Pasteurella multocida Toxin Stimulates Mitogen-activated Protein Kinase via Gq/11-dependent Transactivation of the Epidermal Growth Factor Receptor*

The dermatonecrotic toxin produced byPasteurella multocida is one of the most potent mitogenic substances known for fibroblasts in vitro. Exposure to recombinant P. multocida toxin (rPMT) causes phospholipase C-mediated hydrolysis of inositol phospholipids, calcium mobilization, and activation of protein kinase C via a poorly characterized mechanism involving Gq/11 family heterotrimeric G proteins. To determine whether the regulation of G protein pathways contributes to the mitogenic effects of rPMT, we have examined the mechanism whereby rPMT stimulates the Erk mitogen-activated protein kinase cascade in cultured HEK-293 cells. Treatment with rPMT resulted in a dose and time-dependent increase in Erk 1/2 phosphorylation that paralleled its stimulation of inositol phospholipid hydrolysis. Both rPMT- and α-thrombin receptor- stimulated Erk phosphorylation were selectively blocked by cellular expression of two peptide inhibitors of Gq/11 signaling, the dominant negative mutant G protein-coupled receptor kinase, GRK2(K220R), and the Gαqcarboxyl-terminal peptide, Gαq-(305–359). Like α-thrombin receptor-mediated Erk activation, the effect of rPMT was insensitive to the protein kinase C inhibitor GF109203X, but was blocked by the epidermal growth factor receptor-specific tyrphostin, AG1478 and by dominant negative mutants of mSos1 and Ha-Ras. These data indicate that rPMT employs Gq/11 family heterotrimeric G proteins to induce Ras-dependent Erk activation via protein kinase C-independent “transactivation” of the epidermal growth factor receptor.

Pasteurella multocida is a wide ranging bacterium found in the respiratory tracts of over 60 avian and 40 mammalian species (1). It is a significant veterinary, and occasional human, pathogen and the cause of swine atrophic rhinitis, a condition characterized by osteoclastic bone resorption and severe progressive turbinate damage. The pathogenicity of P. multocida is related to the production of a 146-kDa toxin (2, 3) that exhibits little functional homology to other known toxins or proteins. Both native and recombinant PMT 1 are potently mi-togenic for several cell types in vitro (4). In Rat1 fibroblasts, rPMT produces anchorage-independent DNA synthesis and growth in soft agar (5). In primary osteoblastic cells in culture, rPMT induces cell proliferation and down-regulation of markers of osteoblast differentiation (6).
The mechanisms whereby rPMT exerts its mitogenic effects are poorly understood. The toxin interacts with a gangliosidetype cell surface receptor and is internalized via both coated and noncoated endocytic structures (7,8). The mitogenic effects of rPMT require its internalization, since exposure of cells to rPMT at 4°C or incubation with the weak base methylamine prevents the response (4). Exposure to rPMT stimulates the hydrolysis of inositol phospholipids, calcium mobilization, and phosphorylation of protein kinase C (PKC) substrates, including the myristoylated alanine-rich C kinase substrate protein (4,5,9). In addition, rPMT has been shown to stimulate tyrosine phosphorylation of several proteins, including the focal adhesion kinase p125 FAK and paxillin, and to promote both focal adhesion assembly and the formation of actin stress fibers (8,10).
Several lines of evidence suggest that rPMT exerts at least some its effects by modulating the activity of G q/11 family heterotrimeric G proteins. Treatment of Swiss 3T3 cells with low doses of rPMT strongly potentiates inositol phosphate production following stimulation of G q/11 -coupled receptors, and PMT-induced phosphatidylinositol hydrolysis in permeabilized cells is blocked by guanosine 5Ј-O-(␤-thiodiphosphate) (11). In Xenopus oocytes, rPMT-induced calcium-dependent chloride currents are blocked by the injection of specific antisera against PLC␤1 or the ␣ subunit of G q/11 and by G␣ q antisense RNA, and are dramatically enhanced by overexpression of G␣ q (12).
Recently, many heterotrimeric G protein-coupled receptors (GPCRs) have been shown to activate the Erk mitogen-activated protein kinase pathway. The mechanisms whereby these signals are transduced are characterized by significant heterogeneity. Depending upon cell type, GPCRs have been shown to mediate both Ras-independent Erk activation via stimulation of PKC isoforms, and Ras-dependent Erk activation via activation of receptor and nonreceptor tyrosine protein kinases (13,14). Since rPMT is thought to mediate many of its effect via the regulation of heterotrimeric G proteins, understanding the mechanisms of PMT-induced mitogenesis might enhance our understanding of the roles of heterotrimeric G proteins in the regulation of cell growth. In this study, we have compared the mechanism of rPMT-mediated Erk activation to that employed by endogenous GPCRs in HEK-293 cells. We find that rPMT activates the Erk cascade via the stimulation of G q/11 family G proteins. Subsequently, both rPMT and endogenous G q/11 -coupled ␣-thrombin receptors induce Ras-dependent Erk activation via protein kinase C-independent "transactivation" of the EGF receptor. Thus, rPMT represents a novel bacterial strategy for inducing cell proliferation, by usurping heterotrimeric G protein-regulated mitogenic signals.
Cell Culture and Transient Transfection-HEK-293 cells were from the American Type Culture Collection. HEK-293 cells were maintained in minimum essential medium with Earle's salts (Life Technologies, Inc.) supplemented with 10% fetal bovine serum and 50 g/ml gentamicin. Transient transfections were performed using a modified calcium phosphate precipitation method as described previously (17). Monolayers of transfected cells were incubated in serum-free minimum essential medium supplemented with 10 mM HEPES, pH 7.4, 0.1% bovine serum albumin, and gentamicin for 16 -20 h prior to stimulation.
Inositol Phosphate Production-HEK-293 cells on six-well plates were labeled overnight with 2 Ci/ml myo- [2-3 H]inositol, washed once with phosphate-buffered saline (PBS), and placed in PBS supplemented with 10 mM LiCl and 1 mM CaCl 2 . Cells pretreated with rPMT were incubated for 45 min in Li ϩ /Ca 2ϩ PBS prior to lysis, while cells treated with agonist were preincubated for 15 min with Li ϩ /Ca 2ϩ followed by the addition of agonist for 30 min prior to lysis. Cells were lysed by the addition of 1.0 ml of 0.4 M perchloric acid, and a 0.8-ml aliquot of the resulting cell lysate was neutralized with 0.4 ml of 0.72 M KOH, 0.6 M KHCO 3 . Total inositol phosphates were determined by Dowex anion exchange chromatography as described previously (19).
Erk 1/2 Phosphorylation-Treatment with rPMT or receptor agonist drugs was performed at 37°C in serum-free medium following preincubation with inhibitors, as described in the figure legends. Erk 1/2 activation was assessed by measuring the phosphorylation state of endogenously expressed Erk 1/2 or of HA epitope-tagged Erk 1 immunoprecipitated from transiently transfected cells.
For determination of endogenous Erk 1/2 phosphorylation, cell monolayers on 12-well plates were washed once with ice-cold PBS and lysed in 200 l/well Laemmli sample buffer. Approximately 15 g of whole cell lysate protein/lane was resolved by SDS-polyacrylamide gel electrophoresis. Erk 1/2 phosphorylation was detected by protein immunoblotting using a 1:1000 dilution of rabbit polyclonal phosphospecific MAP kinase IgG (New England Biolabs) with horseradish peroxidaseconjugated goat anti-rabbit IgG (Amersham Pharmacia Biotech) as secondary antibody. Quantitation of Erk 1/2 phosphorylation was performed using scanning laser densitometry. After quantitation of Erk 1/2 phosphorylation, nitrocellulose membranes were stripped of immunoglobulin and reprobed using rabbit polyclonal anti-Erk1/2 IgG (Santa Cruz Biotechnology) to confirm equal loading of Erk2 protein.
In other experiments, transiently expressed HA epitope-tagged Erk 1 was immunoprecipitated from cells coexpressing putative dominant inhibitory mutant proteins. Following stimulation, monolayers on sixwell plates were washed once with PBS and lysed in 1 ml of ice-cold immunoprecipitation precipitation buffer (150 mM NaCl, 50 mM Tris-Cl, pH 8.0, 0.25% w/v sodium deoxycholate, 0.1% v/v Nonidet P-40, 1 mM NaF, 1 mM sodium pyrophosphate, 100 M NaVO 4 , 1 mM phenylmethylsulfonyl fluoride, 10 g/ml leupeptin, 10 g/ml aprotinin). Cell lysates were clarified by centrifugation and immunoprecipitation of HA-Erk1 was performed using 20 l of HA.11 affinity matrix (Berkeley Antibody Co.) with 2 h agitation at 4°C. Immune complexes were washed twice with ice-cold immunoprecipitation precipitation buffer and once with PBS, denatured in 2ϫ Laemmli sample buffer, and resolved by SDSpolyacrylamide gel electrophoresis. Quantitation of HA-Erk 1 phosphorylation was determined by protein immunoblotting using rabbit polyclonal phosphospecific MAP kinase IgG (New England Biolabs) as described. Fig. 1, treatment of HEK-293 cells with rPMT resulted in a dose-and time-dependent increase in Erk 1/2 phosphorylation. Erk 1/2 phosphorylation in rPMT-treated cells reached a maximum 6 -10-fold over basal increase after 24 -48 h of continuous exposure (Fig. 1A). Increasing concentrations of rPMT up to 100 ng/ml produced dose-dependent increases in Erk 1/2 phosphorylation ( Fig. 1B), which approximately paralleled rPMT-induced increases in inositol phosphate hydrolysis (Fig. 1C).

Recombinant PMT Mediates Erk 1/2 Activation via Pertussis Toxin-insensitive G q/11 Proteins-As shown in
Several lines of evidence suggest that the stimulation of inositol phosphate hydrolysis in response to rPMT results from toxin-induced activation of heterotrimeric G proteins (11,12). To examine whether heterotrimeric G proteins contributed to rPMT-induced activation of the Erk cascade, we initially determined whether rPMT-induced Erk 1/2 activation was additive with that induced by endogenous G i -coupled and G q/11 -coupled receptors. As shown in Fig. 2, stimulation of predominantly G q/11 -coupled ␣-thrombin receptors in HEK-293 cells increased Erk 1/2 phosphorylation 8 -10-fold. Pretreatment with rPMT for 24 h prior to stimulation failed to produce an additive response. In contrast, rPMT-mediated Erk 1/2 phosphorylation was additive with that induced by acute stimulation of the predominantly G i -coupled LPA receptor. Interestingly, the rPMT effect was also not additive with Erk 1/2 phosphorylation induced via the EGF receptor tyrosine kinase.
The lack of additivity between rPMT and ␣-thrombin receptor-mediated Erk 1/2 activation suggests that a common G protein pool might mediate the response to both stimuli. To determine the role of heterotrimeric G proteins in rPMT-mediated Erk 1/2 activation, we employed three selective inhibitors of G protein function: B. pertussis toxin, which ADP-ribosylates and inactivates G i/o family G proteins, and two inhibitory polypeptides GRK2(K220R) and G␣ q -(305-359). The catalytically inactive mutant of the G protein-coupled receptor kinase (GRK) 2, GRK2(K220R), which functions as an dominant negative antagonist of GPCR desensitization (20), has also been shown to directly block G q/11 -mediated inositol phosphate production in COS-1 cells (21). This effect results from the direct binding and sequestration of G␣ q/11 subunits by an RGS homology domain in the amino terminus of GRK2 (22). GRK2 also contains a carboxyl-terminal G␤␥ subunit binding motif, which inhibits G protein signaling by sequestering free G␤␥ subunits (23). The G␣ q -(305-359) peptide, derived from the carboxylterminal 55 amino acids of G␣ q , inhibits G q/11 -coupled, but not G i -coupled, receptor-mediated PI hydrolysis in COS-7 cells (24).
As shown in Fig. 3, pertussis toxin treatment had no effect on rPMT-stimulated Erk 1/2 phosphorylation in HEK-293 cells. Stimulation of Erk 1/2 by endogenous ␣-thrombin receptors was minimally affected by pertussis toxin, consistent with Erk activation mediated predominantly via G q/11 family G proteins. In contrast, LPA-stimulated Erk 1/2 phosphorylation was significantly reduced by pertussis toxin, consistent with a predominantly G i -mediated response. EGF-stimulated Erk 1/2 activation was pertussis toxin-insensitive.
In contrast to the lack of pertussis toxin effect, cellular expression of GRK2(K220R) and G␣ q -(305-359) each significantly attenuated rPMT-mediated Erk 1/2 activation. To optimize the detection of signals originating from the transfected cell pool, HEK-293 cells in these experiments were transiently cotransfected with HA epitope-tagged Erk 1 (HA-Erk 1) along with the putative inhibitory peptide, and the phosphorylation state of immunoprecipitated HA-Erk 1 determined. As shown in Fig. 4A, expression of GRK2(K220R) significantly attenuated HA-Erk 1 phosphorylation in response to rPMT treatment, and to stimulation of ␣-thrombin and LPA receptors, with no effect on EGF-stimulated HA-Erk 1 activation. Phosphorylation of HA-Erk 1 resulting from cellular expression of a constitutively activated mutant of G␣ q (G␣ q (Q209L)) was also attenuated by GRK2(K220R) expression, consistent with direct inhibition of G␣ q -dependent signals by GRK2(K220R). Thus, inhibition by GRK2(K220R) is indicative either of a role for G q/11 proteins or for G␤␥ subunits in rPMT-mediated Erk activation.
To discriminate between these alternatives, we employed the G␣ q -(305-359) peptide. Cellular expression of this peptide has been shown to inhibit G q/11 -dependent, but not G i -or Gs-dependent signaling in COS-7 cells (24). As shown in Fig. 4B, expression of G␣ q -(305-359) significantly attenuated inositol phosphate production induced either by acute stimulation of ␣-thrombin receptor or by expression of G␣ q (Q209L), consistent with direct inhibition of G␣ q -dependent signals by G␣ q -(305-359). As shown in Fig. 4C, G␣ q -(305-359) expression markedly inhibited HA-Erk 1 phosphorylation in response to rPMT treatment, G␣ q (Q209L) expression, and stimulation of ␣-thrombin receptors. LPA-stimulated HA-Erk 1 phosphorylation was less sensitive to G␣ q -(305-359) expression, consistent with the greater contribution of pertussis toxin-sensitive G proteins to LPA-stimulated Erk activation in these cells (Fig. 3), and EGFstimulated HA-Erk phosphorylation was insensitive to G␣ q -(305-359) expression.
Collectively, these data suggest that rPMT mediates Erk activation in HEK-293 cells via G q/11 family G proteins. The sensitivity of sustained rPMT-mediated HA-Erk 1 phosphorylation to GRK2(K220R) and G␣ q -(305-359) expression parallels that of the constitutively active G␣ q (Q209L) mutant. In addition, the lack of additivity observed between rPMT-mediated Erk activation and that induced by the predominantly G q/11coupled ␣-thrombin receptor suggests that the two signals converge upon a common intermediate.

Recombinant PMT Induces Erk Activation via G Protein-dependent Transactivation of the EGF Receptor Tyrosine Kinase-
The mechanisms whereby GPCRs activate the Erk cascade are characterized by extensive heterogeneity (13,14). Depending upon cell type, receptors coupled to G q/11 proteins have been shown to mediate both Ras-independent Erk activation via stimulation of PKC isoforms, and Ras-dependent Erk activation via activation of receptor and nonreceptor tyrosine protein kinases. In the latter case, GPCR-induced activation of receptor tyrosine kinases, such as the EGF receptor (25,26), or of focal adhesion kinases (FAK), such as the calcium-dependent FAK kinase Pyk2 (27,28), serves to initiate a tyrosine phosphorylation cascade leading to Ras activation.
The absence of an additive effect of rPMT treatment on EGF-stimulated Erk 1/2 phosphorylation is consistent with a role for EGF receptors in mediating the rPMT response. In addition, rPMT is known to stimulate both inositol phosphate hydrolysis and to induce calcium-and PKC-independent tyrosine phosphorylation of several proteins, including p125 FAK and paxillin (10). Thus, it is plausible that either mechanism could account for rPMT-stimulated Erk activation. To discriminate between these mechanisms, we assessed the sensitivity of rPMT-mediated Erk 1/2 activation to specific PKC and tyrosine kinase inhibitors. As shown in Fig. 5A, inhibition of classical PKC isoforms with GF109203X blocked acute phorbol esterstimulated Erk phosphorylation, but had no effect on the rPMT, ␣-thrombin, and EGF receptor-mediated signals. Similar results were obtained using Ro31-8220, a chemically distinct PKC inhibitor with a similar spectrum (data not shown). As shown in Fig. 5B, inhibition of EGF receptor activity using the EGF receptor-specific tyrphostin AG1478 profoundly reduced rPMT, ␣-thrombin or EGF receptor-mediated Erk phosphorylation, with no effect on the response to phorbol ester. In contrast, treatment of cells with the platelet-derived growth factor receptor-specific tyrphostin AG1295 had no inhibitory effect (data not shown). Thus, rPMT-mediated Erk activation requires EGF receptor activity, but not the activity of classical PKC isoforms. Since rPMT-stimulated Erk activation, like rPMT-stimulatedinositolphosphatehydrolysis(11)andcalciumdependent chloride currents (12), is sensitive to inhibitors of G q/11 proteins, these data suggest that G q/11 -mediated EGF receptor transactivation accounts for the effects of rPMT on Erk 1/2 activity.
UnlikedirectPKC-mediatedErkactivation(29),EGFreceptordependent Erk activation is sensitive to dominant inhibitory mutants of mSos and Ha-Ras (30). As shown in Fig. 6, expression of the two dominant negatives, Sos-PRO and N17Ras, significantly inhibited rPMT, ␣-thrombin, or EGF receptormediated Erk phosphorylation, with no effect on the phorbol ester response. Thus rPMT-mediated Erk activation involves G q/11 -dependent EGF receptor transactivation and leads to the Ras-dependent activation of the Erk cascade. As depicted schematically in Fig. 7, these events are dissociated from rPMTmediated effects on PKC, which contribute at most a minor component to rPMT-stimulated Erk activation. DISCUSSION Despite growing evidence that many of the cellular effects of rPMT are mediated by the activation of G q/11 family heterotrimeric G proteins, the detailed molecular mechanism of rPMT action remains poorly understood. Like other bacterial toxins with intracellular loci of action, binding to cell surface receptors and processing through endocytic vesicles appears to be the rate-limiting step for rPMT action on intact cells. A 3-4-h delay is observed between application of rPMT to intact cells and its maximal effects on inositol phospholipid hydrolysis (9), yet phospholipase C␤-mediated inositol 1,4,5-trisphosphate and calcium responses occur within seconds when the toxin is directly injected into Xenopus oocytes (12).
Unlike several bacterial toxins that modulate G protein activity, rPMT toxin lacks recognizable enzymatic activity. Microinjection of antisera directed against the amino terminus, but not the carboxyl terminus, of rPMT inhibits its biological activity in Xenopus oocytes (12). Although the full-length protein is required for biological activity on intact cells, the aminoterminal 500 residues of the toxin are active when microinjected (15). This region of PMT shares modest homology with the amino-terminal half of the cytotoxic necrotizing factors 1 and 2 from enteropathogenic E. coli (31,32), which are thought to target Rho family small GTP-binding proteins (32). In the Xenopus system, the effect of rPMT on calcium-dependent chloride channel activation is markedly enhanced by G␣ q overexpression and specifically inhibited by microinjection of antisera against G␣ q (12). Oocytes injected with rPMT fail to respond to a second challenge, suggesting that rPMT may bind directly to G␣ q subunits and induce GTP exchange. Our results, demonstrating that cellular expression of peptide inhibitors of G q/11 signaling block rPMT action in intact cells, are consistent with this model.
The activation of G q/11 proteins by rPMT may be sufficient to account for its mitogenic effects, as well as its effects on inositol phospholipid turnover and PKC activation. Transient expression of a constitutively active mutant of G␣ q mimics the effects of rPMT, producing sustained increases in phosphatidylinositol hydrolysis and Erk activation that are blocked by peptide inhibitors of G q/11 signaling. The mechanism of rPMT-stimulated Erk activation in HEK-293 cells parallels that employed by the predominantly G q/11 -coupled ␣-thrombin receptor. Both signals are independent of PKC, but dependent upon the catalytic activity of endogenous EGF receptors to activate Ras-dependent pathways.
Both rPMT exposure and acute stimulation of G q/11 -coupled receptors induce the tyrosine phosphorylation of multiple substrates. In Swiss 3T3 cells, rPMT stimulates tyrosine phosphorylation of p125 FAK and paxillin in a cytochalasin D-sensitive manner (10). Similar results have been obtained following acute stimulation of G q/11 -coupled receptors in Swiss 3T3 (33), Rat 1 (34), and HEK-293 cells (35). Our data indicate that rPMT also mediates growth stimulatory effects via G proteindependent activation of EGF receptors. Indeed, previous work has demonstrated that rPMT exposure promotes the loss of cell surface EGF receptors (36), an effect that may represent EGF receptor down-regulation following rPMT-induced transactivation. G q/11 -coupled receptors have been shown to mediate two, apparently distinct, calcium-dependent tyrosine phosphorylation cascades that converge on Ras and its downstream effectors. In several cell types, GPCR-stimulated Erk activation involves the ligand-independent transactivation of receptor tyrosine kinases, such as the EGF receptor (25,26). In PC12 cells, bradykinin (37) and m1 muscarinic acetylcholine (38) receptors activate EGF receptors via either calcium-or PKC-dependent mechanisms, respectively. In vascular smooth muscle, angiotensin II mediates calcium-dependent EGF receptor transactivation upstream of the Erk cascade (39). Alternatively, bradykinin-dependent Erk activation has been attributed to stimulation of the calcium-dependent FAK family kinase, Pyk2 (27,28). Pyk2-dependent signals require either calcium or PKC (27) and are sensitive to cytochalasin D, suggesting a requirement for intact focal adhesions (40). Activated Pyk2, like p125FAK, associates with c-Src and provides docking sites for Grb2-Sos complexes (28).
Despite similarities in their mechanisms of activation (41), Pyk2-mediated signals appear to be independent of those mediated via EGF receptor transactivation. In PC12 cells, expression of a dominant negative mutant of Pyk2 does not inhibit bradykinin-induced transactivation of the EGF receptor, and expression of a dominant negative EGF receptor mutant has no effect on Pyk2 phosphorylation (42). In HEK-293 cells, treatment with tyrphostin AG1478 (35) and expression of a dominant negative mutant of Pyk2 (43) each cause a partial blockade of G q/11 -coupled receptor-mediated Erk activation, suggesting that both transactivated EGF receptors and Pyk2 contribute independently to Erk activation. While we find that rPMT-stimulated Erk activation is predominantly tyrphostinsensitive, we cannot exclude a contribution from Pyk2 acting via a focal adhesion-based scaffold.
Study of the mechanisms of action of bacterial toxins has led to significant insights into the roles of heterotrimeric G proteins in cellular signaling. Our data suggest that rPMT employs G q/11 proteins in a novel strategy for mitogenic regulation by a bacterial toxin, that of utilizing G proteins to initiate a tyrosine phosphorylation cascade leading to Ras activation. Given the potent mitogenic effects achieved by rPMT, these findings serve to underscore the importance of G protein-catalyzed mitogenic signals to growth regulation. FIG. 7. Model for the proposed mechanism of Erk 1/2 regulation by rPMT. Activation of G q/11 proteins by either rPMT or G qcoupled GPCRs such as the thrombin receptor results in activation of G q/11 effectors such as phospholipase C␤ (PLC␤) isoforms as well as transactivation of the EGF receptor tyrosine kinase. Subsequent activation of the Erk cascade is mediated predominantly via RTK transactivation, with no significant contribution derived from the PLC␤-dependent activation of classical PKC activation protein kinase.