Pertussis toxin-sensitive activation of phospholipase C by the C5a and fMet-Leu-Phe receptors.

Signal transduction pathways that mediate C5a and fMet-Leu-Phe (fMLP)-induced pertussis toxin (PTx)-sensitive activation of phospholipase C (PLC) have been investigated using a cotransfection assay system in COS-7 cells. The abilities of the receptors for C5a and fMLP to activate PLC β2 and PLC β3 through the Gβγ subunits of endogenous Gi proteins in COS-7 cells were tested because both PLC β2 and PLC β3 were shown to be activated by the βγ subunits of G proteins in in vitro reconstitution assays. Neither of the receptors can activate endogenous PLC β3 or recombinant PLC β3 in transfected COS-7 cells. However, both receptors can clearly activate PLC β2 in a PTx-sensitive manner, suggesting that the receptors may interact with endogenous PTx-sensitive G proteins and activate PLC β2 probably through the Gβγ subunits. These findings were further corroborated by the results that PLC β3 could only be slightly activated by Gβ1γ1 or Gβ1γ5 in the cotransfection assay, whereas the Gβγ subunits strongly activated PLC β2 under the same conditions. PLC β3 can be activated by Gαq, Gα11, and Gα16 in the cotransfection assay. In addition, the Gγ2 and Gγ3 mutants with substitution of the C-terminal Cys residue by a Ser residue, which can inhibit wild type Gβγ-mediated activation of PLC β2, were able to inhibit C5a or fMLP-mediated activation of PLC β2. These Gγ mutants, however, showed little effect on m1-muscarinic receptor-mediated PLC activation, which is mediated by the Gq class of G proteins. These results all confirm that the Gβγ subunits are involved in PLC β2 activation by the two chemoattractant receptors and suggest that in COS-7 cells activation of PLC β3 by Gβγ may not be the primary pathway for the receptors.

Heterotrimeric GTP-binding protein (G protein) 1 -mediated signal transduction pathways are involved in a variety of biological processes, ranging from neuronal activities, metabolism, hematopoietic functions, to some sensory processes (1)(2)(3). These pathways can be divided into two groups based on their sensitivities to Pertussis toxin (PTx). PTx is a bacterial toxin that catalyzes ribosylation of C-terminal cysteine residues of some G␣ subunits, including the G␣ i subunits, G␣ o subunits, and transducin ␣ subunits. Modification by PTx prevents the interaction between G␣ subunits and receptors, thus blocking ligand-mediated signal transduction (1,4). The G␣ subunits of the G q and G s class of G proteins lack the C-terminal cysteine residues; hence, signal transduction pathways mediated by these G␣ subunits are PTx-resistant (3).
Many G protein-coupled receptors transduce their signals through the activation of phospholipase C (PLC). Some receptors, such as the ␣ 1 -adrenergic (5,6) and m 1 -muscarinic cholinergic receptors (7), act mainly through the G q class of G proteins and are, thus, resistant to PTx treatment. Other receptors appear to activate PLC in a PTx-sensitive manner. Typical examples are found in leukocytes, where responses to a number of chemoattractants, including interleukin-8 (IL-8), C5a, and f-Met-Leu-Phe (fMLP), are mostly PTx-sensitive (8 -16). A mechanism involved in the PTx-sensitive processes has recently been proposed; ligand-bound receptors may interact with PTx-sensitive G proteins, such as the G i proteins, and release G␤␥, which then activates PLC. This hypothesis is based on the findings that the G␤␥ subunits of G proteins can activate certain isoforms of PLC ␤, while the G␣ i subunits cannot. The G␤␥ subunits were shown to activate PLC ␤2 but not PLC ␤1 or PLC ␤4 in a cotransfection assay (17)(18)(19) and activate PLC ␤3 and PLC ␤2 in reconstitution assays with purified proteins (20 -25). Our previous report on reconstitution of PTx-sensitive, IL-8-induced activation of PLC ␤2 in cotransfected COS-7 cells supports the hypothesis that the chemokine receptor acts through G␤␥ activation of the PLC ␤2 isoform (26).
In this report, we investigated the signal transduction pathways for the C5a and fMLP receptors, which play important roles in inflammation (8). C5a and fMLP receptors have previously been shown to couple selectively to PTx-insensitive G␣ subunit, G␣ 16 to activate PLC (27,28). There must, however, also be a distinct pathway that mediates the PTx-sensitive responses to fMLP and C5a. The PTx-sensitive pathways may be the predominant ones in mature leukocytes, because responses to chemoattractants were found to be largely PTxsensitive in these cells. Since C5a and fMLP, like IL-8, induce Ca 2ϩ efflux and leukocyte chemotaxis that are sensitive to PTx treatment, the C5a and fMLP receptors may utilize the same signal transduction pathways as the IL-8 receptors (26). By using the cotransfection assay, we found that these two receptors can specifically activate PLC ␤2 but not PLC ␤3, presumably through the G␤␥ subunits released from the G i proteins. The finding that the G␥ 2 and G␥ 3 mutants, with substitution of the C-terminal Cys residues by Ser residues, can act as dominant negative inhibitors to block G␤␥-mediated activation of PLC ␤2 in cotransfected COS-7 cells supports the notion that the G␤␥ subunits are involved in the signal transduction processes of these chemoattractant receptors.

MATERIALS AND METHODS
Cell Culture and Transfection-Cos-7 cells were cultured in Dulbecco's modified Eagle's medium containing 10% fetal calf serum under 5% CO 2 at 37°C. For transfection, Cos-7 cells were seeded into 24-well plates at a density of 1 ϫ 10 5 cells/well the day before transfection (32)(33)(34). The media were removed the next day, and 0.25 ml of Opti-MEM (Life Technologies, Inc.) containing 2 l of lipofectamine (Life Technologies, Inc.) and 0.5 g of plasmid DNA were added to each well. Five hours later the transfection media were replaced by the culture media. Then the cells were labeled with 10 Ci/ml myo[2-3 H]inositol the following day, and the levels of inositol phosphates were determined 1 day later as described previously (29). All the cDNAs used in these studies were constructed into expression vectors that were driven by the cytomegalovirus promoters (29).
SDS-Polyacrylamide Gel Electrophoresis and Western Blot--For Western analysis of protein expression in COS-7 cells, equal numbers of cells were directly solubilized in SDS sample buffer. For analysis of G␥ expression, equal numbers of cells were incubated with phosphatebuffered saline containing 1 mM EDTA for 5 min, and they were washed off the plates and collected into microcentrifuge tubes. After brief spinning, the cells were subjected to two freeze-thaw cycles in a hypotonic buffer (10 mM Tris, pH 7.4) containing various protease inhibitors. Then the cytosolic and particulate fractions were separated by high-speed centrifugation. Finally the samples were solubilized in or diluted with the SDS sample buffer and loaded to 12% SDS-polyacrylamide gels. The proteins were subsequently electroblotted onto nitrocellulose membranes and detected with antibodies indicated in the figure legends.
Construction of G␥ Mutants-The G␥ 3CS mutant was generated by polymerase chain reaction with the high fidelity DNA polymerase, pfu (Stratagene), and the mutation was confirmed by DNA sequencing. The G␥ 1CS and G␥ 2CS mutants were kindly provided by Dr. Narasimthan Gautam from the Washington University, MO.
Ligand-binding Assays--COS-7 cells in 24-well plates were cotransfected with cDNAs encoding the receptors and various proteins including PLCs, G␤, G␥, or ␤-galactosidase. Varying amounts of 125 I-labeled C5a (2200 Ci/mmol, DuPont NEN) were incubated with the transfectants for 1 h on ice. Then the cells were washed three times with cold phosphate-buffered saline containing 0.5% bovine serum albumin and solubilized in 0.1 N sodium hydroxide. Aliquots were counted in a ␥-counter. The values derived from cells transfected with the ␤-galactosidase DNA were taken as nonspecific binding. The maximal ligandbinding sites and affinities were determined by Scatchard analysis. RESULTS We used the cotransfection assay in COS-7 cells to characterize signal transduction pathways that mediate PTx-sensitive PLC activation by the receptors for C5a and fMLP. COS-7 cells transfected with the cDNA encoding the C5a receptor or fMLP receptor alone showed no ligand-dependent accumulation of IPs ( Fig. 1), confirming the report that the C5a and fMLP receptors do not couple to G␣ q/11 (27) (COS-7 cells contain endogenous G␣ q/11 but not G␣ 16 (26)). However, when the cells were cotransfected with the PLC ␤2 cDNA and the cDNA encoding the C5a or fMLP receptor, there were marked ligandinduced increases in accumulation of IPs, and these ligandinduced responses were sensitive to PTx (Fig. 1). Cells cotransfected with the cDNA encoding PLC ␤1 instead did not show any ligand-induced responses (Fig. 1). Knowing that G␤␥ can only activate PLC ␤2, but not PLC ␤1 in the cotransfection system, and that COS-7 cells contain PLC ␤1, but not PLC ␤2 (17, 18), we interpret this result to conclude that the receptors for C5a and fMLP may interact with endogenous PTx-sensitive G proteins, presumably G i2 proteins, and cause release of the ␤␥ subunits, which activate the recombinant PLC ␤2. We assume that the interaction is with G i2 because only G i2 , but not other known PTx-sensitive G proteins including the G i1 , G i3 , or G o proteins, was detected in COS-7 cells (17).
Since COS-7 cells contain endogenous PLC ␤3 (18), we predict that the C5a or fMLP receptor would induce IP accumulation by activating endogenous PLC ␤3 through G␤␥. However, this did not occur (Fig. 1), suggesting that the G␤␥ subunits released from the endogenous G i proteins were unable to activate the endogenous PLC ␤3. To investigate whether the C5a or fMLP receptor can activate the recombinant PLC ␤3, we cotransfected COS-7 cells with cDNAs encoding PLC ␤3 and the C5a or fMLP receptor. No ligand-induced accumulation of IPs was observed (Fig. 1), indicating that the receptors cannot activate the recombinant PLC ␤3 either. To test whether the recombinant PLC ␤3 can be activated by recombinant G protein subunits, we cotransfected COS-7 cells with cDNA encoding ␤-galactosidase (as control), G␣ q , G␣ 11 , G␣ 16 , G␤ 1 ␥ 1 , or G␤ 1 ␥ 5 and cDNA encoding PLC ␤3 as well as other PLC ␤ isoforms as controls. As shown in Fig. 2, the recombinant PLC ␤3 as well as PLC ␤1 and ␤2 can all be activated by G␣ q , G␣ 11 , or G␣ 16 (Fig. 2) as cells cotransfected with cDNAs encoding PLC ␤ and G␣ q , G␣ 11 , or G␣ 16 showed marked accumulation of IPs over those transfected with PLC ␤ or G␣ alone. However, G␤ 1 ␥ 1 or G␤ 1 ␥ 5 showed only weak activation of the recombinant PLC ␤3, whereas the G␤␥ subunits markedly activated the recombinant PLC ␤2 (Fig. 2). The G␤␥ subunits did not activate PLC ␤1 or PLC␤4 (Fig. 2), as we have demonstrated previously (17)(18)(19). The weak activation of PLC ␤3 by G␤␥ may explain why the C5a or fMLP receptors could not activate PLC ␤3.
The expression levels of PLC ␤3 were determined. The levels of the recombinant PLC ␤3 are at least 5-fold higher than those of the endogenous PLC ␤3, and the levels of the recombinant PLC ␤3 in various transfectants are rather constant (Fig. 2B). The ligand-binding sites on cells expressing the C5a receptor were also determined using 125 I-labeled C5a. Cells coexpressing the C5a receptor and ␤-galactosidase, PLC ␤3, or PLC ␤2 showed similar numbers of ligand-binding sites (300 -375 fmol per 1 ϫ 10 5 cells) with affinities of 2.5-4 nM. Thus, the inabilities of G␤␥ or the C5a receptor to activate PLC ␤3 were not the result of variations in protein expression. Furthermore, we compared the expression level of the recombinant PLC ␤3 with that of PLC ␤2 in transfected COS cells. Dilutions of cell extracts from cells expressing recombinant PLC ␤2 or PLC ␤3, together with dilutions of purified PLC ␤2 or ␤3 proteins with defined protein concentrations, were analyzed by Western blotting with antibodies specific to PLC ␤2 or ␤3. Based on the signal intensities, we estimated that the expression levels of  (Fig. 2C). Therefore, it is reasonable to conclude that G␤␥ is a poor activator for PLC ␤3 when compared with PLC ␤2 in the cotransfection assay.
In order to obtain more evidence for involvement of G␤␥ in C5a-or fMLP-mediated activation of PLC, we investigated whether the G␥ mutants, with substitution of the C-terminal Cys residues by Ser residues, can act as dominant negative mutants to inhibit G␤␥-mediated PLC activation. The G␥ 1 and G␥ 2 mutants, designated G␥ 1CS and G␥ 2CS , were found to be able to bind to the G␤ subunits and bring them into the cytosol (30,31). These G␥ mutants, when paired with the G␤ subunits, failed to activate PLC ␤2 (17,18), suggesting that the complex of G␥ CS and G␤ is nonfunctional. Recently, we introduced the equivalent mutation into G␥ 3 . The G␥ 3 mutant G␥ 3CS , like G␥ 2CS and G␥ 1CS (17), cannot activate PLC ␤2 (Fig. 3A). How-ever, G␥ 3CS differs from G␥ 1CS and G␥ 2CS in its ability to associate with the particulate fractions; G␥ 3CS , unlike G␥ 1CS and G␥ 2CS (17), can still associate with the particulate fraction (Fig. 3B).
The cotransfection assay was used to test whether these G␥ mutants are capable of acting as dominant negative mutants to inhibit G␤␥-mediated effects. We found that cells cotransfected with cDNAs encoding PLC ␤2, G␤ 1 , G␥ 3 , and G␥ 3CS showed little accumulation of IPs compared with those cotransfected with PLC ␤2, G␤ 1 , and G␥ 3 (Fig. 3A), suggesting that G␥ 3CS inhibited G␤ 1 ␥ 3 -mediated activation of PLC ␤2. To test whether the G␥ mutants can block ligand-mediated responses, we cotransfected COS-7 cells with cDNAs encoding PLC ␤2, the C5a receptor, and one of the G␥ mutants. As shown in Fig. 4A, G␥ 2CS and G␥ 3CS were capable of blocking C5a-mediated accumulation of IPs, whereas the wild types and G␥ 1CS could not. Knowing that COS-7 cells contain G␤ 1 and G␤ 2 (17) and that G␥ 1CS cannot interact with G␤ 2 (30,31), we interpreted the G␥ 1CS results to suggest that G␥ 2CS and G␥ 3CS may be able to scavenge most of the G␤ subunits in the cells to prevent them from activating PLC, whereas G␥ 1CS is unable to scavenge G␤ 2 , thus failing to inhibit C5a-mediated effects. We also determined the expression levels of PLC ␤2 and the numbers of C5a-binding sites on cells expressing the C5a receptors. Coexpression of various G␥ subunits did not significantly affect C5a-binding sites on various transfectants (ϳ350 fmol/1 ϫ 10 5 cells) neither did it affect the expression levels of PLC ␤2 (Fig.  4C). Therefore, inhibition of G␤␥- (Fig. 3A) and C5a (Fig. 4A)mediated PLC activation by the G␥2 2,3 mutants is unlikely to result from changes in the expression levels of the proteins involved in activation of PLC ␤2. The same result was also observed for the fMLP receptor, i.e. the G␥ 2,3 mutants can inhibit fMLP-mediated activation of PLC␤2 in transfected COS-7 cells (data not shown). In addition, it is interesting to note that the G␥ mutants did not appear to affect G␣-mediated effector activation. In cotransfected COS-7 cells, G␥ 3CS (Fig.  4B) and G␥ 2CS (data not shown) did not inhibit m 1 -muscarinic receptor-mediated activation of PLC, which is mediated by the G␣ subunits of the G q class (7). Thus, the G␥ mutants, G␥ 2CS and G␥ 3CS , appear to only affect G␤␥-mediated responses but not G␣-mediated responses. In summary, these results support the idea that the G␤␥ subunits mediate fMLP-and C5a-induced activation of PLC ␤2.

DISCUSSION
In this report, we have demonstrated that the C5a and fMLP receptors, like the IL-8 (18) and m 2 -muscarinic receptors (17), can interact with PTx-sensitive G proteins (presumably G i2 ) causing the release of the G␤␥ subunits, which then specifically activate PLC ␤2 in a PTx-sensitive manner. Although these two receptors can also couple to G␣ 16 , we believe that the G␤␥-mediated pathway may be the predominant one that occurs in leukocytes, because the responses to these chemoattractants in mature leukocytes are mostly PTx-sensitive. Our results are also consistent with previous results demonstrating that the C5a and fMLP receptors can interact with the G i2 proteins (35)(36)(37).
Regulation of PLC ␤1, -␤2, and -␤4 by G␤␥ has been tested before in the cotransfection assay (17)(18)(19), but here PLC ␤3 was tested in this assay system for the first time. To our surprise, PLC ␤3 can only be weakly activated by G␤␥ in the cotransfection assay system. G␤␥ was previously shown to activate PLC ␤3 effectively in in vitro reconstitution assays with purified proteins (23)(24)(25). In fact, there are other examples of apparent specificity in the cellular system that are not seen in vitro. For instance, Schultz and his co-workers (49,50) have demonstrated receptor-mediated specificity for G protein heterotrimers that is not evident in the reconstitution system (23). There are probably other factors, such as substrate compartmentalization, modification, and membrane interaction and the involvement of accessory proteins, which may mediate specificity in the cellular system. Nevertheless, the weak activation of PLC ␤3 by G␤␥ explains the inabilities of the chemoattractant receptors to activate endogenous or recombinant PLC ␤3 in COS-7 cells and may explain the lack of significant PLC activation by non-G q -coupling receptors in many systems where PLC ␤3 is expressed. However, activation of PLC ␤3 by G␤␥ may occur in vivo in cells where the expression levels of PLC ␤3 are higher than those of the recombinant in COS-7 cells or where the subcellular localization of PLC ␤3 or production of accessory proteins is differently regulated.
As we have demonstrated, the G␥ mutants, G␥ 2CS and G␥ 3CS in particular, can serve as dominant negative mutants to inhibit G␤␥-mediated, but not G␣-mediated, activation of PLC. This is presumably due to their abilities to form complexes with all known G␤ subunits. Although the G␥ 1CS and G␥ 2CS mutants were demonstrated to be capable of binding to G␤ (30, 31), we do not know whether the complexes of G␥ CS and G␤ can still interact with PLC ␤2. In other words, it is not clear whether the changes in the G␥ mutants impair the abilities of the G␤␥ complexes to interact with PLC ␤2 or their abilities to activate PLC ␤2. However, the ability of the G␥ 3CS mutant to associate with the particulate fractions indicates that the lipid modification on the Cys residue may not serve only as an anchor for G␤␥, but it may also have other functions. The lipid modification at the C-terminal Cys residue may either participate in effector activation or in orientation of the ␤␥ complexes to allow better access of effectors to their substrates on the membranes. It appears paradoxical that the G␥ mutants had no effect on m 1 -muscarinic receptor-mediated activation of PLC because G␤␥ was shown to be essential for reconstitution of m 1 -receptor-mediated activation of PLC in an in vitro system using purified proteins (7,38). We suggest two possible interpretations. 1) There may still be interactions between receptors and G␣ subunits in the absence of G␤␥ depending on the nature of receptors and G proteins, but G␤␥ may greatly facilitate the interactions. In the COS-7 overexpression system, the requirement for G␤␥ may be overcome by the high expression levels of ␣ subunits and receptors. 2) The G␥ mutants did not scavenge all of the G␤ subunits; hence, there may be enough normal G␤␥ complexes present so that G␣-mediated activation of effectors is largely unaffected. The result that more G␤␥ is required for activation of PLC than G␣ supports the second possibility (24,39).
The G␥CS mutants may not only inhibit G␤␥-mediated activation of PLC, but they should also be able to block other G␤␥-mediated regulation of effectors, including adenylylcyclases (23,40), phosphatidylinositol 3-kinase (41), ion channels (42,43), mitogen-activated kinase (44), and ␤-adrenergic receptor kinase (45,46), etc. Therefore, the G␥ CS mutants, joined with the other G␤␥ antagonists, including phosducin (47) and the G␤␥-binding region of ␤-adrenergic receptor kinase (48), provide useful tools to test whether a specific G protein-coupled signal pathway is mediated by G␤␥ in a variety of systems.