore only the starting residue number is given.
close to the protein-protein interaction site, although the apparent K d does not appear to be affected (see Ref. 3). Because our studies compare the affinities between the PLC enzymes and G protein subunits, perturbation of the absolute affinities because of the probe is not critical.
We reconstituted C-G␤␥ subunits onto PC:PS (1:2) lipid membranes because this high concentration of negatively charged lipids ensures complete membrane binding of both PLC␤ 2 and the PH-PLC␦ 1 chimeras in the absence of PI(4,5)P 2 (data not shown). The studies were carried out at identical lipid concentrations so that apparent affinities can be directly compared. The results of this study are shown in Fig. 3. As can be seen, the G␤␥ affinity of the chimera is closer to PLC␦ 1 than to PLC␤ 2 . We find that the ratio of the apparent G␤␥ dissociation constant of the PH␦PLC␤ to wild type PLC␤ 2 is similar to the ratio of the PH domains of PLC␦ 1 and PLC␤ 2 and also to corresponding ratios of the whole enzymes (8). If other sites in PLC␤ 2 contributed to its association with G␤␥ subunits, we would expect these ratios to differ. Thus, these results are in accord with the idea that the PH domain is the main docking site for G␤␥ subunits and that other domains, if they play a role, do not significantly contribute to the binding energy.
Response of the Chimera to Activators-When the PH domain of PLC␤ 2 was swapped into PLC␦ 1 , the specific activity of this chimera was much lower than PLC␦ 1 and comparable with PLC␤ 2 . This result raised the question of whether the PH domain modulates the activity of the catalytic core. One factor that may contribute to the higher enzymatic activity of PLC␦ 1 as compared with PLC␤ 2 is idea that the specific binding of PI(4,5)P 2 to the PLC␦ 1 -PH domain induces activation of the catalytic core (see Ref. 17). If the PH domain of PLC␦ 1 interacted with the catalytic region PLC␤ 2 in a manner similar to that of wild type PLC␦ 1 , then we would expect the PH␦-PLC␤ 2 chimera to be activated by PI(4,5)P 2 and not by G␤␥ subunits.
To determine the relative activity of the PLC enzymes, we compared the specific activities of PLC␦ 1 , PLC␤ 2 , and PH␦PLC␤ by measuring the hydrolysis of [ 3 H]PI in small, unilamellar vesicles composed of PI:PC:PS (1:5:5) with increasing amounts of cold PI(4,5)P 2 . This substrate concentration has been previously used to show that PI(4,5)P 2 produces an 8 -9fold increase in PLC␦ 1 activity (17).
In the absence of PI(4,5)P 2 activator, we find that the specific activity of PH␦PLC␤ lies between that of PLC␦ 1 and PLC␤ 2 , giving a value only 17-fold less than PLC␦ 1 for the chimera as opposed to a 51-fold less value obtained for PLC␤ 2 relative to PLC␦ 1 (data not shown). Because the chimera contains the catalytic region of PLC␤ 2 , its higher activity toward [ 3 H]PI indicates that the PH␦ 1 has a different interaction with the PLC␤ 2 catalytic core than the wild type PH domain. This observation supports the idea that the PH domain serves to inhibit the core in the basal state in the absence of activators (9).
We then tested whether PI(4,5)P 2 could activate the enzymes. These studies were carried out by measuring [ 3 H]PI hydrolysis with increasing amounts cold PI(4,5)P 2 in the membrane substrate (see Ref. 17). The results, presented in Fig. 4, show that PI(4,5)P 2 activation is close to the level of activation seen in previous studies using PLC␦ 1 (17). PI(4,5)P 2 activation of the PH␦PLC␤ chimera is only slightly lower than PLC␦ 1 , whereas activation of PLC␤ 2 does not significantly occur. These FIG. 6. Sequence comparison of the PH domain residues of PLC␦ 1 and PLC␤ 2 in which the regions proposed to interact with their corresponding catalytic region are given in italics and the sequence of the peptide used in this study is underlined. Identical residues are denoted by asterisks, highly homologous residues are denoted by colons, and homologous residues are denoted by dots.

FIG. 7. Change in the normalized hydrolysis of [ 3 H]PI(4,5)P 2 of PLC enzymes with increasing amounts of the PLC␦ 1 (84 -95) peptide whose sequence is suggested to be at the interface between the PH and catalytic domains where PLC␤ 2 (black bars), PH␤PLC␦ (light gray bars), and PLC␦ 1 (dark gray bars) all show a decrease in activity with increasing peptide concentration.
studies show that the PH domain of PLC␦ 1 regulates PI(4,5)P 2 activation to the catalytic cores of both PLC␤ 2 as well as the wild type enzyme.
The proteins were then tested for their ability to be activated by G protein subunits. Because PH␦PLC␤ displayed an ϳ3-fold weaker binding affinity to G␤␥ subunits than PLC␤ 2 , activation studies were carried out to high concentrations of G␤␥ to ensure complete protein-protein association. Our previous study revealed that the catalytic core of PLC␦ 1 could be activated by G␤␥ when the PH domain of PLC␤ 2 was linked to PLC␦ 1 (9). Based on this study, we predicted that because the PH␦PLC␤ chimera contains a PH domain from a PLC that is not activated by G␤␥ subunits, no changes in activity will be seen. We could not detect activation of the chimera even at concentrations well above the apparent K d for G␤␥ binding (data not shown).
Identification of a Common Site Involved in PLC-␤ 2 and -␦ 1 Activation-The ability of the PH domain of PLC␦ 1 to confer activation to the catalytic domain of PLC␤ 2 and vice versa suggests a common interaction site between the catalytic and PH domains. Although the catalytic domain of the two phospholipases are highly conserved having 50% sequence identity and 82% sequence homology, their PH domains have only 15% sequence identity and 56% sequence homology (see Ref. 12). Nevertheless, we have previously made threading models of the PH domain PLC␤ 2 based on the known PH␦ 1 structure and obtained a predicted 82% structural identity (23). The high sequence homology between the catalytic domains of the two enzymes allowed us to easily make structural models of the catalytic core of PLC␤ 2 . 2 We used these models and the known structures of the PH and catalytic domains of PLC␦ 1 and docked the domains to identify potential interaction sites.
The docking procedure was carried out using GRAMM, which performs an exhaustive six-dimensional search for low intermolecular energy states of the structures by changing the atom-atom potentials of the inputted atomic coordinates of the two molecules (21). We carried out the docking procedure using GRAMM at both high and low resolution because the latter suppresses local minima (false positive fits), giving more appropriate candidates. This procedure resulted in 50 PLC␦ 1 models. Approximately 75% of these were similar and gave fairly consistent structures, and these were used for further investigation. Because the structures of the PH and catalytic domains of PLC␦ 1 are known, we present the results for this enzyme. We discarded all but one model based on the criterion that the PI(4,5)P 2binding site in the PH and catalytic domains of PLC␦ 1 must have the same orientation. This model was docked at high resolution and gives all configurations with a high score stearic fit and is shown in Fig. 5 (a and b), and the homologous model of PLC␤ 2 is shown in Fig 5c. The models in Fig. 5 predict three common regions of the PH domains of PLC␦ 1 and PLC␤ 2 that interact with a common site in the catalytic core as presented in Fig. 6. To test the model in Fig. 5, we synthesized a peptide having the PH domain sequence corresponding to residues 84 -95 of PLC␦ 1 , which lies very close to the catalytic region and has high homology to the corresponding region of PLC␤ 2 . We then determined the effect of this peptide on the basal and G␤␥/ PI(4,5)P 2 -stimulated activity of the PLC enzymes relative to a control peptide with a unhomologous sequence. If the PH domain peptide does mimic the common interaction site between the PH and catalytic domains of PLC␤ 2 and PLC␦ 1 , then it should produce similar changes in activity in both the PLC enzymes and their chimeras.
We monitored the change in PI(4,5)P 2 hydrolysis of PLC␤ 2 , PLC␦ 1 , and a PH␤ 2 -PLC␦ 1 chimera that can be activated by G␤␥ subunits (9). The addition of increasing amounts of the PLC␦ 1 (84 -95) peptide to each of these proteins causes a pro-2 F. Philip and S. Scarlata, unpublished results.

FIG. 8. Effect of the PH-PLC␦ 1 (84 -95) residues on the G␤␥ activation of 4 nM PLC␤ 2 (light gray bars) and 4 nM PH␤PLC␦ (black bars) chimera at 25 nM G␤␥, which gives 1.5-fold activation of both enzymes, and 200 nM G␤␥, which gives 4-fold activation.
gressive reduction in activity (Fig. 7), whereas the control peptides had no affect on the activity. We note that this peptide does not interact with lipid bilayers under our assay conditions. A reduction in activity of the PLC enzymes would be expected if catalysis involves productive interaction between the PH and the catalytic core.
We then focused on activation of the PH-PLC␤ 2 -containing enzymes by G␤␥ by measuring the ability of the PLC␦ 1 (84 -95) peptide to inhibit G␤␥ activation of wild type PLC␤ 2 and the G␤␥-activable chimera PH␤PLC␦. At partially activating concentrations of G␤␥ (i.e. 25 nM G␤␥, which produces a 1.5-fold increase in activation), addition of the peptide could completely abolish G␤␥ activation of both PLC␤ 2 and the PH␤PLC␦ chimera (Fig. 8, left panel). However, at higher G␤␥ concentrations (i.e. 200 nM, which produces a 4-fold increase in activation), the loss of activation by the peptide was greatly reduced (Fig 8, right panel). Thus, raising the G␤␥ concentration eliminates the effect of the peptide.
To determine whether the PLC␦ 1 (84 -95) peptide reduced G␤␥ activation by competing with G␤␥ with the interaction site on the PH domain of PLC␤ 2 , we carried out a series of fluorescence titrations where we measured the association between PLC␤ 2 and G␤␥ subunits at increasing levels of peptide. We find that not only that the peptide binds to isolated G␤␥ and PLC␤ 2 with apparent K d of 40 and 50 nM, respectively (data not shown), but also that the peptide inhibits the association between the two proteins, indicating that it mimics both the interaction site between the PH domain and the catalytic core and between the PH domain and G␤␥ subunits (Fig. 9).
We then determined whether the peptide could specifically inhibit the PH domain-mediated activation of PLC␦ 1 by PI(4,5)P 2 . We found that the addition of 1 M peptide reduces the PI(4,5)P 2 activation by ϳ50% (Fig. 10). Taken together, these results show that a peptide that targets a homologous interaction region between the PH and catalytic regions of PLC␦ 1 can inhibit the basal and stimulated activities of both this enzyme and PLC␤ 2 .

DISCUSSION
In this study, we have shown that we can transfer the activators of the catalytic core of PLC␤ 2 and PLC␦ 1 by transferring their PH domains. Because this activation process must involve a common interface between the PH and catalytic domains in the two enzymes, we constructed structural models that enabled us to identify this interface and tested these using a synthetic peptide inhibitor. Our results also suggest that the interface between the PH and catalytic domains is the region involved in the docking of G␤␥ subunits to PLC␤ 2 .
We first constructed two chimeras that consist of the PH domain of PLC␦ 1 and the remainder of PLC␤ 2 and of the PH-EF domains of PLC␦ 1 and the remainder of PLC␤ 2 . The resulting proteins had electrophoretic mobilities on SDS-PAGE gels matching that of PLC␤ 2 . Western blot analysis showed that the chimeras contain monoclonal antibody epitopes for the N-terminal region of PLC␦ 1 and the C-terminal region of PLC␤ 2 , thus verifying that the proteins are true chimeras. The chimeras showed the same PI(4,5)P 2 -dependent membrane binding behavior as PLC␦ 1 and bound to G␤␥ subunits with identical affinities as PLC␦ 1 , showing that, functionally, the N-terminal region of the protein is correctly folded. The specific activity of the chimeras fell between the high specific activity of PLC␦ 1 and PLC␤ 2 , suggesting that the catalytic core of PLC␤ 2 functions at an enhanced level when the PH domain of PLC␦ 1 is present. We also note that the chimeras bound to unactivated G␣ q with an affinity identical to PLC␤ 2 , strongly suggesting that the C2 and C-terminal extensions are correctly folded (data not shown).
Although the PH␦PLC␤ construct was designed to better define the role of the PH domain of PLC␤ 2 in membrane binding and G protein activation, we also constructed a chimera in which the PH domain, linker, and first EF hand of PLC␦ 1 were swapped into PLC␤ 2 . There is 25% sequence homology of this first EF hand in the two PLCs. Like the N-terminal PH domain, this first EF hand was unresolved in the crystal structure of PLC␦ 1 ; however the structure of the isolated PH domain was later solved (24,25). Although the functional role of the EF hands in PLC␦ 1 are unknown, it has been suggested that the second pair of EF hands interact with the C2 domain to give structural integrity to the protein core because deletion of this region inactivates the enzyme (see Ref. 24). In all of the studies presented here, we found no significant difference in membrane binding, G␤␥ binding, and PI(4,5)P 2 activation between the two chimeras, suggesting that the EF hand does not directly modulate that catalytic core of the protein. The observation that both chimeras have similar activities supports the notion that the gross structural features imparted by this region may assist in enzyme integrity, but the sequence is not important.
In our membrane binding studies, we found that the PH domain of PLC␦ 1 imparts PI(4,5)P 2 -specific membrane binding to the remainder of PLC␤ 2 . Previous studies have shown that the isolated PH domains of PLC␦ 1 and PLC␤ 2 bind to membranes with affinities and specificities similar to the whole enzymes (8). However, studies using purified PLC␤s and model membranes suggest that deletion of the C-terminal extension of PLC␤ 2 weakens membrane affinity, indicating that this region may also interact with membranes (12,13). The observation that the PH domain of PLC␦ 1 does not bind to POPC, whereas the PH␦PLC␤ chimera displays weak binding, shows that the PH domain is a key regulator of PLC␤ 2 membrane association, whereas the C-terminal region works to enhance this association.
We have found that the PH␦PLC␤ chimera can be activated by the presence of PI(4,5)P 2 to a similar extent as PLC␦ 1 . It has been previously shown that binding of PI(4,5)P 2 to the PLC␦ 1 PH domain can confer activation to the catalytic domain (17). Using a PH␤PLC␦ chimera, an analogous model was proposed for the activation of PLC␤ 2 by G␤␥ in which the binding of G␤␥ to the PH domain conferred activation of the catalytic core (9). The observation that the PH domain of PLC␦ 1 could not confer G␤␥ activation of the PLC␤ 2 catalytic core even under conditions where the two proteins are associated correlates well with the model that the PH domain of PLC␤ 2 mediates G␤␥ activation (9). Thus, the PH domains play a similar role in activation of their host core proteins through the binding of an activator. If we consider that the level of PI(4,5)P 2 activation of the chimera is similar to that seen for PLC␤ 2 by G␤␥ subunits, then it appears that each activator stimulates the catalytic core of PLC␤ 2 to a similar extent, which is most likely an intrinsic property of the catalytic site.
The finding that the PH domains of either enzyme can confer activation to the catalytic core suggests a conserved interface between these domains. Molecular modeling and protein docking studies allowed us to identify several potential models of the domain complex. We selected the model in Fig. 5 based on the assumption that the PI(4,5)P 2 -binding sites in PH-PLC␦ 1 and the catalytic domain must be on the same side of the protein to give the proper membrane orientation. This model identifies a PLC␦ 1 -PH domain interface that has high homology to that of PLC␤ 2 and correlates well to the most reasonable docking model of the PH and catalytic domains of PLC␤ 2 . The model in Fig. 5 is in accord with our current knowledge of PH domain-induced activation and activation by G␤␥ subunits. The model shows that the PH domain may be located close to the catalytic core, and it is reasonable to assume that subtle changes in this interface through docking of activators or binding to membrane surfaces could directly affect catalytic activity. The peptide made to this interface (i.e. residues 84 -95 of PLC␦ 1 ) not only inhibited the basal activity of both the PLC␤ 2 and PLC␦ 1 enzymes and inhibited PI(4,5)P 2 activation of PLC␦ 1 , which would be expected if both proteins shared this common site, but it also competed with G␤␥ binding and inhibited G␤␥ activation for the PH-PLC␤ 2 enzymes. This result suggests that G␤␥ subunits bind to this region and confer activation.
Previous studies also based on peptide competition suggested that G␤␥ activation of PLC␤ 2 proceeds through a conserved region in the catalytic domain, which is homologous to residues ϳ516 -541 in PLC␦ 1 (19). The homologous region in PLC␦ 1 lies in close proximity to the PH-catalytic domain interface and can easily contact bound G␤␥ subunits. The high homology between PLC␤ 2 and PLC␦ 1 in the catalytic domain would allow G␤␥ to contact both enzymes as long as the PH domain of PLC␤ 2 is present. Because the strength of G␤␥ association to the PH domain of PLC␦ 1 is only ϳ4-fold lower than PLC␤ 2 and because the PH-PLC␦ 1 domain cannot confer activation, it is likely that the regions outside the interface cannot make the productive contacts needed for G␤␥ activation. Taken together, our studies suggest that G␤␥ subunits initially dock to the PH domain of PLC␤ 2 . The docking site lies in or close to the PH and catalytic domain interface where it can then make the necessary contacts with the catalytic regions, and most likely the region encompasses residues 564 -593 to induce activation (19).