Synergistic Ca2+ Responses by Gαi- and Gαq-coupled G-protein-coupled Receptors Require a Single PLCβ Isoform That Is Sensitive to Both Gβγ and Gαq*

Cross-talk between Gαi- and Gαq-linked G-protein-coupled receptors yields synergistic Ca2+ responses in a variety of cell types. Prior studies have shown that synergistic Ca2+ responses from macrophage G-protein-coupled receptors are primarily dependent on phospholipase Cβ3 (PLCβ3), with a possible contribution of PLCβ2, whereas signaling through PLCβ4 interferes with synergy. We here show that synergy can be induced by the combination of Gβγ and Gαq activation of a single PLCβ isoform. Synergy was absent in macrophages lacking both PLCβ2 and PLCβ3, but it was fully reconstituted following transduction with PLCβ3 alone. Mechanisms of PLCβ-mediated synergy were further explored in NIH-3T3 cells, which express little if any PLCβ2. RNAi-mediated knockdown of endogenous PLCβs demonstrated that synergy in these cells was dependent on PLCβ3, but PLCβ1 and PLCβ4 did not contribute, and overexpression of either isoform inhibited Ca2+ synergy. When synergy was blocked by RNAi of endogenous PLCβ3, it could be reconstituted by expression of either human PLCβ3 or mouse PLCβ2. In contrast, it could not be reconstituted by human PLCβ3 with a mutation of the Y box, which disrupted activation by Gβγ, and it was only partially restored by human PLCβ3 with a mutation of the C terminus, which partly disrupted activation by Gαq. Thus, both Gβγ and Gαq contribute to activation of PLCβ3 in cells for Ca2+ synergy. We conclude that Ca2+ synergy between Gαi-coupled and Gαq-coupled receptors requires the direct action of both Gβγ and Gαq on PLCβ and is mediated primarily by PLCβ3, although PLCβ2 is also competent.

Phosphoinositide-specific phospholipase C (PLC) 5 -dependent Ca 2ϩ responses are stimulated by several types of cell surface receptors, including members of the GPCR family. G␣ q -linked GPCRs stimulate Ca 2ϩ responses in most cells, whereas G␣ i -linked receptor stimulation of Ca 2ϩ release is variably seen, and the factors that determine this variation are not well understood (1)(2)(3). Simultaneous activation of G␣ iand G␣ q -linked GPCRs has revealed pathway cross-talk, resulting in synergistic Ca 2ϩ responses in a variety of primary and cultured cell types (4 -10). Proposed mechanisms have included effects on receptor availability, changes in G-protein turnover, increased availability of the PLC␤ substrate phosphatidylinositol 4,5-diphosphate, increased sensitivity of intracellular stores to inositol 1,4,5-trisphosphate, and enhanced linkage of intracellular Ca 2ϩ release to Ca 2ϩ influx (5). From these disparate results, it is clear that the mechanisms for Ca 2ϩ synergy may vary between systems and contexts.
We recently demonstrated that G␣ i -and G␣ q -coupled GPCRs synergize in stimulating a rise in [Ca 2ϩ ] i in macrophages (11). Synergy required simultaneous receptor activation, and it affected both the initial release of Ca 2ϩ from intracellular stores and the sustained elevation in Ca 2ϩ levels. Similar to synergy in other cell types, synergy in macrophages occurred at the level of inositol 1,4,5-trisphosphate production (10,(12)(13)(14)(15)(16). Although macrophages express multiple PLC␤ isoforms, these were used selectively in Ca 2ϩ signaling and synergy. In particular, PLC␤3 mediated most of the synergy, with a possible minor contribution by PLC␤2, whereas PLC␤4 inhibited synergy (11). In contrast to the synergy in PLC␤ activation, phosphatidylinositol 3-kinase activation was inhibited by the same signal interactions, demonstrating divergence in signaling effects on effector pathways downstream of the GPCRs.
Previous biochemical investigations of PLC␤ isoforms have demonstrated differential sensitivity to activation by G-protein subunits; whereas PLC␤2 and PLC␤3 bind to and are activated by both G␤␥ and G␣ q , PLC␤4 is effectively activated only by G␣ q (17)(18)(19)(20). At the receptor level, G␣ i -linked GPCRs activate PLC␤ through binding of G␤␥ alone, whereas signaling through G␣ q -linked GPCRs may activate PLC␤ through both G␤␥ and G␣ q . Thus, synergy in Ca 2ϩ signaling could involve effects of both G␤␥ and G␣ q on PLC␤2 or PLC␤3. Alternatively, it could involve activation of PLC␤ by only one of the two G-proteins. For example, synergy could involve only G␤␥ interactions between receptors, especially if different G␤␥ subunits were used by each receptor (21). Synergy could also be mediated solely by G␣ q , with increased G␤␥ levels serving to enhance the recycling of GTP-G␣ q , but this is unlikely because this effect should enhance the activity of PLC␤4, and it does not (11).
In the current studies, we further examined the mechanistic requirements for PLC␤-mediated Ca 2ϩ synergy responses. Studies of PLC␤-deficient primary macrophages revealed that only one PLC␤ isoform is required for synergy, either PLC␤3 or, less potently, PLC␤2.
To dissect the molecular mechanisms involved in synergy, we employed a cellular system with a greater efficiency of transfection and higher throughput than primary BMDMs, mouse NIH-3T3 cells expressing C5aR. This model permits G␣ i -mediated signaling in response to C5a together with G␣ q -mediated signaling through endogenous P2Y receptors in response to UDP or UTP.
Studies of the NIH-3T3 model supported and extended the observations in macrophages. They demonstrated a robust synergistic Ca 2ϩ response to C5a and UDP and demonstrated that synergy is dependent on endogenous PLC␤3. In contrast, PLC␤1 or PLC␤4 inhibited synergy, and inhibition by PLC␤4 required its C-terminal, G␣ q binding site. Following knockdown of PLC␤3 by RNAi, synergy could be restored by expression of either human PLC␤3 (hPLC␤3) or mouse PLC␤2 (mPLC␤2), which is not expressed natively in these cells.
Mutational analyses of hPLC␤3 revealed that both G␤␥ and G␣ q binding sites are necessary for full synergistic activation of PLC␤3 in 3T3 cells, confirming the requirement for both activators of PLC␤3 in intact cells. We conclude that synergy between G␤␥ and G␣ q occurs directly at the level of PLC␤3.

EXPERIMENTAL PROCEDURES
Reagents-UDP, UTP, LPA, platelet-activating factor, human C5a, spiradoline, and FITC-dextran were from Sigma-Aldrich. S1P was from Avanti Polar Lipids. Mouse IgG2a was from BD Pharmaceuticals. F(abЈ) 2 fragments of goat anti-mouse IgG were from Jackson Immunoresearch Inc. Anti-PLC␤3 clone B521 was from Paul Sternweis (University of Texas Southwestern) (17). Anti-phospho-Akt and anti-phospho-ERK were from Cell Signaling Technologies. Fura2 was from Molecular Probes. Ionomycin, thapsigargin, and pertussis toxin (PTx) were from Calbiochem. Detailed protocols for reagents, procedures, and solutions are provided upon request and are referenced according to protocol number (e.g. PP00000172).
Mice and Culture of BMDMs-PLC␤3-deficient mice were on a C57BL/6 background after 7 generations of backcrossing (22), and PLC␤2-deficient mice were on 129SvEV background (23). For mice doubly deficient in PLC␤3 and PLC␤2, PLC␤2 Ϫ/Ϫ mice were crossed twice to C57BL/6 mice and a PLC␤2 ϩ/Ϫ mouse was crossed to PLC␤3 Ϫ/Ϫ mice. Mice were bred and housed under approved animal protocols. To obtain BMDMs, femurs and tibias were removed from sex-and agematched mice (4 -20 weeks of age, matched Ϯ4 weeks) (PP00000172). Briefly, marrow was flushed from bones, erythrocytes were lysed, and the white cells were seeded in nontissue culture Petri dishes for differentiation and selection of macrophages by growth and adhesion. After 6 days, over 99% of the surviving cells were macrophages, as assessed by expression of CD11b/␣ M integrin and CD115/c-Fms/MCSF receptor, and these cells were maintained in culture for up to 35 days. Cells were cultured overnight in tissue culture plates prior to use in assays.
Cell Transduction with siRNA, Plasmid Expression Constructs, and Retroviruses-For RNAi, 3T3 cells were plated at 1.25 ϫ 10 5 cells/well in 6-well plates 24 h prior to transfection. Lipofectamine 2000 was used to deliver siRNA oligonucleotides at a final concentration of 100 nM. Media were changed 24 h post-transfection, and cells were plated for assay 48 h post-transfection (PP00000262). Bioassays and sampling for mRNA and protein expression level assessments were performed 72 h post-transfection. Knockdown efficiency of mRNA was quantified by quantitative RT-PCR, and proteins were quantified by Western blotting where antibodies were available.
The cDNA for human C5aR, with an N-terminal FLAG epitope, was generously provided by Henry Bourne (University of California San Francisco) and was transferred to pcDNA4zeo (Invitrogen). The cDNA for the mutant -opioid receptor (RASSL form Ro2 (24)) was generously provided by Bruce Conklin (University of California San Francisco) and was transferred to pMXpie (25). Murine, human, and bovine PLC␤-YFP fusion proteins and mutant constructs were prepared by using previously described cloning methods and constructs (26). The domain structure of PLC␤ is shown in Fig. 1, where previously demonstrated binding sites for G␤␥ and G␣ q are indicated (reviewed in Ref. 27). Residues in the Y-box and the C-terminal domains of either PLC␤3 or PLC␤4 were targeted for mutation to interfere with G␤␥ or G␣ q binding, respectively. Briefly, hPLC␤3 and bovine PLC␤4 cDNAs were cloned in gateway entry vectors, and site-directed mutagenesis was directed at putative G-protein interaction sites by using QuikChange (Stratagene), according to the method of the manufacturer. For both hPLC␤3 and bovine PLC␤4, four amino acids that are predicted to contribute to G␣ q interaction at the C terminus (hPLC␤3: Lys 965 , Arg 969 , Asp 1074 , and Arg 1089 ; bovine PLC␤4: Lys 942 , Lys 946 , Ser 1068 , and Arg 1083 ) were mutated to alanine (28,29). For hPLC␤3, three amino acids that are predicted to contribute to G␤␥ interaction (Glu 622 , Thr 623 , and Lys 624 ) were also mutated to alanine (30,31). For mPLC␤2, a stop codon was introduced after amino acid 840. N-terminal YFP tags were added to PLC␤ sequences by Gateway (Invitrogen) recombination to either pDS_EF1-YFP-XB for mammalian plasmid expression or to pDS-FBneo YFP-X for retroviral expression (AfCS plasmids, available through ATCC). Control constructs were ei-ther empty vector, YFP alone, or YFP-FLAG versions. For in vitro assessment of hPLC␤3 mutants, the recombinant proteins were produced via the baculovirus expression vector system. His tags were added to the C termini, the sequences were transferred to pFastBac1 vectors, SF9 cells were transfected, and the proteins were purified from cell lysates by onestep affinity chromatography on Ni 2ϩ -nitrilotriacetic acidagarose (Qiagen) (28).
3T3 cells were transfected with stable mammalian expression vectors by using Lipofectamine 2000. Amphotropic or ecotropic retroviruses were produced by using the Phoenix (32) or PlatE (33) packaging lines, respectively, with pMXpie or pFB-neo (Stratagene) vectors. 3T3 cells were selected and maintained in 100 g/ml zeocin, 400 g/ml G418, or 1 g/ml puromycin, either alone or in combination as appropriate. Where required, fluorescent cell populations were enriched by cell sorting by using a FACSAria (BD Biosciences). Expression levels were evaluated by Western blotting using anti-PLC␤3. The molecular weight difference between endogenous PLC␤3 and transduced YFP-PLC␤3 allowed differentiation of bands corresponding to each form.
Primary BMDMs were transduced by using retroviruses. Proliferating myeloid progenitor cells were infected on two consecutive days, beginning within 2 days of isolation from mouse femurs. Overnight infections with retrovirus-containing supernatants included Polybrene at 8 g/ml. After expansion, transduced macrophages were harvested and replated on tissue culture plastic or poly-L-lysine-coated glass chamber slides 24 h prior to assay.
In Vitro PLC␤ Activity Assays-PLC activity was measured by hydrolysis of [ 3 H]phosphatidylinositol 4,5-diphosphate as described previously (28,34). Briefly, the reactions of purified PLC␤ enzymes, G-protein subunits, and [ 3 H]phosphatidylinositol 4,5-diphosphate (added as mixed phospholipid vesicles) were allowed to proceed at 30°C for 10 min. Ca 2ϩ was maintained at 0.2 M. G␣ q -stimulated activity was measured in the presence of (0.2 or) 2 nM G␣ q (ϳEC 50 ) that had been previously activated by incubation with GTP␥S. G␤␥-stimulated activity was measured in the presence of 30 nM G␤␥.
Population Calcium Assays-Ca 2ϩ responses were measured by monitoring the fluorescence of Fura-2-loaded cells (PP00000211). Base-line readings were collected for 30 -40 s prior to stimulation with ligands. Calibration steps included the additions of a Ca 2ϩ -minimizing solution (PS00000607) and Fura-2 Ca 2ϩ -saturating solution (PS00000608) at the end of each recording to allow calculation of [Ca 2ϩ ] i values according to the method of Grynkiewicz et al. (52), assuming a cytoplasmic k D value of 250 nM for Fura-2. Ca 2ϩ signals during the response period were quantified by features as indicated, including the peak offset response (difference between base-line Ca 2ϩ level and the maximal Ca 2ϩ level observed, reported in nM) and an integrated response (integrated Ca 2ϩ level above the average base line over the indicated time period, reported in nM ϫ s).
Single-cell Calcium Assays-Cells were plated in chambered coverglasses (Nunc, 8 wells/coverglass), cultured overnight, and loaded with Fura-2/AM as described above. Video microscopy was performed by using a Nikon TE-300 fluorescence microscope equipped with a Photometrics HQ2 camera, 37°C stage incubator (Bionomics), xenon lamp (Sutter), and filter/shutter/ dichroic controllers (Sutter and Conix). SimplePCI software was used to control collection parameters and to extract fluorescence intensity data for individual cells. Intracellular [Ca 2ϩ ] i was estimated as described for population assays.
Statistical Analyses-The error bars in graphs depict the S.E. The statistical significance of each comparison was evaluated by performing Student's t test or a one-way analysis of variance followed by Dunnett's test or individual t tests (with Bonferroni correction). A p value of Ͻ0.05 was considered significant.

Macrophages Deficient in Both PLC␤2 and PLC␤3 Have No
Synergistic Ca 2ϩ Response to C5a and UDP-Synergy in Ca 2ϩ signaling in macrophages between C5a, acting on the G␣ icoupled C5aR, and UDP, acting on G␣ q -coupled P2Y receptors, was previously found to be primarily dependent on expression of PLC␤3, but some synergy remained in PLC␤3-deficient cells (11). Because PLC␤4 appears to have an inhibitory role and PLC␤1 is not expressed in macrophages, the residual Ca 2ϩ synergy in PLC␤3-deficient macrophages was thought to depend on PLC␤2 (11). To test this hypothesis, we examined Ca 2ϩ synergy in mouse macrophages deficient in both PLC␤2 and PLC␤3 (PLC␤2/3 Ϫ/Ϫ ) in comparison with macrophages deficient in PLC␤3 alone (PLC␤3 Ϫ/Ϫ ). In the absence of both PLC␤2 and PLC␤3, BMDMs (which still expressed PLC␤4) lost the residual synergy that was apparent in macrophages deficient in PLC␤3 alone (Fig. 2, A-C). Thus, although synergy in macrophages is FIGURE 1. Schematic of PLC␤ structure and design of G-protein binding site mutants. The theoretical PLC␤ structural domain organization (NCBI) is shown along with sites of modification as described under "Experimental Procedures." YFP was added as an N-terminal fusion protein, and point mutations targeting G␤␥ and G␣ q binding sites, based on homology to PLC␤1, were created as X 3 Ala mutations (*). The arrows indicate the location of amino acids mutated, and the numbers indicate their position relative to the N terminus of hPLC␤3. The species of isoform used is indicated as human (h) and bovine (b). Also used but not shown are YFP-mPLC␤1, -2, -3, and -4 without mutation and mPLCb2 truncated at amino acid 840.
primarily dependent on PLC␤3, PLC␤2 is capable of supporting synergy.
The Ca 2ϩ response to UDP alone in PLC␤2/3 Ϫ/Ϫ macrophages differed little from the response by macrophages deficient in PLC␤3 alone (Fig. 2D), presumably reflecting activation of PLC␤4 by G␣ q , but the Ca 2ϩ response to C5a alone was diminished in PLC␤2/3 Ϫ/Ϫ BMDMs compared with macrophages deficient in PLC␤3 alone, indicating a contribution from PLC␤2.
Reconstitution of PLC␤2/3-deficient BMDMs with PLC␤3 Alone Restores Synergy-The capacity of PLC␤3 alone to support Ca 2ϩ synergy was evaluated by reconstituting wild type or PLC␤2/3-deficient BMDMs with PLC␤3 coupled to YFP or, as a control, with FLAG peptide coupled to YPF. Introduction of wild-type PLC␤3 increased single-ligand responses to both C5a and UDP, and it restored synergy to normal levels ( Fig. 2E and supplemental Fig. S1). Thus, synergy can be supported by either PLC␤2 or PLC␤3 alone.
In these studies, transduction of PLC␤ into BMDMs was inefficient, requiring the study of large numbers of cells to obtain statistically valid results. Therefore, for further studies of the role of PLC␤ isoforms in synergy, we moved to the study of NIH-3T3 cells stably expressing C5aR, beginning with validation of these cells as a model for synergy in Ca 2ϩ signaling between G␣ i -and G␣ q -coupled receptors.
NIH-3T3 Cells Demonstrate Robust Synergy in Ca 2ϩ Signaling in Response to Endogenous Purinergic Receptors (P2YRs) and a Stably Transfected C5a Receptor (C5aR)-Mechanistic studies of the signaling interaction between C5a and UDP or UTP GPCRs at the level of PLC␤ activation were performed in mouse NIH-3T3 cells expressing a stably transfected human C5aR and endogenous P2YRs. Robust synergy between C5a and either UDP (Fig. 3A) or UTP (Fig. 3B) was observed, both in the initial peak and in sustained levels of [Ca 2ϩ ] i . C5a or UDP alone generated low or undetectable Ca 2ϩ responses, even at concentrations up to 100 M ( Fig. 3B and supplemental Figs. S2 and S3). Substituting UTP for UDP, however, consistently gave a detectable response by itself, and this was elevated by C5a to levels 2-4 times the predicted additive value (Fig. 3C). Of note, C5a increased Ca 2ϩ responses to UTP even when UTP was used at maximal concentrations, evidence that C5a recruits signaling pathways that are not utilized by UTP.
To confirm that NIH-3T3 cells serve as an adequate model for synergy in Ca 2ϩ signaling between G␣ i -and G␣ q -linked GPCRs, we evaluated synergy by a number of alternate receptor pairings with or without inhibition of G␣ i signaling by PTx. As expected, the C5a and synergistic responses to UTP ϩ C5a were sensitive to PTx (supplemental Fig. S4), whereas the responses to UTP were not (supplemental Fig. S5A), as seen previously in BMDMs (11). Endogenous LPA and S1P receptors generated PTx-sensitive Ca 2ϩ responses (supplemental Fig. S5A), as did a transfected modified -opioid GPCR, Ro2 (24), which signals solely through G␣ i in response to spiradoline (supplemental Figs. S5C and S6). A screen of dual-ligand Ca 2ϩ responses showed that pairs of receptors with heterologous couplings to G␣ i and G␣ q showed synergy, whereas homologous couplings did not (supplemental Fig. S5, B and D; summarized in supplemental Table S1). Thus, Ca 2ϩ synergy in NIH-3T3 cells reflects G␣ q /G␣ i GPCR cross-talk, as it does in macrophages and several other cell types.
Ca 2ϩ Synergy in NIH-3T3 Cells Depends on Endogenous PLC␤3-We next sought to determine the contribution of different PLC␤ isoforms to Ca 2ϩ signaling in NIH-3T3 cells. By quantitative RT-PCR, NIH-3T3 cells expressed relatively high levels of PLC␤1 and PLC␤3 transcripts but lower levels of PLC␤4 and little or no PLC␤2 (Fig. 4). Expression of each isoform was selectively reduced by RNAi, which reduced transcripts for PLC␤1, PLC␤3, and PLC␤4 by 60-80% (supplemental Fig. S7A). Only the knockdown of PLC␤3 significantly reduced the Ca 2ϩ synergy response (Fig. 5). Knockdown of PLC␤2 had no effect, but this isoform is not significantly expressed in NIH-3T3 cells, so this may be viewed as an addi- C, the responses were quantified by using peak offset measurements, the difference from base line to observed peak. The dual-ligand responses were compared with the predicted additive responses (sum of single-ligand responses) and a synergy ratio calculated as observed/predicted additive peak offsets. For the C5a ϩ UTP experiment shown, the synergy ratio was 2, but the predicted additive level for C5a ϩ UDP was too low for reliable estimation of the synergy ratio. Data shown are from a representative assay of n ϭ 7 with similar results. tional negative control. Knockdown of PLC␤1 or PLC␤4 increased synergy, but this change did not reach statistical significance for either isoform. These results are similar to our results in genetically deficient BMDMs, where the absence of PLC␤3 reduced synergy, whereas the absence of PLC␤4 enhanced synergy (PLC␤1 is the closest isoform to PLC␤4 biochemically (27)). Hence, our results in both BMDMs and NIH-3T3 cells indicate that PLC␤3 contributes to synergy in Ca 2ϩ signaling, whereas PLC␤4 or PLC␤1 do not and instead inhibit synergy. Knockdown of single PLC␤ isoforms did not significantly alter single-ligand Ca 2ϩ responses; responses to C5a were too low to identify changes, whereas UTP responses were not affected by knockdown of PLC␤2 or PLC␤3 and only slightly affected by knockdown of PLC␤1. Presumably, the loss of individual PLC␤ isoforms following RNAi is compensated by signaling through others, all of which can bind G␣ q .
Overexpression of PLC␤ Isoforms Selectively Enhances or Reduces Ca 2ϩ Signaling-To further examine the selectivity of PLC␤ isoforms in contributing to Ca 2ϩ synergy, we overexpressed YFP-tagged PLC␤ isoforms in NIH-3T3 cells. By flow cytometry, the PLC␤ isoforms were expressed at comparable levels (supplemental Fig. S7, B and C). By Western blotting, PLC␤3 levels were 2.7 Ϯ 0.2-fold above levels of endogenous PLC␤3 (data not shown). Ca 2ϩ responses were compared with responses by cells transfected with YFP-tagged FLAG epitope (YFP-FLAG). Overexpression of PLC␤3 had no significant effect on any single-ligand responses, perhaps because endogenous levels of this PLC␤ isoform are already high. Overexpression of PLC␤2, however, markedly enhanced the response to C5a, although it did not alter signaling to UTP, LPA, or S1P. The capacity of PLC␤2 to augment C5a signaling following overexpression in 3T3 cells, whereas overexpression of PLC␤3 did not, suggests that PLC␤2 may differ qualitatively from PLC␤3 in its activation by C5a. Overexpression of PLC␤1 or PLC␤4 augmented G␣ q -coupled UTP responses (Fig. 6A), but it inhibited the G␣ i -coupled responses by C5a, LPA, and S1P, perhaps reflecting differences in PLC␤ isoform activation by G␣ q versus G␤␥ subunits (17,18,20). Overall, the results of PLC␤ overexpression suggest that Ca 2ϩ responses to C5a may be inhibited by PLC␤ isoforms that are less sensitive to G␤␥ (PLC␤1 and PLC␤4), but they are enhanced by isoforms that are activated by G␤␥ (PLC␤2). Synergy in response to dual ligands also differed  between either PLC␤1 or PLC␤4 relative to PLC␤2 or PLC␤3. Overexpression of PLC␤1 or PLC␤4 inhibited synergy following stimulation with C5a plus UTP (Fig. 6B), but this inhibition was lost with mutation of G␣ q -binding domain of PLC␤4 (supplemental Fig. S7D). In contrast, overexpression of PLC␤2 or PLC␤3 did not alter synergy; the synergy ratio was maintained and was not further increased by overexpression of PLC␤2 or PLC␤3. As noted above, signaling by C5a alone rose substantially following overexpression of PLC␤2, and so did dual-ligand signaling by C5a plus UTP, but the synergy ratio was the same as in YFP-transfected cells, suggesting that PLC␤2 overexpression augments the magnitude of synergy responses in parallel with its augmentation of signaling by C5a alone. Thus, results from overexpression of PLC␤ isoforms are consistent with the results from the knockdown experiments in supporting an inhibitory effect of PLC␤1 or PLC␤4 on Ca 2ϩ response synergy, in contrast with dependence for synergy on PLC␤3. Although PLC␤2 is normally not found in NIH-3T3 cells, it appears capable of contributing to synergy, as it can in BMDMs (although its contribution to synergy in BMDMs is low).

Reconstitution of PLC␤3 Expression Restores Ca 2ϩ Synergy
Lost with PLC␤3 RNAi-To further validate the requirement for PLC␤3 expression for Ca 2ϩ synergy, we combined knockdown of endogenous murine PLC␤3 in NIH-3T3 cells by RNAi with restoration of PLC␤3 expression by transfection of human PLC␤3. The RNAi sequence targeted mPLC␤3 but not hPLC␤3. RNAi reduced mPLC␤3 mRNA by 74 Ϯ 7%, as assessed by species-specific quantitative RT-PCR, and it reduced mPLC␤3 protein by 50 -80% (supplemental Fig. S8), as assessed by Western blotting. In accord with this, RNAi reduced Ca 2ϩ synergy by 60 -70% (Fig. 7, A, B, and E). Expression of hPLC␤3 reversed the loss of synergy following knockdown of mPLC␤3 by RNAi (Fig. 7, C-E), supporting the conclusion that the loss of synergy following RNAi of PLC␤3 reflects knockdown of PLC␤3 expression. The restoration of synergy by transfected hPLC␤3 was not due to negation of the RNAi by hPLC␤3 (e.g. by binding siRNA) because mPLC␤3 remained suppressed (supplemental Fig. S8B). Indeed, attempted overexpression of transcripts for mPLC␤3 instead of hPLC␤3 failed to restore synergy (supplemental Fig. S9C), because knockdown of mPLC␤3 persisted (supplemental Fig. S8B). Synergy was also restored by mPLC␤2 although not as robustly as by hPLC␤3 (supplemental Fig. S9D). Thus, PLC␤2, like PLC␤3, has the capacity to mediate synergy in these cells.
Both the G␤␥ and G␣ q Binding Sites of PLC␤3 Are Required for Synergistic Ca 2ϩ Responses-PLC␤3 is responsive to both G␤␥ and G␣ q , and our studies suggest that synergy by G␣ i and G␣ q GPCRs may be mediated by the combined action of G␤␥ and G␣ q on a single PLC␤ isoform. Several alternate mechanisms have been proposed for synergy between G␣ iand G␣ q -linked GPCRs in different systems (5), however, and a direct synergistic activation mechanism does not preclude contributions from other mechanisms. To examine if the mechanism of Ca 2ϩ synergy for dual C5aR and P2YR stimulation shares a requirement for activation of PLC␤3 by both G␤␥ and G␣ q , we evaluated the ability of PLC␤3 mutated in residues critical for activation by these respective G-proteins to mediate synergy. Residues in the Y-box of PLC␤3 homologous to those critical for the binding of G␤␥ to PLC␤1 or residues in the C terminus homologous to those critical for the binding of G␣ q to PLC␤1 were targeted with X 3 Ala mutations (28 -31). These sites are depicted in Fig. 1 and described under "Experimental Procedures" as E262A,T263A,K264A in the Y-box and K965A,R969A,D1074A,R1089A in the C terminus. Mutation of the Y-box disrupted activation by G␤␥ in vitro by ϳ80%, but activation by G␣ q remained intact ( Table  1). Mutation of the C terminus diminished activation by G␣ q by ϳ65%, whereas activation by G␤␥ was largely intact. Thus, the functional defects of these mutants were relatively selective.
We transfected NIH-3T3 cells with the different YFP-hPLC␤3 mutants or with a double mutant in which both sites were targeted. Expression of PLC␤3 mutated in the Y-box failed to reconstitute synergy lost by knockdown of endogenous mPLC␤3 (Fig. 8), indicating a requirement for G␤␥ activation of PLC␤ for synergy. In contrast, reconstitution with WT hPLC␤3 actually enhanced synergy in these experiments, suggesting that greater activity exists for this construct, perhaps due to YFP fusion or human versus mouse sequence effects. Such an enhancement was suggested, but not statistically significant, in Fig. 7E. Expression of PLC␤3 mutated in the C-terminal domain only partially restored synergy (Fig. 8) because synergy remained at base-line levels and did not show the enhancement seen with WT PLC␤3. The C-terminal mutant appeared ϳ50% as effective as intact hPLC␤3 (supplemental Fig. S9C). This partial effect is consistent with the incomplete effect of the mutation in disrupting G␣ q activation ( Table 1). Disruption of both Y-box and C-terminal sites dis-rupted synergy comparable with the Y-box alone. Thus, both Y-box and C-terminal sites in PLC␤3 contribute to synergistic Ca 2ϩ responses, supporting the premise that synergy in intact cells depends on activation of PLC␤3 by both G␤␥ and G␣ q . These data support a mechanism of Ca 2ϩ synergy resulting from the synergistic activation of PLC␤3 by the combined activity of G␤␥ and G␣ q released by G␣ i -and G␣ q -linked GPCRs.

DISCUSSION
In the current studies, we demonstrate that the synergistic increases in intracellular Ca 2ϩ following simultaneous stimulation of G␣ i -and G␣ q -linked GPCRs in cells can be mediated by PLC␤3 or PLC␤2. PLC␤1 and PLC␤4 are instead inhibitory. Ca 2ϩ synergy depends on binding sites in PLC␤3 for both G␤␥ in the Y-box domain and G␣ q in the C terminus. Based on these findings, we present a model of Ca 2ϩ synergy resulting from the direct, synergistic activation of PLC␤3 by combined stimulation with G␤␥ and G␣ q .
PLC␤3 is the predominant isoform used for single-ligand and synergistic Ca 2ϩ responses in macrophages (11). Despite the fact that macrophages have abundant transcripts for PLC␤2, PLC␤2 does not effectively substitute for PLC␤3 when the latter is absent (11). We now show that PLC␤2 is nonetheless required for the low level of residual synergy detected in PLC␤3-deficient macrophages because macrophages deficient in both PLC␤3 and PLC␤2 demonstrated no synergy in their Ca 2ϩ response to C5a and UDP. Also, we found that PLC␤2 could substitute for PLC␤3 in NIH-3T3 cells (where it is not normally expressed at detectable levels); either endogenous PLC␤3 or exogenous PLC␤2 produced a robust synergistic Ca 2ϩ response in these cells. In contrast, PLC␤4 did not  mediate synergy in either system. Instead, either PLC␤4 or PLC␤1 isoforms tended to inhibit synergy when they were overexpressed. These results may be viewed in light of known features of the PLC␤ isoforms in their binding to G-protein subunits (reviewed in Ref. 27). G␤␥ has been shown to interact both with residues in the N-terminal pleckstrin homology domain and in the Y-box of the PLC␤ catalytic domain (30,31,35). Only PLC␤3 and PLC␤2 show high affinity binding and activation by G␤␥, and varied combinations of ␤ and ␥ subunits can differ in their effectiveness (36 -40). In our studies, PLC␤3 and PLC␤2 were the only isoforms able to reconstitute synergy in cells, and mutation of the Y-box G␤␥ binding site alone led to a substantial loss of the synergistic activation of PLC␤3.
Like G␤␥, G␣ q has at least two potential binding sites on PLC␤, one in the C terminus and another in the C2 domain (28,(41)(42)(43)(44). Although G␣ q can activate all isoforms of PLC␤, PLC␤2 is the least sensitive (45,46). Mutation of the C terminus of PLC␤1 can disrupt its activation by G␣ q but not G␤␥ (28,29). In our studies, homologous mutations of PLC␤3 also disrupted G␣ q activation, but not fully; activation was reduced by ϳ65%. This reduction in G␣ q activation of the hPLC␤3 mutant in vitro correlated with a ϳ50% reduced magnitude of synergy observed in cells. The residual G␣ q activation in the PLC␤3 C-terminal mutant suggests either that alternate residues in this area contribute to G␣ q activation or that there is a role for the other known G␣ q -binding site in the C2 domain. Although our C-terminal mutation of PLC␤3 did not fully ablate G␣ q activation, our data indicate a contribution of G␣ q binding to the C terminus of PLC␤3 in the synergy mechanism. A role for G␣ q in the synergy mechanism was further suggested by the inability of truncation mutants of PLC␤2, which eliminated the C-terminal G␣ q binding region (truncation at amino acid 840), to mediate synergy in 3T3 cells (data not shown). Overall, our results demonstrate a direct role for G␤␥ in PLC␤3-mediated synergy via binding in the Y-box catalytic domain, and they support a role for G␣ q at the targeted C-terminal binding site.
The capacity to test the activation of purified PLC␤3 in vitro by individual or combined G-protein subunits has opened the path to exploring mechanisms by which G␤␥ and G␣ q may synergize in activating this PLC␤ isoform. Recent studies by Philip and Ross (47). demonstrate synergistic activation of PLC␤3 by simultaneous G␤␥ and G␣q stimulation in vitro, in accord with our studies in cells. These studies further show that the combined action of G␤␥ and G␣ q on PLC␤3 enhances conversion of the enzyme to the active state when both ligands are bound rather than merely enhancing ligand affinity.
The selective use of PLC␤ isoforms in GPCR Ca 2ϩ signaling and synergy may involve mechanisms in addition to their binding of G-proteins, such as the selective use of different scaffolding or other interacting proteins (48,49). PDZ domain-containing proteins may be of central importance in the recruitment of PLC␤s and other regulators of GPCR-mediated Ca 2ϩ signaling (48 -50). Although all isoforms of PLC␤ have a PDZ domain-binding motif (X(S/T)X(V/L)) at the C terminus, their peptide sequences vary and may thus make differential use of PDZ domains. In support of this, PLC␤3 preferentially binds to specific PDZ domain-containing scaffolding proteins, including NHERF1, E3KARP, and SHANK2 (48,49,51). The selective use of PLC␤3 in GPCR signaling by macrophages could relate to the selective expression of such scaffolding proteins in these cells, 6 whereas synergy in NIH-3T3 cells may involve scaffolding proteins that are recognized more equivalently by both PLC␤3 and PLC␤2. Further examination of PLC␤3 and PLC␤2 structural requirements for synergy are necessary to evaluate these contributions.