Structural Determinants for Phosphatidic Acid Regulation of Phospholipase C-β1*

Signaling from G protein-coupled receptors to phospholipase C-β (PLC-β) is regulated by coordinate interactions among multiple intracellular signaling molecules. Phosphatidic acid (PA), a signaling phospholipid, binds to and stimulates PLC-β1 through a mechanism that requires the PLC-β1 C-terminal domain. PA also modulates Gαq stimulation of PLC-β1. These data suggest that PA may have a key role in the regulation of PLC-β1 signaling in cells. The present studies addressed the structural requirements and the mechanism for PA regulation of PLC-β1. We used a combination of enzymatic assays, PA-binding assays, and circular dichroism spectroscopy to evaluate the interaction of PA with wild-type and mutant PLC-β1 proteins and with fragments of the Gαq binding domain. The results identify a region that includes the αA helix and flexible loop of the Gαq-binding domain as necessary for PA regulation. A mutant PLC-β1 with multiple alanine/glycine replacements for residues 944LIKEHTTKYNEIQN957 was markedly impaired in PA regulation. The high affinity and low affinity component of PA stimulation was reduced 70% and PA binding was reduced 45% in this mutant. Relative PLC stimulation by PA increased with PLC-β1 concentration in a manner suggesting cooperative binding to PA. Similar concentration dependence was observed in the PLC-β1 mutant. These data are consistent with a model for PA regulation of PLC-β1 that involves cooperative interactions, probably PLC homodimerization, that require the flexible loop region, as is consistent with the dimeric structure of the Gαq-binding domain. PA regulation of PLC-β1 requires unique residues that are not required for Gαq stimulation or GTPase-activating protein activity.

Phosphatidic acid (PA) 2 is a novel signaling phospholipid that has been implicated in the regulation of multiple cell func-tions including cell proliferation, cytoskeletal reorganization and membrane trafficking (1,2). PA regulates the activity of signaling proteins in vitro, including phospholipases (3)(4)(5), kinases (6,7), G protein regulators (8), and cyclic AMP-phosphodiesterases (9,10). The importance of PA in the regulation of cellular effectors is shown by in vivo studies that demonstrate loss of cellular regulation in cells overexpressing mutant PA-impaired effectors (6,7,9,10).
Intracellular PA levels are low under resting conditions but increase rapidly upon activation of a number of receptor signaling pathways, including GPCR-PLC-␤ signaling (11). GPCR-PLC-␤ signaling is mediated both by the G q/11 family of GTPbinding proteins acting through the G␣ subunit and by G␤␥ subunits supplied primarily by G i family members. Rapid hydrolysis of phosphatidylinositol 4,5-bisphosphate (PIP 2 ) by PLC-␤ generates inositol 1,4,5-trisphosphate and diacylglycerol (DAG). Inositol 1,4,5-trisphosphate increases the levels of regulatory Ca 2ϩ . DAG activates the protein kinase Cs and other effectors, including DAG kinases, which convert DAG to PA. Increases in intracellular Ca 2ϩ levels and protein kinase C activity stimulate phospholipase D (PLD) activity, which generates PA from phosphatidylcholine. PLD is also stimulated in response to GPCR-G 12/13 signaling and activation of monomeric GTP-binding proteins, including RhoA. Nine DAG kinase isoforms and two major PLD isoforms have been identified; they differ in subcellular localization and regulation (12,13). The contribution of the DAG kinase-and PLD-generated pools of PA in the GPCR regulation of cellular function, and the mechanism for PA regulation of effectors remains poorly understood.
In addition to the role of PLC-␤ 1 in mediating GPCR-dependent increases in PA levels, PLC-␤ 1 is also regulated by PA in vitro. These data suggest a potential role for PA in the feedback modulation of GPCR-PLC-␤ 1 signaling (14 -16). PLC-␤ is a member of a large family of PIP 2 -specific PLC enzymes that include PLC-␤ 1-4 , PLC-␥ 1-2 , PLC-␦ 1-4 , PLC-⑀, PLC-, and PLC-1-2 (17)(18)(19)(20)(21). PIP 2 -PLCs are grouped by sequence homology and coupling to specific receptor-delineated mechanisms, including those mediated by GPCR and tyrosine kinase receptors. The mechanisms that regulate many of these enzymes are still not well understood, and both specific signaling phospholipids and small molecular weight GTP-binding proteins have recently been implicated in regulation. PLC-␤ enzymes share many of the structural features present in other members of the PIP 2 -PLC family, including conserved catalytic X and Y boxes as well as two membrane-phospholipid binding regions, the pleckstrin homology and C2 domains. PLC-␤ enzymes, however, are distinguished by the presence of an elongated C terminus consisting of ϳ450 residues, which contains many of the determinants for interaction with G␣ q (22).
GPCR stimulation of PLC-␤ 1 activity is tightly regulated at the G protein level by cycling of G␣ q between inactive and active states (23). Under basal conditions, G␣ q exists in the relatively inactive GDP-bound state. Agonist binding to GPCR activates G␣ q by promoting exchange of GTP for bound GDP. The active G␣ q -GTP stimulates PLC-␤ 1 activity. The duration of the G␣ q -GTP state is determined by its intrinsic GTPase activity, which hydrolyzes G␣ q -bound GTP to GDP and thereby terminates activity. GTPase-activating proteins (GAPs), including regulators of G protein signaling and PLC-␤, stimulate the intrinsic GTPase activity of G q , accelerating deactivation (24). PLC-␤ thus functions as both an effector of G␣ q and a GAP specific for the G␣ q family of G proteins (24). The intracellular mechanisms that regulate and integrate these dual functions of PLC-␤ are not fully understood.
We have shown that PA stimulates PLC-␤ 1 activity through a mechanism that requires PA binding (14,16). Unlike other PAregulated effectors, PA also modulates PLC-␤ 1 activation in response to known cellular mediators, suggesting that PA targets a mechanism common to both basal and regulated PLC-␤ 1 activity. PA stimulation of PLC-␤ 1 activity is synergistic with G␣ q stimulation in vitro (16). The combination of PA and G␣ q -GTP␥S results in greater stimulation of PLC-␤ 1 activity than that due to either PA or G␣ q -GTP␥S alone. PA also potentiates GPCR stimulation of PLC-␤ 1 activity in membranes, consistent with a role for PA in enhancing receptor-G␣ q stimulation (14,15). Furthermore, PA antagonizes net inhibition of PLC-␤ 1 activity by protein kinase C␣, suggesting a potential role for PA in the regulation of negative feedback inhibition by protein kinase C␣ (16).
Two observations suggest that PA regulation occurs through binding to a site within the PLC-␤ 1 C terminus and that it is isoform-dependent. First, a calpain-generated 100-kDa PLC-␤ 1 fragment, which lacks 336 residues C-terminal to His 880 , does not bind PA and is insensitive to stimulation by 7-15 mol % PA. Stimulation of the 100-kDa fragment requires Ͼ15 mol % PA and is not coupled to measurable PA binding (14). Second, PA stimulation of the PLC-␤ 3 isoform requires Ͼ15 mol % PA and is also not associated with PA binding (16).
Sequence diversity within the C-terminal domain of PLC-␤ isoforms is thought to contribute to isoform differences in sensitivity to stimulation by G protein subunits and regulation by protein kinases (21). The C-terminal domain contains determinants for ␣ q stimulation (25,26), GAP activity (27), electrostatic dependent association with membrane lipids (25,26), nuclear localization (26), and phosphorylation/regulation by protein kinase C␣ (28,29). Isolated PLC-␤ 1 C-terminal tails show complex GAP behavior, which suggests that dimerization/oligomerization has an important role in the regulation of PLC-␤ 1 function (27). Consistent with this hypothesis, the crystal structure of the turkey PLC-␤ 2 C-terminal domain shows it to be a homodimer of three-stranded coiled coils in an antiparallel orientation (30). The dimer interface includes the ␣B and ␣C helices, and G␣ q is predicted to bind to the dimer interface, further implicating dimerization in G␣ q interaction (31).
To further delineate the role of PA in the modulation of GPCR-G␣ q -PLC-␤ 1 signaling, the present studies used truncated PLC-␤ 1 C-terminal fragments and corresponding PLC-␤ 1 C-terminal mutants (31) to localize the determinants for PA binding and to address the mechanism for PA regulation of PLC-␤ 1 activity. The data show that the region mapping to the ␣A helix and flexible ␣A-␣B loop of the G␣ q binding domain is required for both PA binding and stimulation of PLC-␤ 1 activity. This loop is conserved among mammalian PLC-␤ 1 s but diverges in the other isoforms. In addition, our data suggest a mechanism for PA regulation of PLC-␤ 1 that involves cooperative interactions, probably homodimerization. Regulated homodimerization is a potential novel mechanism for the integration of signaling from multiple cellular networks to the regulation of GPCR-G␣ q -PLC-␤ 1 signaling.
Binding Assay-Binding to phospholipids was performed essentially as described (14) except that large multilamellar vesicles (liposomes) were used that can be efficiently sedimented by centrifugation at 40,000 ϫ g for 60 min at 5°C (34,35). PLC fragments or PLC-␤ 1 were added to a 100-l assay mixture (liposomes consisting of 400 M 100 mol % PC or 25 mol % PA/75 mol % PC or, as indicated, 30 mM HEPES (pH 7.0), 5 g of fatty acid-free bovine serum albumin plus 0.1 M KCl or, as indicated, final concentration). The lipid-bound protein was separated from free protein by centrifugation, the pellet was resuspended in 25 mM Tris (pH 7.0) diluted with 4ϫ SDS-PAGE sample buffer, and proteins were resolved by electrophoresis on 10 -20% SDS-PAGE (for C-terminal fragments) or 7.5% SDS-PAGE (for PLC-(1-880) and PLC-␤ 1 ). Load and internal protein concentration standards were included in each gel. PLC fragments were detected by Coomassie Blue staining of the gel. PLC-␤ 1 was detected by Coomassie Blue staining or Western blot using anti-PLC-␤ 1 antibody (Santa Cruz Biotechnology, Inc., Santa Cruz, CA), as indicated. Protein bands were quantitated by ScionImage.
PLC Assay-PLC activity was measured as described (14). PLC-␤ 1 was incubated in 50 l of buffer (0.5 mM MgCl 2 , 100 mM KCl, 50 mM HEPES (pH 7.0)) and 150 M phospholipid vesicles consisting of 50 M [ 3 H]PIP 2 in combination with PC and PA to achieve the appropriate mol % PA. Vesicles were prepared by drying the lipids under a stream of nitrogen followed by the addition of 50 mM HEPES buffer (pH 7.0) and sonication. Free Ca 2ϩ concentration was maintained at 33 nM or as indicated with a Ca-EGTA buffer. Incubation was at 30°C for 10 -20 min.
Other Materials-[ 3 H]PIP 2 was from PerkinElmer Life Sciences. PIP 2 was from Sigma. Brain PC and dipalmitoyl phosphatidic acid, from Sigma or Avanti Polar Lipids, were used in the binding and PLC studies. DPPC, DPPA, DPPC, and Di6-PA were from Avanti Polar Lipids. Calpain-1 was from EMD (Calbiochem).
Statistical Analysis-Paired Student's t tests were performed as indicated. A value of p Ͻ 0.05 was considered significant.

RESULTS
Localization of PA Binding Determinants in PLC-␤ 1 C-terminal Fragments-Calpain cleaves PLC-␤ 1 after His 880 to generate a 100-kDa PLC-␤ 1 fragment, PLC-(1-880), that lacks the 336 C-terminal residues (14,32). The activity of PLC-(1-880) is not stimulated by low mol % PA, and PLC-(1-880) does not bind PA, indicating that high affinity PA regulation requires the PLC-␤ 1 C terminus (14). The binding of full-length PLC-␤ 1 to PA is salt-resistant, consistent with binding mechanisms other than electrostatic (16). To directly address the involvement of the PLC-␤ 1 C terminus in salt-resistant PA binding, we measured PA binding to a purified recombinant PLC C-terminal fragment corresponding to residues 878 -1141 of PLC-␤ 1 , PLC-(878 -1141), under highly stringent conditions that included 0.4 M KCl. As shown in Fig. 1, the phospholipid binding profile of PLC-(878 -1141) was comparable with that of full-length PLC-␤ 1 . Binding of both PLC-(878 -1141) and fulllength PLC-␤ 1 to liposomes that contain PA was resistant to displacement by 0.4 M KCl, indicating that residues 878 -1141 contain determinants for salt-resistant PA binding. In contrast, binding of either protein to PS-containing liposomes was inhibited 100% by 0.4 M KCl, showing that PS binds through a predominantly electrostatic mechanism. Less than 5% of either protein bound to PC liposomes.
To further localize the determinants for salt-resistant PA binding, we studied the PA binding of PLC-(1-880) and a series of PLC C-terminal fragments corresponding to residues 878 -1030, 878 -1010, 878 -980, and 878 -950 (Fig. 2). Neither PLC-(1-880) nor PLC C-terminal fragments bound appreciably to PC liposomes, less than 5% binding. The binding profile for PLC-(1-880) and PLC C-terminal fragments to PA is shown in Table 1 1 and 4), 25 mol % PA/75 mol % PC (lanes 2 and 5), or 25 mol % PS/75 mol % PC (lanes 3 and 6) in the presence of either 0.1 M KCl (lanes 1-3) or 0.4 M KCl (lanes 4 -6). Binding was determined as described under "Experimental Procedures." Gels were stained with Coomassie Blue. PA binding to the PLC-␤ 1 C-terminal domain was also demonstrated by measurement of PA-promoted changes in secondary structure using far UV CD. Phospholipid binding frequently alters the secondary structure of proteins, and this effect provides a complementary measure of binding. DPPC and corresponding phospholipid analogs produce less light scattering than other lipids (36) and were therefore used. As shown in Fig.  3A, the secondary structure of PLC-(878 -980) was mostly random coil in the presence of DPPC vesicles, as is evident from the single minimum at ϳ200 nm. DPPA at 25 mol % induced a significant increase in secondary structure, as shown by a pair of minima at ϳ222 and ϳ208 nm and a maximum near 190 nm. The increase in ␣ helical content was saturable, with half-maximum at 15 mol % and maximum at ϳ25 mol % DPPA (Fig. 3B). In contrast to PA, 25 mol % DPPS had little effect on secondary structure, showing that structural changes were specific for DPPA. Binding of PLC-(878 -980) to DPPA vesicles was verified by sedimentation of the reaction mixture and quantitation of protein associated with the phospholipid pellet. There was little binding of PLC-(878 -980) to DPPC vesicles, but 90% of the PLC-(878 -980) bound to the DPPA-containing vesicles. DPPS promoted binding by only 30%. Electrostatic binding of the sort driven by PS is thus not sufficient to cause a structural change in this region.
Localization of Determinants for PA Regulation of PLC-␤ 1 -Residues 878 -980 in PLC-␤ 1 , which confer salt-resistant PA binding, include part of the flexible linker that connects the C2 and the G␣ q -binding GAP domains, the first ␣ helix of the GAP domain, and part of the following loop. We further localized the structures in this region that are important for PA binding and stimulation of PLC activity using a series of previously characterized mutants with runs of multiple alanine/glycine replacements in or adjacent to the ␣A-␣B loop (31). As shown in Table  2, PA binding to the mutant A-(944 -957) was reduced by 45% in the presence of 0.3 M KCl and statistically significant. Binding to the A-(944 -957) mutant was equivalent to wild-type PLC-␤ 1 in 0.1 M KCl, suggesting that electrostatic binding mediates a greater proportion of total A-(944 -957) binding to PA relative to wild-type PLC-␤ 1 (supplemental Fig. 1). Binding to mutant A-(965-975) was marginally reduced, and binding to the other mutants was unaltered. The low affinity, salt-sensitive binding was not altered in any of the mutants (data not shown). The sequences of the alanine/glycine PLC-␤ 1 mutants are summarized (supplemental Table 1).
Stimulation of the phospholipase activity of the mutants by PA paralleled their PA binding. PA stimulation of the A-(944 -957) and A-(965-975) mutants was reduced 70 and 30% relative to wild-type PLC-␤ 1 , respectively. Again, the decrease in stimulation was statistically significant only for the A-(944 -957) mutant. PA stimulation of the other mutants was similar to that of wild-type PLC-␤ 1 at about 200% of basal level. Interaction with fatty acyl chains appears to contribute appreciably to A-(944 -957) binding to PA. When binding was carried out with the short chain PA analog, Di6-PA, binding of A-(944 -957) to PA liposomes was 1.9 Ϯ 2 fmol as compared with 17.2 Ϯ 2 fmol for wild-type  PLC-␤ 1 (n ϭ 3), a 90% reduction in mutant PA binding. The greater contribution of nonspecific electrostatic binding and fatty acyl interaction in A-(944 -957) relative to wild-type PLC-␤ 1 probably underestimates the actual decrease in PA binding. Together, these data identify one mutant with a significant impairment in PA regulation and implicate residues 944 -957 in the mechanism for PA binding and stimulation of phospholipase C activity. The A-(944 -957) mutation, which replaces LIKEHTTKYNEIQN with AGAG-AGAGAGAGAG, was characterized further. Phospholipase Activation by PA Depends on the Concentration of PLC-␤ 1 -The relative stimulation of phospholipase activity by PA depends on the concentration of PLC-␤ 1 . As shown in Fig. 4, the specific activity of PLC-␤ 1 increased markedly with increasing enzyme concentration up to about 50 pM in either the presence or absence of PA. For wild-type PLC-␤ 1 , specific activity increased about 4-fold over the range of enzyme concentrations shown. A qualitatively similar increase was observed for the A-(944 -957) mutant. Such behavior suggests that mostly inactive PLC-␤ 1 monomers associate to form a more active dimer or higher oligomer. In the presence of PA, increased specific activity occurred at lower PLC-␤ 1 concentrations than in its absence, which suggests that PA binding promotes dimerization.
The interactive effects of PA and PLC-␤ 1 on self-association are more evident when the PA concentration dependence is examined at low and high PLC-␤ 1 concentrations (Fig. 5). PA stimulated dilute wild-type PLC-␤ 1 more than the concentrated enzyme. Maximal stimulation of 50 pM PLC-␤ 1 was about 8-fold, whereas that of 150 pM PLC-␤ 1 was less than 3-fold. Further, 50 pM PLC-␤ 1 was half-maximally stimulated at about 0.4 mol % PA, but stimulation of 150 pM PLC-␤ 1 did not obviously saturate even at 15 mol %, reflecting a greater contribution of ionic and nonspecific effects. It is therefore likely that at low PLC-␤ 1 concentrations, PA stimulation occurs through a high affinity interaction and that only low affinity interactions occur at higher PLC-␤ 1 concentrations. Stimulation by PA was nearly absent in the A-(944 -957) mutant.
Taken together, the data of Figs. 4 and 5 suggest that high affinity PA binding activates PLC-␤ 1 primarily at low enzyme concentrations and that activation by oligomerization sup-plants the PA effect. Conversely, PA decreases the concentrations of PLC-␤ 1 needed for the display of increased activity. At higher enzyme concentrations, high affinity activation by PA is lost, and the behavior of the A-(944 -957) mutant argues that this region of the C-terminal domain is involved in this interaction. The data support the idea that PA stimulates phospholipase activity by promoting its association to form an active oligomer, probably a dimer.

DISCUSSION
Previous work from this laboratory has shown that PA regulates PLC-␤ 1 activity through a mechanism that requires PA binding to the C-terminal domain that includes

binding and stimulation of PLC activity in PLC-␤ 1 mutants
Wild-type PLC-␤ 1 and mutants, purified by chromatography on Ni 2ϩ -nitrilotriacetic acid-agarose, were assayed for PA binding and PA stimulation of PLC activity. PA binding was determined with 0.3 nM PLC-␤ 1 and either 25 mol % PA/75 mol % PC or 100 mol % PC liposomes in buffer containing 0.3 M KCl. Binding to PC vesicles was negligible and subtracted from the total binding. PLC activity was determined with 0.3 nM PLC-␤ 1 at 33 nM Ca 2ϩ and 15 mol % PA/85 mol % PC. Basal PLC activity for wild-type PLC-␤ 1 , A-(928 -943), A-(944 -957), A-(965-975), A-(976 -990), A- (998 -1007), and A-(1018 -1028) was 1.3, 1.6, 1.9, 1.4, 1.3, 1.4, and 1  the G␣ q binding site (14,16). The observation that PA modulates PLC-␤ 1 regulation by known cellular regulators suggests that PA constitutes part of an important isoform-specific feedback mechanism for regulation of GPCR-PLC-␤ 1 signaling (16). Further insight into the role and mechanism for PA regulation requires identification of the PA binding region. The present studies used PA binding, CD spectroscopy, and PLC activity measurements and a combination of PLC C-terminal fragments and full-length PLC-␤ mutants to delineate the region required for interaction with PA. All of the experimental data localized the PA interaction site to the ␣A helix and the flexible ␣A-␣B loop of the G␣ q binding domain. Most notably, mutation of residues 944 -957, which includes parts of ␣A and the loop, caused a significant 45% reduction in PA binding and 70% reduction in PA stimulation. Mutation of residues 965-975, within the loop, also marginally decreased PA binding and PA stimulation but was not statistically significant. Other mutations in the C-terminal domain, A-(928 -943), A-(976 -990), A-(998 -1007), and A-(1018 -1028), displayed wild-type or nearly wild-type regulation by PA. Mutants A-(944 -957) and A-(965-975) have been characterized previously (31) ( Table  3). They show wild-type behavior with respect to Ca 2ϩ -stimulated phospholipase C activity and wild-type G q GAP activ-ity. The A-(944 -957) mutant shows only a minor decrease in stimulation by G␣ q , and stimulation of A-(965-975) mutant is unaltered. Most of the mutations that decrease the response to G␣ q or G␣ q GAP activity map to the interface of the C-terminal dimer that occurs between the ␣B and ␣C helices or to a proposed G␣ q -binding site (30,31). The present data thus show that interaction with PA is mediated by residues within the G␣ q -binding domain that largely do not overlap with those required for G␣ q binding or regulation.
The flexible ␣A-␣B loop is highly conserved among mammalian PLC-␤ 1 enzymes but diverges among the four different PLC-␤ isoforms. This pattern suggests a possible molecular basis for the isoform specificity in PA regulation. The flexible loop is solvent-exposed and conformationally flexible in full-length PLC-␤, and its role in regulation of PLC-␤ is not understood (30). Solving the crystal structure of the turkey PLC-␤ C terminus required removal of 76% of the turkey flexible loop (residues 946 -978 in turkey PLC-␤ 2 ) (30). This corresponds to rat PLC-␤ 1 residues 961-993 and lies outside the sequence required for PA regulation as identified in the present study. Removal of this region had little effect on the ability of the turkey PLC-␤ 2 C-terminal fragment to inhibit intrinsic PLC-␤ 1 GAP activity toward G␣ q . These residues also do not contribute to PA regulation of PLC-␤ 1 , because A-(965-975) and A-(976 -993) mutants have wild-type PA regulation. Flexible loops do not appear to have structure-stabilizing roles, but they can accommodate regulatory ligands into binding pockets. In addition, flexible loops allow conformational flexibility, which can govern intramolecular interactions and potentially modulate sensitivity to physiological stimuli (37). The present data show that a portion of the PLC-␤ 1 flexible loop is required for interaction with PA and that these residues lie outside the dimer interface.
The structural nature of the interaction of PA with PLC-␤ 1 remains to be determined. Although PA binding regions have been identified in several proteins (2), there appears to be little overall similarity in the binding sequence or obvious binding determinants except for the presence of positively charged residues that suggest a role for electrostatic binding. The C-terminal domain of the turkey PLC-␤ shows a large concentration of positive charge at the interface between the ␣A and ␣B helices, and this charged region has been proposed to have a role in localizing PLC-␤ to the membrane (30). The major concentration of charge lies outside the region shown by this study to be required for PA interaction, which suggests that electrostatic interactions are not the  major determinant for PA binding. Additional experimental data support this conclusion. First, PS is also negatively charged but does not produce effects comparable with PA, which is inconsistent with a purely electrostatic mechanism. In addition, PS binding is inhibited 100% by 0.4 M KCl, whereas PA binding is more salt-resistant (Fig. 1). Second, although both DPPA and DPPS promote binding of the PLC-(878 -980) fragment to lipid vesicles, only DPPA induces an increase in the secondary structure of the protein (Fig. 3). Third, the most effective PA mutation, A-(944 -957) does not change the net charge within PLC. In contrast, both A-(928 -943) and A-(965-957), which have no effect or produce a minor decrease in PA regulation, remove 3 and 2 net charges, respectively. The data presented here suggest that PLC-␤ 1 dimerization, specific (salt-resistant, high affinity) PA binding, and stimulation of phospholipase activity are allosterically coupled. The specific activity of PLC-␤ 1 increases markedly with increasing concentration of the enzyme, and this increase is promoted by PA (Fig. 4). Consistent with this model, stimulation of phospholipase activity by PA is diminished at high PLC-␤ 1 concentration, where specific activity is also high. The effect of PA on protein concentration dependence is diminished in the A-(944 -957) mutant, which also displays decreased high affinity PA binding. The quantitative relationships among these effects (i.e. the coupling free energies) await more detailed measurements, and it is likely that these interactions will depend in detail on the lipid surface that contains the PA activator and the PIP 2 substrate. It may be that PLC-␤ 1 dimerization is the primary source of activation and that dimerization is driven by PA binding. Alternatively, PA binding might cause activation independently of dimerization. Regardless, the processes of activation, dimerization, and PA binding are clearly thermodynamically linked. Ultimately, the PLC-␤ 1 dimer appears to offer the optimal surface to which G␣ q binds to cause full activation (30,31).
At least two structural models for PA-induced dimerization are suggested by these data. Two binding sites for PA may exist, a high affinity and a low affinity binding site near the ␣A helix and adjacent loop. Cooperative interactions or dimerization might then modulate the contribution of each binding region to overall binding. Alternatively, only one PA binding site may exist, with its affinity for PA regulated by intramolecular interactions that occur in the ␣A helix/loop region. Nonspecific ionic PA binding may also be structurally delocalized and mechanistically unrelated to the interactions described here. Although the binding data suggest the existence of only one high affinity, salt-resistant PA binding site (supplemental Fig. 1), more detailed characterization of binding at different protein concentrations is necessary to address this question.
These studies suggest a novel role for PA in the regulation of GPCR signal transduction that is mediated through PLC-␤ 1 homodimerization. Activation of G q -PLC-␤ 1 signaling increases PA levels through activation of both DAG kinases and phospholipase Ds. The PA species thus generated would be chemically distinct because of their different precursors, and they may each exert unique regulation of PLC-␤ 1 . PA increases PLC-␤ 1 dimerization and promotes PLC-␤ 1 localization to membrane-associated signaling proteins. PA regulation is mediated by residues that include the flexible ␣A-B loop and ␣A helix of the G␣ q binding domain but are not required for G␣ q stimulation or GAP activity. PLC-␤ 1 binding to G␣ q , stimulation of phospholipase activity by G␣ q , and regulation by protein kinase C are all altered as a consequence of PA binding. PA thus coordinately modulates spatial and temporal GPCR signaling through PLC-␤ 1 . Cross-talk with PA generated by activation of other receptors may confer additional levels of regulation on GPCR-PLC-␤ 1 signaling.