Pleiotropic Effects of Pasteurella multocida Toxin Are Mediated by G q -dependent and -independent Mechanisms INVOLVEMENT OF G q BUT NOT G 11 *

Pasteurella multocida toxin (PMT) is a highly potent mitogen for a variety of cell types. PMT has been shown to induce various cellular signaling processes, and it has been suggested to function through the heterotrimeric G-proteins G q /G 11 . To analyze the role of G q /G 11 in the action of PMT, we have studied the effect of the toxin in G a q /G a 11 double-deficient fibroblasts as well as in fibro- blasts lacking only G a q or G a 11 . Interestingly, formation of inositol phosphates in response to PMT was exclu-sively dependent on G a q but not on the closely related G a 11 . Although G a q /G a 11 double-deficient and G a q -defi- cient cells did not respond with any production of inositol phosphates to PMT, PMT was still able to induce various other cellular effects in these cells, including the activation of Rho, the Rho-dependent formation of actin stress fibers and focal adhesions, as well as the stimulation of c-Jun N-terminal kinase and extracellular signal-regulated kinase. These data show that PMT leads to a variety of cellular effects that are mediated only in part by the heterotrimeric G-protein G q . Pasteurella

. Exposure of cells to PMT results in tyrosine phosphorylation of various proteins including focal adhesion kinase and paxillin as well as in actin stress fiber formation and focal contact assembly (9,12,13). Several lines of evidence suggest that some of these effects are mediated by the small GTPbinding protein Rho, which plays a major role in actin cytoskeleton dynamics (14). Disruption of Rho function by the C3 exoenzyme of Clostridium botulinum, which ADP-ribosylates and thereby inhibits Rho abolishes focal contact formation in response to PMT, and incubation of endothelial cells with C3 exoenzyme blocks PMT-induced actin stress fiber formation (12,13).
PMT has also been shown to induce a robust increase in inositol phosphate levels, mobilization of intracellularly stored calcium, production of diacyl glycerol, and activation of protein kinase C, suggesting that it leads to an activation of phospholipase C (10,(15)(16)(17)(18). PMT potentiates the production of inositol phosphates induced by various agonists that function through receptors coupling to G-proteins of the G q/11 family, and PMTinduced formation of inositol phosphates can be inhibited by guanosine 5Ј-O-(␤-thiodiphosphate) (18). It has therefore been proposed that heterotrimeric G-proteins of the G q/11 family may be involved in the action of PMT. The G q/11 family contains four members, of which two, G q and G 11 , are expressed in almost all tissues of the mammalian organism and couple heptahelical receptors in a stimulatory fashion to ␤-isoforms of phospholipase C (18). Further evidence for a possible role of G q /G 11 in cellular effects of PMT came from studies in Xenopus oocytes. A PMT-induced Ca 2ϩ -dependent chloride current could be suppressed by injection of a G␣ q antisense RNA and an antiserum recognizing both G␣ q and G␣ 11 (20). In addition, PMT-induced phosphorylation of ERK-1 was reduced by expression of a Cterminal peptide of G␣ q that is believed to interfere with receptor-G q interaction (21).
To determine the exact role of G q /G 11 in various cellular responses of PMT, we have studied the effect of PMT in G␣ q / G␣ 11 double-deficient fibroblasts as well as in fibroblasts lacking only G␣ q or G␣ 11 . Surprisingly, we found that the formation of inositol phosphates in response to PMT is dependent on G␣ q but not on the closely related G␣ 11 . In addition, exposure of cells to PMT induced Rho activation, Rho-dependent stress fiber formation, and activation of MAP kinases in a manner independent of G␣ q /G␣ 11 . These data show that PMT leads to pleiotropic effects in a G␣ q -dependent and -independent manner. zyme was a donation from I. Just and K. Aktories (Freiburg, Germany) or was purchased from Upstate Biotechnology. PMT was purchased from Sigma.
Cell Culture-Wild-type fibroblasts and fibroblasts lacking both Gprotein ␣-subunits were derived from embryonic day 10.5 mouse embryos originating from intercrosses of G␣ q (Ϫ/ϩ) and G␣ 11 (Ϫ/ϩ) mice. The generation of G␣ q and G␣ 11 mutant mice has been described previously (22,23). Fibroblasts lacking G-protein ␣-subunits were prepared and cultured as described previously (24).
Microinjection-For microinjection studies, cells were seeded at a density of ϳ10 3 cells/mm 2 on glass coverslips imprinted with squares to facilitate the localization of injected cells and grown overnight. To obtain quiescent and serum-starved fibroblasts, cultures were rinsed in serum-free DMEM and incubated in DMEM supplemented with 25% Ham's F-12 medium, 0.2% NaHCO 3 , 10 mM Hepes, and 0.1% fetal bovine serum (modified DMEM) for 24 h, followed by a 48-h incubation in modified DMEM devoid of fetal bovine serum. Plasmids were injected into the nucleus together with Texas Red dextran (5 mg/ml; Molecular Probes) to visualize injected cells. C. botulinum C3 exoenzyme was comicroinjected with the cDNAs at a concentration of 100 g/ml. About 150 cells/field were injected in each case, using a manual injection system (Eppendorf, Hamburg, Germany).
Visualization of Actin Cytoskeleton-Microinjected cells were stimulated with 100 ng/ml PMT overnight, fixed in 4% paraformaldehyde for 20 min, and permeabilized in 0.2% Triton X-100 for 5 min. To visualize the cytoskeleton, cells were stained for polymerized actin by incubation with 0.5 g/ml fluorescein isothiocyanate-phalloidin (Sigma) for 40 min. The coverslips were mounted on glass slides and examined using an inverted microscope (Zeiss Axiovert 100). Quantification of actin stress fibers was performed as described (24).
Determination of Inositol Phosphate Levels-Cells were labeled for 20 -24 h with 120 pmol of myo-[2-3 H]inositol (758.5 Gbq/mmole; PerkinElmer Life Sciences)/well in the absence or presence of PMT. For determination of receptor-mediated inositol phosphate production, cells were washed with inositol-free medium and then incubated for 10 min at 37°C with 0.25 ml of inositol-free medium containing 10 mM of LiCl. Thereafter, medium was aspirated, the indicated agents were added in medium containing 10 mM LiCl, and cells were incubated for 20 min. Inositol phosphate production was stopped by addition of 0.2 ml of 10 mM ice-cold formic acid. After keeping the samples on ice for 20 min, 0.45 ml of 10 mM NH 4 OH was added, and the whole sample was loaded onto a column containing 0.75 ml of anion exchange resin (AG 1-X8; Bio-Rad) equilibrated with 5 mM borax and 60 mM sodium formate. Total inositol phosphates were then separated and measured as described (25). If not stated otherwise, measurements were done in triplicates.
Determination of Activated Cellular RhoA-The amount of activated cellular Rho was determined by precipitation with a fusion protein consisting of GST and the Rho-binding domain of Rhotekin (amino acids 7-89; GST-Rho-binding domain) as described (26). Cells were washed with ice-cold Hank's buffer and lysed in RIPA buffer (50 mM Tris, pH 7.2, 1%, Triton X-100, 0.5% sodium deoxycholate, 0.1% SDS, 500 mM NaCl, 10 mM MgCl 2 , 10 g/ml each of leupeptin and aprotinin, and 1 mM PMSF). Clarified cell lysates were incubated with GST-Rho-binding domain (20 g of beads at 4°C for 45 min. The beads were washed four times as described (26), and the precipitated Rho was detected by Western blotting using a monoclonal antibody against RhoA (Santa Cruz Biotechnology).
Determination of ERK Phosphorylation and c-Jun Kinase Activity-For determination of ERK phosphorylation, serum-starved (48 h 0.5% fetal calf serum, 24 h 0.1% fetal calf serum) cells grown in 12-well dishes were washed once with phosphate-buffered saline and lysed in Laemmli sample buffer. Cell lysates were separated by SDS-polyacrylamide gelelectrophoresis, and phosphorylation of ERK was determined by immunoblotting with an anti-phospho-ERK antiserum (New England Biolabs). Blots were reprobed with an anti-ERK antiserum (New England Biolabs).
c-Jun kinase activity was determined in a solid phase assay using GST-c-Jun as a substrate (27,28). GST-c-Jun phosphorylated in the presence of [␥-32 P]ATP was subjected to SDS-polyacrylamide gelelectrophoresis, and phosphorylation of c-Jun was determined by autoradiography of dried gels (29). JNK1 and JNK2 were detected with an anti-JNK-antiserum (Santa Cruz Biotechnologies).

RESULTS
For studies on the possible role of G q /G 11 in the cellular effects of PMT, we employed fibroblast cell lines derived from mouse embryos deficient in either G␣ q or G␣ 11 or lacking both G-protein ␣-subunits. The absence or presence of G␣ q and G␣ 11 was verified by immunoblotting ( Fig. 1). Treatment of wildtype mouse fibroblasts with 100 ng/ml PMT for increasing time periods resulted in a marked and time-dependent accumulation of inositol phosphates that could be observed 8 h after addition of the toxin and reached a maximum after about 20 h ( Fig. 2A). In contrast, incubation of G␣ q /G␣ 11 double-deficient fibroblasts for various time periods did not result in any increase in the formation of inositol phosphates ( Fig. 2A). Inositol phosphate production in wild-type cells could be induced with 10 ng/ml PMT and increased dose-dependently up to a concentration of 1000 ng/ml of the toxin (Fig. 2B). However, even at PMT concentrations that were maximally effective in wild-type fibroblasts, no effect on inositol phosphate levels could be observed in G␣ q /G␣ 11 double-deficient fibroblasts (Fig. 2B). This indicates that G-proteins of the G q /G 11 -family are indeed required for PMT-induced inositol phosphate formation.
The ␣-subunits of G q and G 11 are highly homologous, and so far no functional differences between G␣ q and G␣ 11 either with regard to their activation through receptors or their regulation of effectors have been reported. To test whether both G q and G 11 are involved in PMT-induced formation of inositol phosphates, we tested the effect of PMT on inositol phosphate production in cells that lack only G␣ q or G␣ 11 (Fig. 3). The expression of either G q or G 11 was sufficient to mediate receptordependent phospholipase C activation because inositol phosphate production could be induced by thrombin and bradykinin in G␣ q (Ϫ/Ϫ) as well as in G␣ 11 (Ϫ/Ϫ) cells, but not in G␣ q /G␣ 11 double-deficient fibroblasts (Fig. 3). However, although G␣ 11deficient fibroblasts still responded with an increased inositol phosphate production to PMT, G␣ q -deficient cells behaved like G␣ q /G␣ 11 double-deficient fibroblasts and were completely unresponsive, indicating that the effect of PMT was mediated solely by G␣ q and not by G␣ 11 .
PMT has also been shown to induce a Rho-dependent actin stress fiber formation and focal adhesion formation in fibroblasts and endothelial cells (9,12,13). There are conflicting data as to the ability of G q /G 11 to induce the formation of actin stress fibers (30,31). However, in some systems G q /G 11 have been shown to be able to regulate Rho-dependent processes (32)(33)(34). To study the involvement of G q /G 11 in PMT-induced actin stress fiber formation and focal adhesion assembly, we tested the effect of the toxin on the actin cytoskeleton in wildtype and G␣ q /G␣ 11 double-deficient fibroblasts (Fig. 4). Actin filaments were visualized by fluorescein isothiocyanate-labeled phalloidin, and focal adhesions were stained with an antivinculin antibody. In serum-starved fibroblasts lacking G␣ q and G␣ 11 , PMT induced a pronounced formation of actin stress FIG. 1. Expression of G␣ q and G␣ 11 in fibroblasts derived from G␣-deficient mouse embryos. Shown is an immunoblot of cholate extracts from plasma membranes of fibroblasts derived from wild-type, G␣ q -, G␣ 11 -, and G␣ q /G␣ 11 -deficient embryos. Shown is an auto-luminogram of a blot stained with an antiserum recognizing both G␣ q and G␣ 11 . The position of the respective G-protein ␣-subunit is indicated on the left.
fibers and a formation of focal adhesions indistinguishable from its effect in wild-type cells (Fig. 4). This indicates that the PMT-induced reorganization of the actin cytoskeleton and focal adhesion assembly occurred independently of the G q /G 11 -mediated signaling pathway. Preincubation of cells with pertussis toxin did not affect PMT-induced actin stress fiber formation, indicating that G-proteins of the G i family were not involved. Induction of actin stress fiber formation through various receptors in fibroblasts has been shown to involve a Rho/Rho kinasemediated signaling pathway (35)(36)(37). The actin stress fiber formation by PMT in G␣ q /G␣ 11 double-deficient fibroblasts could be blocked by cytosolic injection of C3 exoenzyme of C. botulinum, which ADP-ribosylates Rho at residue Asn 41  To test whether PMT indeed induces the activation of Rho, we directly determined Rho activation by precipitation of endogenous GTP-bound Rho using a fusion protein consisting of glutathione S-transferase and the Rho-binding domain of rhotekin (26). As shown in Fig. 6 PMT induced a pronounced activation of Rho that was indistinguishable in wild-type and G␣ q /G␣ 11 double-deficient cells, indicating that PMT-induced Rho activation is independent of the G q /G 11 -dependent signaling pathway. Similar to the G q -mediated PMT-induced inositol phosphate production, the G q /G 11 -independent activation of Rho could only be observed after a lag period of several hours following exposure of cells to PMT (data not shown).
Various G q /G 11 -coupled receptors have been demonstrated to mediate the activation of JNK and ERK (41)(42)(43)(44)(45)(46), and constitutively active mutants of G q -family members have been shown to stimulate JNK activity in some cellular systems (47,48). In addition, PMT has been shown to induce stimulation of ERK, an effect that could be inhibited by a dominant negative G␣ q mutant (21). To delineate the role of G␣ q /G␣ 11 in PMT-induced MAP kinase activation, we compared its effect on JNK and ERK activity in wild-type and G␣ q /G␣ 11 double-deficient fibroblasts (Fig. 7). PMT induced activation of JNK as well as of ERK in both wild-type and G␣ q /G␣ 11 double-deficient fibroblasts, demonstrating that PMT leads to JNK and ERK activation in a manner independent of G q . DISCUSSION PMT has been shown to induce a variety of cellular effects. The precise molecular mechanism by which PMT acts is, however, still poorly defined. The toxin has only moderate homology to other proteins, and so far no enzymatic activity has been detected. Similar to many other toxins, PMT requires internalization and intracellular processing to exhibit cellular effects. This results in a lag period of a few hours between exposure of PMT to intact cells and the occurrence of cellular changes (16).
It has been suggested that G-proteins of the G q/11 family are involved in the cellular action of PMT. The two main members of this family, G q and G 11 , are structurally and functionally highly homologous and couple receptors in a stimulatory fashion to ␤-isoforms of phospholipase C (19,49). An antiserum recognizing the ␣-subunits of both G␣ q and G␣ 11 blocks a PMT-induced Ca 2ϩ -dependent Cl Ϫ current in Xenopus oocytes, which involves PLC-␤ (20). Stimulation of this current by PMT could also be inhibited by the injection of G␣ q antisense RNA, whereas sense RNA potentiated the effect of PMT (20). These data were obtained after injection of PMT into oocytes, which results in a rapid response within seconds after injection. This, however, is a situation completely different from the action of the toxin on intact cells that requires internalization and proc-essing of PMT. In a recent study using HEK-293 cells, it was shown that PMT-induced activation of Erk-1 can be reduced by about 70 -80% upon expression of a C-terminal fragment of G␣ q that is supposed to act in a dominant negative fashion (21). Although these studies support an involvement of G q /G 11 in some of the cellular effects of PMT, the evidence provided remains indirect.
To study the role of G-proteins of the G q /G 11 family in the cellular action of PMT, we used fibroblast cell lines derived from mouse embryos that are deficient in G␣ q /G␣ 11 . In wildtype embryonic fibroblasts PMT induced a robust time-and dose-dependent increase in the production of inositol phosphates that could not be observed in fibroblasts lacking both G␣ q and G␣ 11 (Figs. 2 and 3). Embryonic fibroblasts lacking only G␣ q did not respond to PMT with inositol phosphate production, whereas PMT lead to a strong response in G␣ 11 -deficient cells, indicating that PMT-induced inositol phosphate production is mediated by G q and not by G 11 . This is surprising because evidence collected from biochemical, pharmacological, and somatic cell genetic studies suggested that G␣ q and G␣ 11 have very similar, if not identical, characteristics. G␣ q and G␣ 11 couple to the same set of seven transmembrane receptors with the same effector specificity for phospholipase C-␤ isoforms (50 -54).
Although our results in G␣ q /G␣ 11 -deficient cells clearly show that the G q /PLC-␤ pathway plays an important role in the action of PMT, it has been suggested that PMT can also act independently of PLC-mediated Ca 2ϩ mobilization and protein kinase C activation (12). We therefore tested whether PMT can still induce other cellular effects in the absence of G q -dependent signaling. Exposure of G␣ q /G␣ 11 -deficient embryonic fibroblasts to PMT resulted in actin stress fiber formation and focal adhesion assembly (Figs. 4 and 5). Actin stress fiber formation was inhibited by C3 exoenzyme of C. botulinum as well as by a dominant negative form of Rho kinase and the Rho kinase inhibitor Y-27632. This indicates that a Rho/Rho kinase mediated but G q /PLC-␤-independent pathway is involved in this cellular response to PMT. A Rho/Rho kinase-mediated pathway resulting in the inhibition of myosin phosphatase and subsequent increase in myosin light chain phosphorylation has recently been proposed to underlie PMT-induced reorganization of the actin cytoskeleton in endothelial cells (13). Actin rearrangement induced by PMT in endothelial cells could be completely blocked by an inhibitor of the Ca 2ϩ /calmodulin-regulated myosin light chain kinase, suggesting that dual regulation of myosin light chain phosphorylation through Ca 2ϩdependent myosin light chain kinase activation and Rho/Rho kinase-mediated myosin phosphatase inhibition is involved in the effect of PMT on the actin cytoskeleton. Our data, however, suggest that the inhibition of myosin phosphatase through Rho/Rho kinase is sufficient to induce a rearrangement of the actin cytoskeleton in embryonic fibroblasts because it can be observed in the absence of a G q -mediated inositol phosphate production and subsequent Ca 2ϩ mobilization.
The involvement of Rho in PMT-induced cellular effects could be directly demonstrated by precipitation of active Rho from cell lysates of PMT-exposed wild-type and G␣ q /G␣ 11 -deficient embryonic fibroblasts. Thus, activation of Rho by PMT occurs independently of G␣ q /G␣ 11 . The N-terminal half of cytotoxic necrotizing factor (CNF) 1 and 2 from Escherichia coli show moderate homology with N-terminal regions of PMT, and CNF1 has been shown to inhibit the GTPase activity of RhoA by deamidation of glutamine residue 63 resulting in constitutive activation of Rho (55,56). However, the catalytic activity of CNF1 appears to reside in the C-terminal part of the toxin (57,58), and it is unlikely that PMT functions analogous to CNF1 (12,13,20). PMT may act upstream of Rho by regulating the activity of a guanine nucleotide exchange factor or a GTPaseactivating protein specific for Rho. G-proteins of the G 12 family that have been shown to be able to mediate Rho activation (30,59,60) are not involved because PMT-induced Rho activation could also observed in fibroblasts lacking G␣ 12 /G␣ 13 (data not shown).
Various MAP kinases including JNK and ERK have been shown to be regulated through G q /G 11 -coupled receptors, and PMT has been reported to activate ERK via a pathway involving the epidermal growth factor receptor (21). Incubation of wild-type fibroblasts with PMT resulted in a clear JNK and ERK activation. This effect of PMT obviously did not involve G q /G 11 because G␣ q /G␣ 11 -deficient cells also responded with activation of JNK and ERK after exposure to PMT to a comparable extent as did wild-type cells (Fig. 7). It has previously been suggested that PMT-induced ERK activation is mediated by G q /G 11 (21). Our data demonstrate that G q /G 11 are not required for JNK and ERK activation by PMT in embryonic fibroblasts. However, we cannot exclude the possibility that G q contributes to PMT-induced MAP kinase activation in wildtype cells.
In summary, we show that PMT induces a remarkable array of cellular effects including the activation of phospholipase C, which is entirely dependent on G q but not on the closely related G-protein G 11 . However, the pleiotropic actions of PMT are only in part mediated by G q . Activation of Rho and MAP kinases can be induced by PMT in a G q /G 11 -independent manner, suggest-ing that other, possibly G-protein independent processes are involved in the cellular action of PMT.