Activation of Phospholipase C-ε by Heterotrimeric G Protein βγ-Subunits*

PLC-ε was identified recently as a phosphoinositide-hydrolyzing phospholipase C (PLC) containing catalytic domains (X, Y, and C2) common to all PLC isozymes as well as unique CDC25- and Ras-associating domains. Novel regulation of this PLC isozyme by the Ras oncoprotein and α-subunits (Gα12) of heterotrimeric G proteins was illustrated. Sequence analyses of PLC-ε revealed previously unrecognized PH and EF-hand domains in the amino terminus. The known interaction of Gβγ subunits with the PH domains of other proteins led us to examine the capacity of Gβγ to activate PLC-ε. Co-expression of Gβ1γ2 with PLC-ε in COS-7 cells resulted in marked stimulation of phospholipase C activity. Gβ2 and Gβ4 in combination with Gγ1, Gγ2, Gγ3, or Gγ13 also activated PLC-ε to levels similar to those observed with Gβ1-containing dimers of these Gγ-subunits. Gβ3 in combination with the same Gγ-subunits was less active, and Gβ5-containing dimers were essentially inactive. Gβγ-promoted activation of PLC-ε was blocked by cotransfection with either of two Gβγ-interacting proteins, Gαi1 or the carboxyl terminus of G protein receptor kinase 2. Pharmacological inhibition of PI3-kinase-γ had no effect on Gβ1γ2-promoted activation of PLC-ε. Similarly, activation of Ras in the action of Gβγ is unlikely, because a mutation in the second RA domain of PLC-ε that blocks Ras activation of PLC failed to alter the stimulatory activity of Gβ1γ2. Taken together, these results reveal the presence of additional functional domains in PLC-ε and add a new level of complexity in the regulation of this novel enzyme by heterotrimeric G proteins.

PLC 1 -catalyzed hydrolysis of polyphosphoinositides is a necessary target cell response in the physiological action of many hormones, neurotransmitters, growth factors, and other extra-cellular stimuli (1). Three classes of PLC isozymes have been considered historically to underlie these signaling responses (2). A panoply of seven transmembrane-spanning receptors activate isozymes of the PLC-␤ class through release of ␣-subunits of the G q family of heterotrimeric G proteins (3)(4)(5)(6). Certain PLC-␤ isozymes also are activated by G␤␥ (7)(8)(9), which has been proposed largely to originate from activated Gi family G proteins. PLC-␥ isozymes are activated by protein phosphorylation following from activation of tyrosine kinase receptors or from receptors linked to tyrosine kinases (10 -12). Although there is evidence for Ca 2ϩ -and transglutaminase II-promoted regulation of PLC-␦ (13,14), the importance of hormonal regulation of this PLC isozyme remains unclear (15).
The recent identification of PLC-⑀ revealed a fourth class of PLC isozymes (16 -19). PLC-⑀ contains both a CDC25 homology domain at its amino terminus and a pair of RA domains at the carboxyl terminus, suggesting direct involvement of this PLC isozyme in signaling promoted by Ras superfamily GTPases. Indeed, GTPase-deficient, constitutively active, mutants of Ras promote increases in inositol phosphate accumulation in cells cotransfected with PLC-⑀ (17). Moreover, this PLC isozyme may function at a nexus between Ras GTPase signaling and signaling through heterotrimeric G proteins because G␣ 12 , but not G␣ q , also activates PLC-⑀ in cotransfection experiments (18).
We illustrate here an additional layer in the architecture of PLC-⑀-mediated signaling. A previously unrecognized PH domain followed by an EF-hand domain exists in the amino terminus of PLC-⑀. In light of previous observations that G␤␥ interacts with PH domains in other signaling proteins, we speculated that G␤␥ might act as an efficacious activator of PLC-⑀ and found that PLC-⑀ catalytic activity was stimulated by co-expression with various G␤␥ dimers. This activity was not dependent on activation of PI3-kinase or on an intact RA domain. These results add further complexity to the signaling pathways that may converge on PLC-⑀.

EXPERIMENTAL PROCEDURES
Materials-Rat PLC-⑀ (17), Ras, G␣ 12 , and G␣ 13 (20), and G␤ and G␥ expression vectors (21) were as previously described. The expression vector for the carboxyl-terminal domain of GRK2 was generously provided by Dr. Robert J. Lefkowitz, Duke University. Wortmannin and LY294002 were purchased from Biomol (Plymouth Meeting, PA). Inositol-free DMEM was purchased from ICN Biomedicals (Costa Mesa, CA). All other reagents were from sources previously noted (22,23).
Computational Tools-PLC-⑀ protein sequences were analyzed for structural relationships using the three-dimensional-PSSM method (24) as implemented on the ICRF Fold Recognition Server. 2 Threedimensional -PSSM analyses were performed using the following data base and software versions: fold library version 1.53.6230, sequence data base version 2001.9.7, 3D-PSSM binary version 2.2, and server version 2.6.0. Multiple sequence alignments of PLC-⑀/PLC210 sequences were first generated using the PILEUP program of the GCG Wisconsin Package (version 10.2, Genetics Computer Group, Madison, WI). Manual addition of rat PLC-␦1 and human GRK2 sequences to these alignments was performed using the GCG SeqLab Editor with data from three-dimensional-PSSM-generated pairwise alignments. Sequence alignments were then exported to the Macintosh implementation of the BOXSHADE program 3 for highlighting positions of conserved amino acid residue chemistry.
Transfection of COS-7 Cells and Quantitation of PLC Activity-COS-7 cells were subcultured in 12-well culture dishes at a density of 20,000 cells/cm 2 and maintained in DMEM supplemented with 10% fetal bovine serum at 37°C in an atmosphere of 95% air/5% CO 2 . The indicated plasmid DNA vectors were transfected utilizing Fugene 6 (Roche Molecular Biochemicals) transfection reagent (2 l/well) according to the manufacturer's protocol. Approximately 24 h after the addition of DNA and the transfection agent, the medium was changed to inositol-free DMEM containing 1 Ci of [ 3 H]inositol/well. After an additional 12-h incubation, [ 3 H]inositol accumulation was initiated by addition of LiCl to a final concentration of 10 mM. The reaction was stopped after 60 min by aspiration of the medium and the addition of 50 mM ice cold formic acid. [ 3 H]Inositol phosphates were quantitated by Dowex chromatography as described previously (23).

Identification of Pleckstrin Homology and EF-hand Domains
within PLC-⑀-PLC-⑀ contains the core catalytic domains (X, Y, and C2) common to all PLC isozymes (Fig. 1A) as well as CDC25 homology and RA domains unique to PLC-⑀ and specifying upstream and downstream interactions with Ras superfamily GTPases (16 -18). To identify additional functional modules in PLC-⑀, the three-dimensional-PSSM method of Kelley and colleagues (24) was applied to the polypeptide sequences that separate known structural domains within rat and human PLC-⑀ and PLC210, the PLC-⑀ orthologue of Caenorhabditis elegans (25). The top two predictions for structural motifs between the CDC25 and X domains of rat PLC-⑀ (aa 829 -1373) and human PLC-⑀ (aa 832-1392) were the PH domain (PDB code 1mai) and third and fourth EF-hands of rat PLC-␦1 (PDB code 1qas) (26). The top prediction for the analogous region (aa 334 -910) of PLC210 was the rat PLC-␦1 EF-hand domain; the PLC-␦1 PH domain match was tenth on the list (the other eight predictions were EF-hands of other proteins including troponin-C, calmodulin, and aequorin). The scoring matrix method was repeated with the isolated NH 2 -and COOH-terminal halves of this region of the three PLC-⑀ isozymes. Significant three-dimensional-PSSM E-values were obtained for the PH domain assignment (rat PLC-⑀, E ϭ 0.01, 95% certainty; human PLC-⑀, E ϭ 0.01, 95% certainty; PLC210, E ϭ 0.18; 80% certainty) and the EF-hand assignment (rat PLC-⑀ (aa 1001-1372), E ϭ 0.09, 90% certainty; human PLC-⑀ (aa 1001-1392), E ϭ 0.14, 80% certainty; PLC210 (aa 501-893), E ϭ 0.01, 95% certainty).
Multiple sequence alignments of the predicted PLC-⑀/ PLC210 PH and EF-hand domains are presented in Fig. 1, B

FIG. 1. Domain organization of PLC isozymes and secondary structure predictions of the PH and EF-hand domains of PLC-⑀.
A, structural motifs present within PLC-␤, PLC-␥, PLC-␦, and PLC-⑀ isozymes are outlined, including the PH, EF-hand, catalytic TIM-barrel (X, Y), and C2 domains shared among all isozymes. CT, regulatory COOH terminus of PLC-␤ responsible for specific binding and activation by G-protein ␣ q/11 subunits; SH2 and SH3, an internal Src-homology-2 and -3 domain array present within the second PH domain of PLC-␥ isozymes; CDC25, Ras superfamily guanine nucleotide exchange factor domain. B, multiple sequence alignment of the PH domains of rat PLC-␦1 (SwissProt PID1 Rat), rat and human PLC-⑀ (GenBank TM accession numbers AF323615 and AF190642), C. elegans PLC210 (GenBank TM AF044576), and human GRK2 (GenBank TM U08438). Known secondary structures within the PH domains of rat PLC-␦1 (26) and GRK2 (28) are overlined and underlined, respectively. The term molten-helix refers to a highly mobile extension of the PH domain COOH-terminal ␣-helix (28) required for GRK2 binding to G␤␥ (29). Conserved positions are denoted by black boxes. Predicted (based on PSI-Pred) ␣-helices are highlighted in red and predicted ␤-sheets in blue. C, multiple sequence alignment of the EF-hands of rat PLC-␦1, rat and human PLC-⑀, and C. elegans PLC210. Known boundaries of rat PLC-␦1 ␣-helices (26) are overlined. The color scheme for amino acids is the same as described for panel B. and C, respectively. These alignments include known boundaries of secondary structure elements (␣-helices and ␤-strands) within the rat PLC-␦1 PH and EF-hand-3/-4 regions (26) as well as secondary structure predictions of the corresponding PLC-⑀/PLC210 regions as derived from the PSI-Pred algorithm (27). Overall concordance was high between the predicted secondary structures and known structural elements within both domains. However, differences were observed in the predicted ␣-helical character within rat and human PLC-⑀ corresponding to ␤-strands ␤4 and ␤5 of the PLC-␦1 PH domain (Fig. 1B) and in a short stretch of predicted ␤-strand within rat and human PLC-⑀ corresponding to ␣-helix E4␣Ј of the fourth PLC-␦1 EFhand (Fig. 1C). The predicted loops between ␤-strand pairs ␤3/␤4 and ␤5/␤6 in the PLC-⑀/PLC210 PH domains are longer than their counterparts within PLC-␦1; in the case of the ␤3/␤4 intervening region, additional secondary structure elements are predicted by the PSI-Pred algorithm. The intervening sequence between the PH and EF-hand domains is also much longer in PLC-⑀/PLC210 proteins (e.g. 274 amino acids in human PLC-⑀) than in other PLC isozymes (10 ϳ 30 amino acids; Fig. 1A). This intervening polypeptide may fold to form additional EF-hands currently unidentifiable by primary sequence searches or structure threading algorithms, 4 because all other PLC isozymes encode four EF-hand ␣-helical pairs carboxylterminal to the PH domain.
An additional feature of the putative PH domain within PLC-⑀/PLC210 proteins is a predicted extension of the carboxyl-terminal ␣-helix well beyond the ␣1 helix of PLC-␦1 (aa 115-128; Fig. 1B). This ␣-helical extension is rich in arginine and lysine residues and thus reminiscent of the highly mobile, positively charged "molten helix" observed in the NMR structure of the human GRK2 PH domain (PDB code 1bak; Fig. 1B) and critical for GRK2 binding to G-protein ␤␥ dimers (28,29).
Activation of PLC-⑀ by G␤␥-PH domains serve as recognition motifs for phosphoinositides or proteins such as G␤␥. The presence of a PH domain in PLC-⑀ suggested that, in addition to regulation by G␣ 12 and Ras, PLC-⑀ also is regulated by G␤␥.  Fig. 2A). However, cotransfection of PLC-⑀ with G␤ 1 ␥ 2 produced marked increases in [ 3 H]inositol phosphate accumulation to levels similar to those observed with G␣ 12 or G␣ 13 (Fig. 2A). Cotransfection of maximally effective concentrations of G␤ 1 ␥ 2 and G␣ 12 (or G␣ 13 ) with PLC-⑀ produced no greater activity than with cotransfection of PLC-⑀ with each subunit alone (data not shown). Similarly, co-expression of G␤ 1 ␥ 2 together with GTPase-deficient, constitutively active mutants of G␣ 12 (Q229L) or G␣ 13 (Q226L) resulted in activity similar to that observed with the GTPasedeficient mutant alone (Fig. 2B).
Two separate experiments were employed to illustrate that free G␤␥ is likely necessary for activation of PLC-⑀. First, whereas G␣ i1 alone had no effect on PLC-⑀ activity (data not shown), coexpression of G␤ 1 ␥ 2 with G␣ i1 resulted in loss of capacity of G␤ 1 ␥ 2 to activate the enzyme (Fig. 3A). Similarly, the GRK2 carboxyl terminus, which was previously demonstrated to bind to G␤␥ (31,32), also reduced the capacity of G␤ 1 ␥ 2 to activate PLC-⑀ upon coexpression in COS-7 cells (Fig. 3 A). Activation of PI3-kinase-␥ by G␤␥ is not likely involved in activation of PLC-⑀, because neither wortmannin nor LY294002, which are known inhibitors of PI3-kinase-␥ (33, 34), inhibited G␤ 1 ␥ 2stimulated [ 3 H]inositol phosphate accumulation (Fig. 3B).
Ras stimulation of PLC-⑀ is dependent on intact RA domains (17). Therefore, a PLC-⑀ construct with two point mutations (K2150E and K2152E) in the RA domain nearest the COOH terminus (17) was used to determine whether the effects of G␤␥ on the enzymatic activity of PLC-⑀ are indirect and due to activation of Ras. Cotransfection of COS-7 cells with G12Vactivated H-Ras and wild-type PLC-⑀ but not with mutant PLC-⑀ resulted in marked increases in inositol phosphate accumulation (Fig. 4, inset). In contrast, coexpression of G␤ 1 ␥ 2 with either wild-type PLC-⑀ or PLC-⑀ with mutated RA domain resulted in similar levels of G␤␥-stimulated activity (Fig. 4). Thus, the action of G␤␥ on PLC-⑀ is apparently a separate event from the activation of this isozyme by Ras. DISCUSSION Identification of a PH domain and EF-hand domain in rat, human, and C. elegans PLC-⑀ orthologues and demonstration of G␤␥-mediated activation of rat PLC-⑀ suggest an additional mechanism for the regulation of this PLC isozyme. The recently established dual activation of PLC-⑀ by Ras and by the G12 family of heterotrimeric G protein ␣-subunits now must be expanded to include potential input from all G protein-coupled receptors through release of G␤␥.
PLC-⑀ initially was reported as a PLC isozyme that possesses the TIM-barrel X and Y domains and C2 domains conserved in the PLC-␤, -␥, and -␦ isozyme families. However, PLC-⑀ is notably unique by virtue of a large amino-terminal region that contains a CDC25 homology domain and by a carboxyl-terminal region consisting largely of two RA domains. Our illustration here of a PH domain and EF-hand domain interspersed in the large amino terminus between the CDC25 homology domain and X domain adds an additional commonality in structure with the other PLC isozyme families.
Ras apparently activates PLC-⑀, in part, by promoting its translocation to the plasma membrane (17). Consequently, it will be important to resolve whether G␤␥ activates PLC-⑀ by recruitment of the isozyme to the plasma membrane, by direct stimulation of catalytic activity, or by both mechanisms. Activation of PLC-⑀ by G␤␥ apparently is independent of PI3kinase-␥ or the Ras signaling pathways. Our working hypothesis is that activation by both G␤␥ and G␣ 12 /G␣ 13 is direct and requires independent domains in the enzyme. However, lack of additivity in activation by G␤ 1 ␥ 2 and GTPase mutants of G␣ 12 and G␣ 13 may presage involvement of interacting or mutually exclusive domains in the activation by G␤␥-and G␣-subunits. Neither GTP␥S-bound Ras nor G␤ 1 ␥ 2 -activated purified PLC-⑀ when mixed with [ 3 H]PtdIns(4,5)P 2 -containing vesicles under conditions in which marked activation of PLC-␤2 by G␤ 1 ␥ 2 was observed (17). Our data illustrate robust G␤␥-promoted phosphoinositide hydrolysis by PLC-⑀, but whether G␤␥ modulates the catalytic activity of the PLC-⑀ CDC25 homology domain, which serves as a guanine nucleotide exchange factor for Ras GTPases (18,19), also will be important to establish.
The presence of a PH domain in PLC-⑀ together with the well established interactions of G␤␥ with PH domains of GRK2 and GRK3 (15,35,36), Bruton's tyrosine kinase, IRS-1, and other proteins (37,38) prompted the testing of G␤␥ as a potential activator of PLC-⑀. Although our studies revealed a marked stimulation of PLC-⑀ by G␤␥, they did not address the direct role of the PH domain in this activation. Isozymes of the PLC-␤, PLC-␥, and PLC-␦ families all contain PH domains (15). However, only PLC-␤2 and PLC-␤3 are robustly activated by G␤␥, whereas PLC-␤1, PLC-␤4, and members of the PLC-␥ and PLC-␦ families, all of which possess PH domains, are not. That the PH domain is involved in activation of PLC-␤2 by G␤␥ is suggested by the observation that a chimeric PLC-␦ containing the PH domain of PLC-␤2 is activated by G␤␥ (35). Similarly, sequence in PLC-␤3 carboxyl-terminal to the PH domain was shown by Barr et al. (36) to be important for interaction with and activation by G␤␥. This region exhibits similarity to the basic residues in the extended PH domain of GRK2, i.e. the molten helix, involved in interaction with G␤␥ (28,29). Other studies have illustrated the importance of sequence in the Y-domain of PLC-␤2 in G␤␥-promoted activation (39), and therefore the precise structural determinants for activation of PLC-␤ isozymes by G␤␥ remain to be resolved.
In summary, observation of activation of PLC-⑀ by G␤␥ extends the possibilities for activation of this unique inositol lipid-hydrolyzing effector protein to a broad range of heptahelical receptors for extracellular stimuli. This ubiquitously expressed multifunctional protein may provide a major point of integration between heterotrimeric G protein and Ras GTPase signaling pathways.