Action of Pasteurella multocida toxin depends on the helical domain of Galphaq.

Pasteurella multocida produces a 146-kDa protein toxin (PMT), which activates multiple cellular signal transduction pathways, resulting in the activation of phospholipase Cbeta, RhoA, Jun kinase, and extracellular signal-regulated kinase. Using Galpha(q)/Galpha(11) -deficient cells, it was shown that the PMT-induced pleiotropic effects are mediated by Galpha(q) but not by the highly related Galpha(11) protein (Zywietz, A., Gohla, A., Schmelz, M., Schultz, G., and Offermanns, S. (2001) J. Biol. Chem. 276, 3840-3845). Here we studied the molecular basis of the unique specificity of PMT to distinguish between Galpha(q) and/or Galpha(11). Infection of Galpha(q) -deficient cells with retrovirus-encoding Galpha(q) caused reconstitution of PMT-induced activation of phospholipase Cbeta, whereas Galpha(11) -encoding virus did not reconstitute PMT activity. Chimeras between Galpha(q) and/or Galpha(11) revealed that a peptide region of Galpha(q), covering amino acid residues 105-113, is essential for the action of PMT to activate phospholipase Cbeta. Exchange of glutamine 105 or asparagine 109 of Galpha(11), which are located in the all-helical domain of the Galpha subunit, with the equally positioned histidines of Galpha(q), renders Galpha(11) capable of transmission PMT-induced phospholipase Cbeta activation. The data indicate that the all-helical domain of Galpha(q) is essential for the action of PMT and suggest an essential functional role of this domain in signal transduction via G(q) proteins.

The Pasteurella multocida is a facultative pathogen, which cause bite wound infections, pneumonia, endocarditis, and septicemia in men. In pigs, the pathogen induces atrophic rhinitis, which is characterized by a loss of nasal turbinate bone (1,2). The 146-kDa protein toxin P. multocida toxin (PMT) 1 is the major virulence factor of the pathogen, the causative agent of atrophic rhinitis, and is responsible for the osteolytic activity of bacteria (1,(3)(4)(5). PMT consists of 1285 amino acid residues. It is generally accepted that the toxin is structured according to a typical AB toxin. Initial studies revealed that the N terminus of PMT is involved in the binding and in translocation of the toxin into target cells (6), whereas the biologically active domain is located in the C-terminal part of the protein (6,7). This concept is in line with a significant sequence similarity of PMT at its N terminus with the N-terminal part of the cytotoxic necrotizing factor of Escherichia coli that is also involved in binding and translocation. According to the hypothesis that the C terminus of PMT carries the biological activity, an essential cysteine residue (Cys 1165 ) was identified at the C terminus (8). The change of Cys 1165 to serine blocked toxin activity but not cell binding. In addition, histidine residues (His 1205 and His 1223 ) in this part of the toxin were recognized to be essential for the activity of PMT (9).
PMT activates numerous cellular signal transduction pathways. It is a strong mitogen and stimulates DNA synthesis and proliferation in several cell lines (10 -14). The mitogenic actions of PMT appear to depend on the stimulation of the extracellular signal-regulated kinase (ERK) (15). PMT stimulates phospholipase C␤1 (PLC␤1) in a G q -dependent manner (16), resulting in calcium mobilization, accumulation of diacylglycerol, and activation of protein kinase C (11). In addition, PMT activates the small GTPase RhoA, thereby inducing formation of stress fibers, focal adhesions, and tyrosine phosphorylation of focal adhesion kinase and paxillin (12,17). The activation of Rho, extracellular signal-regulated kinase, and Jun kinase appears to be independent of G q , indicating that PMT activates signaling pathways in a G q -dependent and -independent manner (18). So far, however, the precise mode of molecular action of PMT is not known.
Recently, it was shown by gene deletion of the ␣-subunits of G q and G 11 that PMT acts on PLC␤ via G␣ q but not via G␣ 11 (18). This is remarkable because G␣ q and G␣ 11 are highly related and share 89% of their amino acid residues.
We studied the molecular basis for this unique difference between G␣ q and G␣ 11 . Using G protein chimeras, we identified the helical domain of G␣ q to be essential for mediating the activation of PLC␤ by PMT. We obtained evidence that the region of helix ␣B of G␣ q is involved in the PMT effect. Moreover, exchange of residues Gln 105 and Asn 109 in G␣ 11 to that of G␣ q renders G␣ 11 sensitive toward PMT effects. These findings show that not only the Ras-like GTPase domain of G␣ q but also the helical domain, which is inserted into the GTP-binding domain before switch 1, are functionally important and essential for mediating the activating effects of PMT. These findings give important new information on the site of action of PMT and suggest that the all-helical domain is important for signal transduction processes from G␣ q to PLC␤.
Constructions of Mutant G␣ q and G␣ 11 -Mutated proteins were constructed by site-directed mutagenesis using pLNCX2 constructs as template and the respective oligonucleotides using the QuikChange kit according to the manufacturer's instructions. The nucleotide changes responsible for mutation are underlined. From the two complementary primers used for each mutation, only one is listed: G␣ q , 5Ј-GAG CAC AAC AAG GCC CAT GCA CAA CTG GTT CGG GAG-3Ј. The plasmids were transformed in E. coli XL1 cells. All mutations were confirmed by DNA sequencing with the ABI PRISM TM dye terminator cycle sequencing ready reaction kit (PerkinElmer Life Sciences, Ü berlingen, Germany).
Analysis of Total Inositol Phosphate-Infected G␣ q /G␣ 11 -deficient cells were grown in 24-well plates for 3 days and labeled with 2 Ci/ml [2-3 H]inositol in inositol-and serum-free medium (minimal essential medium) for 12 h. Subsequently, PMT and LiCl (20 mM) were added, and the cells were incubated for the indicated times. For cell lysis and extraction of inositol phosphate, the medium was replaced with 750 l of ice-cold formic acid (10 mM, pH 3). After 30-min incubation on ice, the extract was neutralized with 3 ml of NH 3 (5 mM, pH 8.5). Analysis of total inositol phosphate was done by anion exchange chromatography using AG1-X8 resin (200 -400 mesh; Bio-Rad, Mü nchen, Germany) as described previously (7).
Expression and Purification of PMT Protein-Recombinant PMT protein was expressed as glutathione S-transferase fusion protein and purified according to the manufacturer's instructions (Amersham Biosciences). In brief, glutathione S-transferase fusion protein was isolated by affinity chromatography with glutathione-Sepharose, followed by proteolytic cleavage using 3.25 units of thrombin/mg of recombinant glutathione S-transferase fusion protein. Thrombin was removed by incubation with benzamidine-Sepharose (Amersham Biosciences).

FIG. 1. Effect of PMT on inositol phosphate production in wildtype mouse embryonic fibroblasts and in G␣ q /G␣ 11 -deficient cells.
A, wild-type cells (triangles) and G␣ q /G␣ 11 -deficient cells (squares) were incubated for 6 h at the indicated concentrations of PMT. The total amount of inositol phosphate was measured as described under "Experimental Procedures." Data are given as fold induction over buffer control. B, expression of G␣ q /G␣ 11 in wild-type mouse embryonic fibroblasts and in G␣ q /G␣ 11 -deficient cells. Shown is an immunoblot of RIPA extracts of wild-type cells and G␣ q /G␣ 11 -deficient cells. The immunoblot was performed as described under "Experimental Procedures" with a rabbit polyclonal antibody against a common C-terminal sequence of G␣ q and G␣ 11 .

FIG. 2. Retroviral infection of G␣ q /G␣ 11 -deficient cells with G␣ q -encoding virus recovers PMT-induced inositol phosphate
production. G␣ q -or G␣ 11 -encoding retrovirus was produced, and G␣ q / G␣ 11 -deficient cells were transduced with the resulting retrovirus as described under "Experimental Procedures." A, G␣ q /G␣ 11 -deficient cells, transduced with G␣ q (filled rhombus)-or G␣ 11 (open rhombus)encoding retrovirus were incubated for 6 h at the indicated concentrations of PMT. The total amount of inositol phosphate was measured as described under "Experimental Procedures." Data are given as fold stimulation over buffer control. B, expression of G␣ q or G␣ 11 in retroviral transduced G␣ q /G␣ 11 -deficient cells was examined by Western blot. Shown is an immunoblot of RIPA extracts of G␣ q /G␣ 11 -deficient cells. The immunoblot was performed as described under "Experimental Procedures." polyacrylamide gel electrophoresis and visualized by chemiluminescence detection using goat anti-rabbit antibodies (Biotrend Chemikalien, Cologne, Germany) coupled to horseradish peroxidase and were visualized using ECL reagent. Rabbit polyclonal antibody against G␣ q/11 (C-19) was purchased from Santa Cruz Biotechnology (Heidelberg, Germany). Fig. 1 shows the effects of PMT on wild-type and G␣ q /G␣ 11deficient cells. Whereas PMT causes a strong activation of inositol phosphate accumulation in wild-type cells, the toxin had no effect in G␣ q /G␣ 11 -deficient cells. The Western blot analysis revealed that wild-type cells expressed G␣ q and G␣ 11 , whereas in the G␣ q and G␣ 11 gene-deficient cells, no expression of the ␣-subunits of G q and G 11 were detected. Next, we studied whether the PMT response could be reconstituted in G␣ q /G␣ 11deficient cells by introduction and expression of the cDNA of the G proteins into the cells. For this purpose, we used a retroviral system, harboring the cDNA of G␣ q or G␣ 11 . As shown in Fig. 2, infection with G␣ q but not with G␣ 11 caused reconstitution of the PMT effect. This confirms recent results showing that G␣ q but not G␣ 11 is able to mediate PMT-induced activation of PLC␤.

RESULTS
To investigate the molecular mechanism responsible for this unique specificity of PMT, we decided to construct G␣ q /G␣ 11 chimeras. G␣ q and G␣ 11 are about 89% identical in their amino acid sequences. At first, we constructed a chimera consisting of the N-terminal part of residues 1-120 of G␣ q and 121-359 of G␣ 11 (G␣ q -G␣ 11 121-359 ). Conversely, we constructed the chimera G␣ 11 -G␣ q 121-359 (Fig. 3). These constructs were introduced into the retrovirus, and the G␣ q /G␣ 11 -deficient cells were infected with the retrovirus. As shown in Fig. 4, only the construct with the N-terminal part of G q e.g. G␣ q -G␣ 11 121-359 caused activation of PLC␤ after PMT addition to infected cells.
Next, we further confined the region responsible for PMT sensitivity of G␣ q and constructed the chimera G␣ 11 -G␣ q 67-120 (Fig. 3). As a control, we constructed the inverse chimera G␣ q -G␣ 11 67-120 . Western blot analysis showed that these proteins were expressed in infected cells (Fig. 5). When cells, which expressed these chimeric G proteins, were treated with PMT, an increase in inositol phosphate accumulation was observed with G␣ 11 -G␣ q 67-120 but not with the other constructs. This finding indicated that the structural requirements for mediation of PMT activity are located within a peptide of G␣ q that covers residues 67-120. This region harbors the helix ␣A and ␣B of the all-helical domain, and only a few residues are different between G␣ q and G␣ 11 (Fig. 6A). To identify the region essential for PMT-mediated PLC␤ activation, we exchanged helix ␣B of G␣ q and G␣ 11 by sequential site-directed mutagenesis. Thereby, we generated chimeric G␣ q protein, harboring helix ␣B from G␣ 11 (G␣ q -G␣ 11 105-113 ), and G␣ 11 protein, harboring helix ␣B from G␣ q (G␣ 11 -G␣ q 105-113 ) (Fig. 3). Both constructs were tested for the ability to mediate PMT-induced activation of PLC␤. As Fig. 6B shows PMT-stimulated PLC␤ only via G␣ 11 -G␣ q 105-113 . Next, we exchanged glutamine at position 105 and aspara-

FIG. 3. Schematic representation of G␣ q /G␣ 11 chimeras.
To determine the region in G␣ q , which is essential for the PMT-induced stimulation of the G␣ q -PLC␤ pathway, several chimeric constructs of G␣ q and G␣ 11 were generated as described under "Experimental Procedures." G␣ q -G␣ 11 121-359 contains the N terminus of G␣ q , including helices ␣A and ␣B of the helical domain. The C-terminal part of this construct (after position 120) consists of G␣ 11 . G␣ 11 -G␣ q 67-120 : helices ␣A and ␣B of the helical domain of G␣ q were introduced in G␣ 11 . G␣ 11 -G␣ q 105-113 : only helix ␣B of the helical domain of G␣ q was inserted into G␣ 11 . For each described chimeric construct a converse construct was generated (G␣ 11  (open rhombus)-encoding retrovirus were incubated for 6 h at the indicated concentrations of PMT. The total amount of inositol phosphate was measured as described under "Experimental Procedures." Data are given as fold stimulation over buffer control. B, expression of G␣ q -G␣ 11 121-359 or G␣ 11 -G␣ q 121-359 in retroviral transduced G␣ q /G␣ 11 -deficient cells was examined by Western blot. Shown is an immunoblot of RIPA extracts of G␣ q /G␣ 11 -deficient cells. The immunoblot was performed as described under "Experimental Procedures." gine at position 109 in G␣ 11 with the equally positioned histidines in G␣ q . The mutants G␣ 11 Q105H and G␣ 11 N109H were introduced into the retrovirus and the G␣ q /G␣ 11 -deficient cells infected with the virus. As shown in Fig. 7, PMT treatment of G␣ q /G␣ 11 -deficient cells, which expressed G␣ 11 Q105H or G␣ 11 N109H , resulted in stimulation of inositol phosphate production and reconstituted the stimulation of PLC␤. Surprisingly, a converse mutant, e.g. G␣ q H105Q/H109N , did not lose the ability to activate PLC␤ by PMT treatment. Taken together, these results indicate that the minimal requirement to recover PMT-mediated response via G␣ 11 is the introduction of a histidine residue at position 105 or 109. However, to render G␣ q inactive, the additional exchange of helix ␣B to that of G␣ 11 is required. DISCUSSION PMT has pleiotropic effects on target cells, including activation of the mitogen-activated protein kinase pathway, RhoA activation, and activation of PLC␤. Although the molecular mechanism of PMT is not known, evidence has been obtained that G␣ q activation is involved in the stimulation of PLC␤ by PMT. Recently, it was shown that activation of PLC␤ by PMT is mediated by G␣ q but not by G␣ 11 . This finding was surprising because both G proteins are closely related, and G␣ q -coupled receptors were constantly found to activate also G␣ 11 . Because both G proteins share a sequence identity of 89%, we were prompted to identify the region that is responsible for this unique specificity of PMT. For this purpose, we used G␣ q /G␣ 11deficient cells, which are unresponsive toward PMT in regard to PLC␤ stimulation. Introduction of G␣ q but not of G␣ 11 into these cells reconstituted the sensitivity of these cells for PLC␤ activation by PMT. We constructed several chimeras and checked the expression of the chimeras by Western blotting. These studies showed that a small region, covering residues 105-113, defined the structural requirement of G␣ q as compared with G␣ 11 to mediate the PLC␤ stimulatory effects of PMT. Because G␣ q and G␣ 11 are very similar in this region, we further equalized both sequences by changing amino acid residues in G␣ 11 to that of G␣ q .
Exchange of Gln 105 or Asn 109 in G␣ 11 to histidine rendered G␣ 11 capable of mediating PMT-induced activation of PLC␤. These findings are surprising because residues 105 and 109 of G␣ q and of G␣ 11 are positioned in the helical domains of the G proteins. So far almost all functions of G␣ q are related to the Ras-like GTPase domain of the G proteins. The functional role of the helical domain is still largely enigmatic. However, our findings indicate that this region, especially helix ␣B, is involved in the PMT effects. Moreover, the finding that exchange of one amino residue in the region largely effects the functional properties of the G␣ q/11 proteins suggested that minor structural changes have major consequences in the function of G proteins.
During recent years some findings suggested that the allhelical domain of other G proteins might have a role in signal transduction. For example, it was shown by crystal structure analysis that the G o LOCO motif (also called G protein regulatory motif), which is found in some GTPase-activating proteins of the RGS (regulator of G protein signaling) family (e.g. RGS12 and RGS14), binds to the all-helical domain of G proteins (22,23). Also AGS proteins (activators of G protein signaling), which function as guanine nucleotide dissociation inhibitors, possess a G o LOCO domain (24). AGS3 protein binds to the GDP-bound forms of all G i isoforms, to G␣ t , and weakly to G␣ q . Recently, Natochin et al. (25) identified residues 144 -151 of G␣ i , which are located in the helical domain, to be essential for interaction with AGS3. However, it appears that AGS3 has activity as a guanine nucleotide dissociation inhibitor only toward G␣ i and G␣ t . Thus, so far, no evidence has been presented that the helical domain of G␣ q is somehow involved in signaling processes.
Comparing the activation of heterotrimeric G proteins with the mode of activation of small GTPases and GTP-binding elongation factors, recently, Cherfils and Chabre (26) suggested a novel model of G protein activation. The main point of the model is an increase in the interaction and contact area of G␤/␥ and G␣ to stabilize a nucleotide-free complex. In contrast to the dissociation model of G protein activation, Cherfils and Chabre (26) proposed that upon activation the N terminus of the G␥-subunit interacts in a hook-like manner with the helical domain, especially with the ␣A or ␣B helices of G␣. Moreover, by fluorescence resonance energy transfer-based studies, it has been proposed recently that subunit rearrangement rather than dissociation is involved in G i protein activation (27). It is fascinating to speculate that the same region, e.g. ␣A or ␣B helix of G␣, which is important for G protein activation, is involved in PMT-induced activation of G q .
The molecular mode of action of PMT is still not known. The high potency of the toxin suggests that an enzyme activity is involved as reported for many other bacterial protein toxins and bacterial effectors targeting eukaryotic cells (28). Several bacterial effectors target G proteins, including pertussis toxin and cholera toxin, which ADP-ribosylate G i , and G s , respec- -encoding retrovirus was produced, and G␣ q /G␣ 11 -deficient cells were transduced with the resulting retrovirus as described under "Experimental Procedures." A, G␣ q /G␣ 11deficient cells, transduced with G␣ 11 -G␣ q 67-120 (filled squares)-or G␣ q -G␣ 11 67-120 (open squares)-encoding retrovirus were incubated for 6 h at the indicated concentrations of PMT. The total amount of inositol phosphate was measured as described under "Experimental Procedures." Data are given as fold induction over buffer control. B, expression of G␣ 11 -G␣ q 67-120 or G␣ q -G␣ 11 67-120 in retroviral transduced G␣ q /G␣ 11 -deficient cells was examined by Western blot. Shown is an immunoblot of RIPA extracts of G␣ q /G␣ 11 -deficient cells. The immunoblot was performed as described under "Experimental Procedures." FIG. 6. Introduction of helix ␣B of G␣ q into G␣ 11 is sufficient to mediate PMT-induced activation of the resulting G␣ q /G␣ 11 chimera. A, an alignment of G␣ q and G␣ 11 harboring the Nterminal ␣A and ␣B helices of the helical domain is shown. The regions with predicted helical structures are indicated, and amino acids, which are different in G␣ q and G␣ 11 , are highlighted. In the region of helix ␣B four amino acids differ. These amino acids were sequentially exchanged in G␣ q and G␣ 11 by site-directed mutagenesis as described under "Experimental Procedures." B, retrovirus encoding for G␣ 11 -G␣ q 105-113 (filled triangles) and G␣ q -G␣ 11 105-113 (open triangles) were generated and used to transduce G␣ q / G␣ 11 -deficient cells. The ability to induce inositol phosphate formation by treatment with PMT was determined as before. C, expression of G␣ 11 -G␣ q 105-113 or G␣ q -G␣ 11 105-113 in retroviral transduced G␣ q /G␣ 11 -deficient cells was examined by Western blot. Shown is an immunoblot of RIPA extracts of G␣ q /G␣ 11 -deficient cells. The immunoblot was performed as described under "Experimental Procedures. "   FIG. 7. Exchange of single amino acid residues in helix ␣B of G␣ 11 to corresponding amino acid residues of G␣ q enables the resulting G␣ 11 construct to mediate PMT signaling. A, retroviral transfer vector with inserts for G␣ q or G␣ 11 was mutated by site-directed mutagenesis to generate the following constructs: G␣ 11 Q105H (rhombus), G␣ 11 N109H (squares) and G␣ q H105Q/H109N (triangles). G␣ q /G␣ 11 -deficient cells, transduced with retrovirus encoding for described constructs, were used to test the ability to activate PLC␤ by PMT treatment. B, expression of G␣ 11 Q105H , G␣ 11 N109H , and G␣ q H105Q/H109N in retroviral transduced G␣ q /G␣ 11 -deficient cells was examined by Western blot. Shown is an immunoblot of RIPA extracts of G␣ q /G␣ 11 -deficient cells. The immunoblot was performed as described under "Experimental Procedures." tively (29,30). Low molecular mass GTPases, especially of the Rho/Ras family, are modified by ADP-ribosylation, glucosylation, (e.g. Clostridium difficile toxins), and by deamidation (e.g. cytotoxic necrotizing factors from E. coli) (28,31). Moreover, several bacterial effectors activate or inactivate small GTPases by mimicking the activity of eukaryotic guanine nucleotide exchange factors (e.g. SopE from Salmonella enterica) and GT-Pase-activating proteins (e.g. SptP from S. enterica) (28,32,33). So far no evidence is available that PMT belongs to one of these toxin families. Recently, it was shown that PMT induces tyrosine phosphorylation of G␣ q (34). Tyrosine phosphorylation of G q has been claimed to regulate the activity of the G protein (35,36). However, PMT-induced tyrosine phosphorylation of G q , which occurs at the very C terminus, is also observed with the cysteine mutant that is defective in activation of PLC␤ (34).
Here, we report that exchange of only one amino acid residue in G␣ 11 (e.g. Gln 105 or Asn 109 ) allows PMT-induced activation of PLC␤. Surprisingly, the reverse is not true; exchange of the equivalent residues in G␣ q (e.g. His 105 or His 109 ) did not block its ability to mediate PMT-induced activation of PLC␤. One explanation of this finding is that the interaction of PMT (or that of a putative PMT-effector) with G␣ q and the subsequent activation of the G protein is supported by several amino acid residues, some of which are located in the ␣B helix. Exchange of one of these residues did not prevent the PMT effect. However, only one of these crucial residues allows mediation of PMT activation of PLC␤ in G␣ 11 .
Taken together, our data indicate that the ␣-helical region, especially the ␣B helix, is essential for the functional difference of G␣ q and G␣ 11 to serve as a target of PMT. These data support the view that the helical domain of G␣ q has important functions in G protein activation and signaling.