Modification of Heterotrimeric G-Proteins in Swiss 3T3 Cells Stimulated with Pasteurella multocida Toxin

Many bacterial toxins covalently modify components of eukaryotic signalling pathways in a highly specific manner, and can be used as powerful tools to decipher the function of their molecular target(s). The Pasteurella multocida toxin (PMT) mediates its cellular effects through the activation of members of three of the four heterotrimeric G-protein families, Gq, G12 and Gi. PMT has been shown by others to lead to the deamidation of recombinant Gαi at Gln-205 to inhibit its intrinsic GTPase activity. We have investigated modification of native Gα subunits mediated by PMT in Swiss 3T3 cells using 2-D gel electrophoresis and antibody detection. An acidic change in the isoelectric point was observed for the Gα subunit of the Gq and Gi families following PMT treatment of Swiss 3T3 cells, which is consistent with the deamidation of these Gα subunits. Surprisingly, PMT also induced a similar modification of Gα11, a member of the Gq family of G-proteins that is not activated by PMT. Furthermore, an alkaline change in the isoelectric point of Gα13 was observed following PMT treatment of cells, suggesting differential modification of this Gα subunit by PMT. Gs was not affected by PMT treatment. Prolonged treatment with PMT led to a reduction in membrane-associated Gαi, but not Gαq. We also show that PMT inhibits the GTPase activity of Gq.


Introduction
Heterotrimeric G-proteins are a family of key signal transduction proteins that intercede between the many G-protein coupled receptors (GPCR) that the cell uses to interrogate its local environment and downstream signalling pathways that ultimately regulate fundamental cellular choices [1]. G-proteins are divided into 4 classes (G q , G 12 , G i and G s ) according to their constituent alpha subunit, which is a guanine nucleotide binding protein that can exist in an inactive GDP-bound or an active GTP-bound form [2]. Activation of a GPCR causes a conformational change in its cognate Ga subunit that triggers GDP to be exchanged for GTP. The activated state persists until GTP is hydrolysed to GDP by the intrinsic GTPase activity of the Ga subunit. G-proteins are also subject to reversible tyrosine phosphorylation and lipid modifications during their activation cycle, but the regulatory role of these events is not fully understood [3]. Each G-protein class activates a characteristic set of downstream targets. The G s and G i families activate or inhibit adenylate cyclase, respectively [4]. The G q family activates phospholipase C (PLC) [5], while the G 12 family is particularly linked to activation of the Rho GTPase [6].
Intracellularly-acting bacterial protein toxins enzymatically modify a limited and precise set of cellular proteins to modulate their function. The Pasteurella multocida toxin (PMT) activates multiple signalling pathways in cultured cells leading characteristically to a strong mitogenic response [7]. PMT has been shown to activate members of the G q , G 12 and G i families [8][9][10][11][12][13]. PMT catalyses the deamidation of recombinant G i at Gln-205 to inhibit its intrinsic GTPase activity [14]. We describe here the effects of PMT on all four classes of heterotrimeric G-proteins in Swiss 3T3 cells using two-dimensional (2-D) gel electrophoresis and other techniques.
USA) [16]. All other chemical reagents were of analytical grade and were obtained from Sigma-Aldrich, unless otherwise stated.

Cell culture
Swiss 3T3 cells, originally developed by Todaro and Green [17], and kindly provided by Theresa Higgins (Cancer Research UK, London, UK) were cultured as described [9]. Cells were grown to confluence and used when quiescent, before the addition of PMT or bombesin (Calbiochem-Novabiochem). The tyrosine kinase inhibitors Su6656 and St638 (Calbiochem-Novabiochem) were prepared in DMSO, diluted in DMEM containing 0.1% DMSO and added to cell cultures to give a final concentration of 100 mM 1 h prior to treatment with PMT.

Preparation of Swiss 3T3 membranes and cytoplasmic fractions
Swiss 3T3 cells were grown in 145 mm dishes, rinsed twice with ice cold PBS and scraped into 2 ml of PBS containing proteinase inhibitors (Complete TM , Roche Diagnostics). Cells from 10 dishes were pooled, collected by centrifugation (200 g, 10 min, 4uC), and washed cell pastes were frozen at 270uC until required. The frozen cell pastes (,5 mg) were thawed on ice and suspended in 5 ml of membrane buffer (10 mM Tris-HCl, 10 mM MgCl 2 , 0.1 mM EDTA, pH 7.4, containing proteinase inhibitors). The cells were ruptured by 25 passes through a 23-gauge needle, and the resulting homogenate was centrifuged at 800 g for 10 min to remove unbroken cells and nuclei. The supernatants were transferred to fresh tubes and centrifuged at 50,000 g for 10 min. The supernatant containing cytoplasmic proteins was transferred to a fresh tube, snap frozen in liquid nitrogen and stored at 270uC. The pellet was washed and suspended in 10 ml of membrane buffer. After a second centrifugation step the membrane pellet was suspended in membrane buffer to a protein concentration of 1 mg/ml and stored at 270uC.

SDS PAGE and urea gel electrophoresis
Membrane proteins were resolved by SDS PAGE on 12.8% acrylamide/0.06% bis acrylamide gels, or on these same gels containing 6M urea to separate the closely migrating Ga 11 and Ga q subunits as described [18]. Proteins were transferred to PVDF membranes and immunoblotted as described below.

2-D gel electrophoresis
Swiss 3T3 membrane proteins were resolved by 2-D gel electrophoresis, as described [19]. The immunodetection of Ga subunits was performed by incubating the membrane overnight at 4uC with primary antibody at a dilution of 1:1000, followed by incubation with horseradish peroxidase-coupled secondary antibody at a dilution of 1:10000 (SouthernBiotech) for 1 h at room temperature. The membrane was incubated with ECL TM chemiluminescent substrate (GE HealthCare) and signals were detected using an automatic X-Ray film processor (Jungwon Precision Industries Co.).

Calcium microfluorimetry
Intracellular calcium was recorded as given previously [20]. Briefly, Swiss 3T3 cells were plated onto 19 mm glass cover slips and incubated in 5 mM Indo -AM (1 hour, 37uC, in the dark, Calbiochem). Cover slips were placed in a custom built chamber allowing gravity fed superfusion (10-12 ml/min) of a modified Krebs solution. Bombesin was applied by switching a multiway tap to a solution containing it and was removed by switching back to a bombesin free solution. The waste was removed by a peristaltic pump. Recordings were performed at room temperature by subtraction of background light and recording the emitted light from individual cells at 405 and 488 nm. The emission ratio (R) was converted to a calcium concentration after calibration (see reference 20] in which [Ca]i (nM) = 1028(R-0.86)/(12-R) and autofluorescence was less that 4%. .

Trypsin protection assay
The trypsin protection assay was adapted from Evanko et al. [21]. Briefly, membrane fractions (100 mg) were incubated with PMT, bombesin, GTPcS or GTP at the required concentrations at 37uC for times indicated. Membrane fractions were centrifuged at 18,0006g for 10 min at 4uC and the pellet was resuspended in 12.8 ml of solubilisation buffer (20 mM Tris-HCl, pH 7.5, 100 mM NaCl, 2 mM MgCl 2 , 0.1 mM EDTA, 1 mM dithiothreitol, 10% glycerol, 1% C 12 E 10 (polyoxyethylene 10-lauryl ether), 0.1 mM phenylmethylsulfonyl fluoride), vortexed, incubated on ice for 20 min and centrifuged at 18,0006g for 10 min at 4uC. The supernatant was then transferred to a new microfuge tube, treated with 4 ml of trypsin mixture (100 mM GDP, 1.5 mg/ ml trypsin in solubilisation buffer) for 30 min at 30uC. The trypsin activity was neutralised with 3 ml of soybean trypsin inhibitor (3 mg/ml). Trypsin-resistant fragments were resolved by SDS-PAGE, and detected by immunoblotting using antiserum against Ga q/11 . The induction of trypsin protection by GTPcS and GTP alone or in the presence of bombesin or PMT were quantified relative to untrypsinised G q using scanning densitometry (Gene-Tools, Syngene). Data were analysed using factorial analysis of variance (ANOVA) by Dr Ron Wilson (King's College London). Unactivated Ga subunits (GDP-bound) are highly susceptible to tryptic digestion; however tryptic cleavage is inhibited when Gproteins are activated (GTP-bound) as most cleavage sites are conformationally protected, and a product resulting from a small N-terminal cleavage can be visualised [22].

Measurement of high-affinity GTPase activity
Determination of GTPase activity was essentially as described [23]. High-affinity GTPase activity was determined by subtraction of P i release in membranes incubated with 50 mM of GTP (lowaffinity GTPase activity) from that with 0.5 mM GTP (total GTPase activity).

Results
PMT stimulates an acidic modification of Ga q and Ga i family proteins Ga q/11 antiserum detected both Ga q and Ga 11 subunits at an apparent molecular mass of 42 kDa in membranes prepared from quiescent Swiss 3T3 cells. Separating these subunits on a urea gel showed that Ga q (which aberrantly runs slower in this system than G 11 [24]) was more abundantly expressed than Ga 11 in these cells (Fig. 1A). A similar relative abundance has been shown in rat neurons [25]. Four distinct Ga q/11 molecular isoforms, designated q-II, q-III, q-V and q-VI, were resolved by 2-D gel electrophoresis followed by immunoblotting with anti-G q/11 antibody in membranes derived from untreated cells (Fig. 1B). These plus two additional isoforms, q-I and q-IV, were detected by 2D PAGE and Western blot analysis of membrane fractions derived from cells treated with PMT (150 pM) for 4 h ( Fig. 1C; Table 1).
Antiserum directed only against Ga q detected two isoforms with pI values corresponding to q-II and q-III ( Fig. 1D; Table 1) in untreated cells. The Ga q antiserum detected an additional isoform with a pI value corresponding to q-I in PMT-treated cells ( Fig. 1E; Table 1). Antiserum directed only against Ga 11 detected two isoforms in untreated cells with pI values corresponding to the isoforms q-V and q-VI ( Fig. 1F; Table 1) and one additional isoform with a pI value corresponding to q-IV in PMT-treated cells ( Fig. 1G; Table 1).
We excluded the possibility that the Ga 11 antibody could react with Ga q by testing the ability of the Ga q and Ga 11 antibodies to react with a recombinant Ga q subunit. The Ga q but not the Ga 11 antiserum could detect the Ga q subunit (Fig. 1H). The experimentally determined pI values for the Ga q/11 isoforms (Table 1) are similar to the predicted pI values of 5.48 and 5.70 for murine Ga q and Ga 11 , respectively [26,27].
The expression of the Ga i-1 , Ga i-2 and Ga i-3 subclasses, which have the widest tissue expression pattern of this family [28], was analysed in Swiss 3T3 cells using specific antisera. The Ga i-1-2 (directed against Ga i-1 and Ga i-2 ) and Ga i-1-3 antisera (directed against Ga i-1 , Ga i-2 and Ga i-3 ) each detected an abundant protein band at 40 kDa in membranes from Swiss 3T3 cells. The antiserum specific for only Ga i-1 detected a weak band ( Fig. 2A), although this antiserum could be shown to react strongly with a recombinant Ga i-1 subunit (Fig. 2B), demonstrating a low abundance of Ga i-1 in Swiss 3T3 cells.
Ga i-1 isoforms were present at low abundance in membranes prepared from either untreated or PMT-treated Swiss 3T3 cells as determined by 2-D gel electrophoresis followed by immunoblotting (Fig. 2C, D; Table 2). The Ga i-1-2 antiserum detected two Ga i isoforms in untreated and PMT-treated cells, designated i-I and i-II (Fig. 2E, F; Table 2), with a reproducible change in the relative abundance of the isoforms after PMT treatment. The Ga i-1-3 antiserum detected 3 Ga i isoforms in untreated cells, two of which appeared to correspond to i-I and i-II; the third isoform was designated i-IV ( Fig. 2G; Table 2). The Ga i-1-3 antiserum also detected these and one additional isoform, i-III in PMT-treated cells ( Fig. 2H; Table 2).
The predicted pI values of murine Ga i-1 , Ga i-2 and Ga i-3 are 5.69, 5.28 and 5.50, respectively [29]. It seems probable that isoforms i-I and i-II detected by the Ga i-1-2 antiserum belong to the Ga i-2 subclass, as isoforms of the Ga i-1 subclass are expected to have a more basic pI, and Ga i-1 was not detected in Swiss 3T3 cells. Isoforms i-III and i-IV are therefore likely to belong to the Ga i-3 subclass. Orth et al. resolved Ga i-1 and Ga i-2 from mouse embryonic fibroblast cells by 2-D gel electrophoresis at an unspecified pI value and showed that PMT treatment of these cells caused an acidic pI shift consistent with deamidated recombinant Ga i-2 [14]. Our results suggest that PMT catalyses the acidic covalent modification of Ga i-2 and Ga i-3 .

PMT induces an alkaline modification of Ga 13
The two members of the Ga 12 family, Ga 12 and Ga 13 , are ubiquitously expressed [30]. Ga 13 was detected in Swiss 3T3 membranes using antiserum against Ga 13 (Fig. 3A). Three Ga 13 isoforms, 13-I, 13-III and 13-IV, were identified in membranes from Swiss 3T3 cells (Fig. 3B). Two additional isoforms, 13-II and 13-V, were detected in membranes derived from PMT-treated cells ( Fig. 3C; Table 3). The additional Ga 13 isoforms seem to be the result of an alkaline pH shift, in contrast to the effect of PMT on Ga q/11 and Ga i isoforms. Under our experimental conditions, Ga 12 could not be resolved by 2-D gel electrophoresis.

PMT does not induce any modification of Ga s
The alpha subunits of the ubiquitously expressed G s family can be expressed as four distinct forms as a result of alternative mRNA splicing [31]. Swiss 3T3 cells were shown to express both large (55 kDa) and small (52 kDa) forms of Ga s , with Ga s -large being more abundantly expressed than Ga s -small (Fig. S1A). Six isoforms of Ga s -large (s-I to s-VI) and two isoforms of Ga s -small (s-VII and s-VIII) were resolved in membranes derived from Swiss 3T3 cells by 2-D gel electrophoresis, followed by immunoblotting with the Ga s/olf antiserum ( Fig. S1B; Table S1). The Ga s -large isoforms were detected at a more acidic pI than the Ga s -small isoforms, which concurs with previous findings [19]. PMT showed no discernable effect on the pI or molecular mass of the Ga s subunits ( Fig. S1C; Table S1).

PMT stimulates the stable covalent modification of Gproteins
It was important to establish whether the additional isoforms detected in PMT-treated cells arose as a consequence of normal activation induced by PMT or if they were directly PMTmodified. Cells were challenged with the neuropeptide bombesin, which acts through a G q -coupled receptor to stimulate phospholipase C (PLC) activation culminating in the release of Ca 2+ from intracellular stores [32]. Bombesin at a concentration of 30 nM effectively stimulated Ca 2+ release from cells (Fig. 4A), but no additional Ga q/11 isoforms were detected by 2D PAGE and Western blot analysis of a membrane fraction derived from cells exposed to bombesin (Fig. 4B, C). This suggested that Ga qcoupled receptor activation did not stimulate the stable covalent modification of Ga q .
We have previously demonstrated that PMT induced the phosphorylation of Ga q on Tyr369 [9]. We stimulated membrane fractions with bombesin in the presence of sodium vanadate, a potent tyrosine phosphatase inhibitor, in order to prevent the dephosphorylation of Ga q/11 . Bombesin activation of Ga q/11 in the membrane fractions was confirmed by the trypsin protection assay. Bombesin significantly enhanced GTPcS binding to Ga q/11 (p = 0.002), by up to 50% in some cases, the most likely explanation being that its action accelerated the rate of nucleotide exchange (Fig. 4D). The additional isoforms detected in membranes stimulated with bombesin appeared to be identical to those found in PMT-treated cells. However, the additional Ga q/11 and Ga i isoforms were also found in membranes derived from unstimulated cells, that had been treated with sodium vanadate alone (Fig. 4E, F and H). These findings suggest that PMT modification of Ga q/11 and Ga i produces a similar pI shift as the tyrosine phosphorylation of these Ga subunits.
The appearance of the additional isoforms observed in PMTtreated cells could not be blocked by the competitive kinase inhibitors Su6656 or St638, although these inhibitors were effective at blocking pervanadate-induced phosphorylation of focal adhesion kinase (FAK) (Fig. S2). We have previously shown that a mutant PMT (PMT C1165S ) can stimulate the tyrosine phosphorylation of G q , although it does not activate G q downstream signalling [9]. Treatment of Swiss 3T3 cells with PMT C1165S did not result in the covalent modification of Ga q or Ga i (Fig. S3). Moreover, tyrosine phosphorylation is a transient reversible modification that cannot be readily detected unless tyrosine phosphatases are inhibited.. The PMT-induced modification of Ga subunits was detected in the absence of sodium vanadate, indicating that the PMT-induced modification was covalent and stable.
Prolonged treatment of cells with PMT has differential effects on G-proteins PMT treatment decreased the abundance of some of the preexisting Ga q/11 and Ga i isoforms in membrane fractions. To explore if PMT caused G-protein removal from membranes, Swiss 3T3 cells were treated with PMT at a concentration of 1 nM for 16 h. This treatment did not cause loss of Ga q/11 from the membrane (Fig. 5A), but resulted in the complete loss of the most basic isoforms of Ga q and Ga 11 , q-III and q-VI, respectively (Fig. 5B, C, and D), while isoforms q-II and q-IV did not undergo an evident change in abundance. We speculate that the loss of detection of Ga q/11 isoforms q-III and q-VI is a result of the covalent modification of these isoforms to q-I and q-IV, respectively, induced by PMT.
In contrast, prolonged treatment of Swiss 3T3 cells with PMT generally resulted in the almost complete loss of Ga i from membranes (Fig. 5E, F). It is unlikely that the failure to detect the Ga i isoforms reflects a modification that interferes with the Ga i-1-3 antigen recognition site, which is at the C-terminus of Ga i , as the loss of Ga i from membranes could also be demonstrated with an antiserum against an internal epitope of Ga i-2 (Fig. 5F). Cytoplasmic extracts of cells that had received prolonged treatment with PMT were probed with anti-Ga i-1-3 antibody, but no increase in Ga i subunits could be detected in these fractions (Fig. 5G). It appears that the sequential loss of Ga i from membranes proceeds by covalent modification of Ga i isoforms i-II and i-IV to produce isoforms i-I and i-III, respectively, followed over time by the loss of isoforms i-I and i-III from the membranes (Fig. 5H-J). In some cases only partial loss of Ga i isoforms was observed over this time period (data not shown).
PMT inhibits the GTPase activity of G q PMT did not significantly enhance GTPcS binding to Ga q/11 in contrast to bombesin (data not shown). Due to its enzymatic nature, PMT required a longer incubation time to promote GTP binding to Ga q compared to bombesin [9]. Therefore, it is likely that during the course of the incubation, Ga q was gradually saturated by GTPcS, thereby preventing the detection of PMTenhanced GTPcS binding to Ga q above background levels. When GTP was used instead of GTPcS, PMT significantly enhanced GTP binding to Ga q as measured by trypsin protection (p = 0.03), by up to 30% (Fig. 6A), in contrast to bombesin (Fig. 6B). This The samples were as described in the legend to Figure 1 and the results are expressed as the mean 6 standard error of the mean (n = 3). doi:10.1371/journal.pone.0047188.t001 finding suggested that PMT might inhibit the GTPase activity of Ga q , to prevent the hydrolysis of GTP to GDP. Bombesin stimulated the steady-state GTPase activity in Swiss 3T3 membrane preparations by up to 30%, whereas pretreatment of cells with PMT at 150 pM for 4 h reduced the basal and bombesin-stimulated GTPase activity in membrane preparations (Fig. 6C). To further decrease the basal steady-state GTPase level, cells were pre-treated with cholera toxin, which ADPribosylates Ga s to inhibit its GTPase activity. Cholera toxin caused an increase in the molecular weight of both the long and short forms of Ga s , due to the addition of ADP ribose (Fig. 6D). Pretreatment of cells with both cholera toxin and PMT further decreased the basal GTPase activity in membrane preparations, compared to cells pre-treated with PMT alone. Bombesin stimulated the steady state GTPase activity by up to 50% in cells pre-treated with cholera toxin, whereas the additional pretreatment of cells with PMT reduced the bombesin-stimulated GTPase activity in membrane preparations, indicating that PMT inhibits the GTPase activity of Ga q but not G s (Fig. 6C).

Discussion
PMT executes its cellular effects through the activation of the heterotrimeric G-proteins, G q , G 12 and G i [8][9][10][11][12][13]. This has been shown to occur in recombinant G i by PMT-induced deamidation of Gln-205 to glutamic acid, which inhibits its intrinsic GTPase activity [14]. The work we report here complements these studies by investigating covalent modifications of G-proteins in Swiss 3T3 cells treated with PMT. PMT treatment consistently led to the appearance of new isoforms at a lower pI for both Ga q and Ga 11 . PMT also stimulated the covalent modification of members of the G i family. The Ga 12 family proteins, unlike the other G-protein families, have predicted pI values within the alkaline pH range (.pH 8) and such proteins are difficult to resolve by 2-D gel electrophoresis [33]. Ga 13 , but not Ga 12 , subunits displayed a reproducible pattern and PMT treatment led to new Ga 13 isoforms at slightly higher pI values. We found no evidence that PMT stimulates the covalent modification of Ga s , although the glutamine residue targeted by PMT is conserved in all G-proteins.
Stimulation of G q -coupled receptors by bombesin only resulted in the detection of the additional Ga q/11 isoforms observed in PMT-treated cells when vanadate was present. The addition of sodium vanadate per se led to a similar pattern of isoforms to those observed in PMT-treated cells. However it is likely that these different treatments lead to different modifications. The modification of Ga q/11 and Ga i stimulated by PMT was detected without sodium vanadate, and is thus indicative of a stable covalent modification such as deamidation, whereas tyrosine phosphorylation is a transient covalent modification. We previously showed that a src family kinase mediates the phosphorylation of G q in response to PMT [9]. However, pre-treatment of cells with a specific src kinase family inhibitor, SU6656, or a broad spectrum kinase inhibitor, St638, did not prevent PMT from stimulating the covalent modification of Ga q and Ga i , despite each kinase inhibitor being effective at blocking FAK phosphorylation. It is possible that the kinase inhibitors failed to completely block PMT-stimulated phosphorylation of G-proteins, due to their competitive nature and the enzymatic nature of PMT. However, this would suggest that deamidation by PMT results in the stable phosphorylation of these Ga subunits that is not reversed by the action of phosphatases, which is unlikely.
Deamidation and tyrosine phosphorylation of a Ga subunit would have a similar effect on the isoelectric point. The PMTinduced deamidation of in-vitro translated Ga q and recombinant Ga i-2 was reported to cause an acidic pI shift of 0.05 and 0.07, respectively [14]. This compares with the acidic pI shift of approximately ,0.15 for both Ga q and Ga i that we have observed. There are various possible interpretations of this apparent discrepancy. First, pI shifts are known to be variable and depend on the overall pI of a protein and its local context [34], and thus Ga i expressed in E. coli may behave differently because of the absence of post-translational modifications. Alternatively, the PMT-induced modification in cells may differ from that observed following expression in E. coli.
PMT is reported not to activate G 11 , as PMT could not induce the activation of PLC in G q -deficient cells [12], and further analysis using Ga q /Ga 11 chimeras also confirmed that PMT did The samples were as described in the legend to Figure 2    The samples were as described in the legend to Figure 4 and the results are expressed as the mean 6 standard error of the mean (n = 3). doi:10.1371/journal.pone.0047188.t003 not lead to G 11 -linked stimulation of PLC [35].We were therefore surprised that PMT stimulated the covalent modification of Ga 11 . Ga q and Ga 11 each contain Gln-209 that is functionally equivalent to Gln-205 in Ga i-2 and it would be unlikely that Ga 11 could be deamidated and yet not activated by PMT, as the loss of the functional Gln would affect the GTPase activity of the G-protein. While this manuscript was in preparation, Kamitani et al. published evidence that an antibody against deamidated Ga subunits recognised Ga 11 in PMT-treated mouse embryonic fibroblasts that were deficient in Ga q/11 but transfected to express Ga 11 [36]. This result provides further evidence that G 11 is also a substrate for PMT. In their experiments there was a small stimulation of PLC in cells expressing Ga 11 . All the other papers addressing this issue have used the same source of G q/11 -deficient MEF cells, whereas our work uses Swiss 3T3 cells. Further investigation of these puzzling and partially contradictory results is required. PMT treatment of cells led to new Ga 13 isoforms at slightly higher (0.09-0.15) pI values. The PMT catalytic triad has high structural similarity to eukaryotic transglutaminases [14], and it is possible that PMT can also function as a transglutaminase, in a similar manner to the cytotoxic necrotizing factor (CNF) which was originally considered to be a deamidase, but was later found to cause transglutamination in cells [37]. Transglutaminases catalyse the acyl transfer between the c-carboxyamide of a peptide bound glutamine (acyl donor) to a primary amine (acyl acceptor). When water functions as an acyl acceptor the result is glutamine deamidation [38]. The choice between deamidation and transglutamination is influenced by the environment of the targeted glutamine residue [39,40]. As transglutamination would impart a The proteins were then analysed for trypsin protection as described under Materials and Methods, and activated Ga q/11 was separated by SDS PAGE and Western blotted with anti-Ga q/11 antibody. Quantification of activated Ga q/11 (lower panel) was determined by densitometric scanning and these data were analysed using factorial analysis of variance (ANOVA). The induction of activation shown is relative to the density of the band without GTPcS or bombesin. Bombesin significantly enhanced GTPcS binding to Ga q/11 (* p = 0.002). Membrane proteins from Swiss 3T3 cells were incubated with (E) 1 mM sodium vanadate for 20 min at 37uC or (F) 1 mM sodium vanadate and 30 nM bombesin for 20 min at 37uC, proteins were separated by 2-D gel electrophoresis and Western blotted with anti-Ga q/11 antibody. Membrane proteins from Swiss 3T3 cells were incubated (G) without or (H) with 1 mM sodium vanadate for 20 min at 37uC, proteins were separated by 2-D gel electrophoresis and Western blotted with anti-Ga i-1-3 antibody. Samples from at least 3 independent membrane preparations were resolved with similar results. doi:10.1371/journal.pone.0047188.g004 positive charge to produce an alkaline shift, it is possible that PMT preferentially transglutaminates Ga 13 in cells.
The removal of G-proteins from the membrane is a regulatory phenomenon that can follow prolonged G-protein activation [41]. The ADP-ribosylation of G s by cholera toxin leads to its downregulation, although ADP-ribosylation of Ga i by pertussis toxin does not result in its degradation [42]. We observed that prolonged treatment of cells with PMT caused the loss of Ga i , but not Ga q , from membranes prepared from Swiss 3T3 cells. Furthermore Ga i could not be detected in the cytoplasm following prolonged PMT treatment. Orth et al. had suggested that overnight treatment of Swiss 3T3 cells with 1 nM PMT uncoupled Ga i from its receptor, as the G i -linked agonist lysophosphatidic acid could not stimulate GTPcS binding to Ga i in membranes derived from these cells [10]. The loss of G i from the membrane that we observed over this time period would provide a more likely explanation for their observation. Furthermore, the site of the PMT-induced modification, Gln-205, is not thought to be linked to receptor interaction. A similar differential degradation has been observed with Rho proteins following modification by CNF [43].
We found that PMT could promote the binding of GTP to Ga q/11 , whereas bombesin could not, which suggested that the action of PMT inhibits the GTPase activity of Ga q/11 . PMT significantly inhibited the bombesin-mediated stimulation of steady-state GTPase activity in Swiss 3T3 membrane preparations. These results complement the demonstration that PMT inhibits the GTPase activity of E. coli-expressed Ga i [10,14]. Furthermore, pre-treatment of cells with cholera toxin and PMT resulted in a greater inhibition of GTPase activity, supporting the view that PMT does not affect G s .
In conclusion, our results demonstrate that treatment of Swiss 3T3 cells with PMT induces the irreversible modification of Gproteins belonging to the G i and G q families resulting in an acidic pI shift, which is consistent with the observation that PMT catalyses deamidation of recombinantly expressed G i causing a similar shift in pI. We found that PMT inhibits the intrinsic GTPase activity of G q , which complements the finding that PMTstimulated deamidation of Ga i-2 inhibits its GTPase activity. We showed that stimulation of cells with PMT results in the degradation of G i which provides an explanation for the observation that PMT-treatment blocks G i activation by a receptor agonist. The unexpected modification of Ga 11 requires further investigation. We demonstrated that PMT treatment causes an alkaline pI shift in Ga 13 and speculate that PMT might preferentially transglutaminate Ga 13 . Working with cells enables the PMT/G-protein interaction to be investigated in a more natural context than when working with recombinantly expressed proteins. However, the further interpretation of results is impeded by the near impossibility of purifying these low abundance proteins in a modified form from cell lines, and thus both in-vitro and invivo studies are required to unravel the complexity of the toxin/Gprotein interactions. Table S1 Analysis of pI values of G s family isoforms after treatment with PMT. The samples were as described in Figure 6. PMT inhibits the GTPase activity of G q . (A) Membrane proteins were incubated in the presence or absence of 150 pM PMT for 1 h with 0.5 nM GTP and tested in a trypsin protection assay as described in Materials and Methods. Proteins were separated by SDS PAGE and Western blotted with anti-Ga q/11 antibody. Quantification of activated Ga q/11 (lower panel) was determined by densitometric scanning and the data were analysed using factorial analysis of variance (ANOVA). The induction of activation shown is relative to the density of the band without GTP or PMT. PMT significantly enhanced GTP binding to G q (* p = 0.03). (B) Membrane proteins were incubated in the presence or absence of 30 nM bombesin 20 min with 0.5 nM GTP and tested in a trypsin protection assay as described in Materials and Methods. Proteins were separated by SDS PAGE and Western blotted with anti-Ga q/11 antibody. Samples from at least 3 independent membrane preparations were resolved with similar results. (C) Membranes derived from Swiss 3T3 cells that had either been treated or untreated with 150 pM PMT for 4 h or 100 ng cholera toxin, or both, were treated with or without 30 nM bombesin for 20 min in the presence of [c-32 P] GTP. All the experimental conditions were repeated three times, and all data are presented as mean 6 standard deviation (SEM). The results for the groups were compared using single-factor analysis of variance (one-way ANOVA), followed by Newnan-Keuls test used to determine differences between groups. Significant changes are indicated by an asterisk (* P,0.05, *** P,0.001).