Pseudomonas aeruginosa exoenzyme S ADP-ribosylates Ras at multiple sites.

Pseudomonas aeruginosa exoenzyme S (ExoS) ADP-ribosylated Ras to a stoichiometry of approximately 2 molecules of ADP-ribose incorporated per molecule of Ras, which suggested that ExoS could ADP-ribosylate Ras at more than one arginine residue. SDS-polyacrylamide gel electrophoresis analysis showed that ADP-ribosylated Ras possessed a slower mobility than non-ADP-ribosylated Ras. Analysis of the ADP-ribosylation of in vitro transcribed/translated Ras by ExoS identified two electrophoretically shifted forms of Ras, which was consistent with the ADP-ribosylation of Ras at two distinct arginine residues. Analysis of ADP-ribosylated in vitro transcribed/translated Ras mutants possessing individual Arg-to-Ala substitutions showed that Arg-41 was the preferred site of ADP-ribosylation and that the second ADP-ribosylation event occurred at a slower rate than the ADP-ribosylation at Arg-41, but did not occur at a specific arginine residue. Analysis of bacterially expressed wild-type RasDeltaCAAX and RasDeltaCAAXR41K supported the conclusion that Arg-41 was the preferred site of ADP-ribosylation. Arg-41 is located adjacent to the switch 1 region of Ras, which is involved in effector interactions. Introduction of ExoS into eukaryotic cells inhibited Ras-mediated eukaryotic signal transduction since infection of PC-12 cells with an ExoS-producing strain of P. aeruginosa inhibited nerve growth factor-stimulated neurite formation. This is the first demonstration that ExoS disrupts a Ras-mediated signal transduction pathway.

Conditions such as cystic fibrosis, leukemia, neutropenia, and burn wounds predispose individuals to infections by P. aeruginosa (6). Numerous factors, including the production of exoenzyme S (ExoS 1 ; 453 amino acids), may contribute to the virulence of P. aeruginosa (7). ExoS, a member of the family of bacterial ADP-ribosyltransferases (8), requires the presence of a eukaryotic protein termed factor-activating exoenzyme S (FAS), a 14-3-3 protein, for expression of ADP-ribosyltransferase activity (9). Unique to most bacterial toxins, ExoS does not have rigid target protein specificity. In vitro, ExoS has been shown to ADP-ribosylate a number of targets, including IgG 3 (10), apolipoprotein A-I (10), vimentin (11), and several members of the Ras superfamily (12). Analysis of chemical sensitivity indicated that ExoS ADP-ribosylates these proteins at arginine residues (10,13).
Initial studies implicated a role for ExoS in the dissemination of P. aeruginosa from burn wounds (14) and in tissue destruction in chronic lung infections (15). However, recent studies have shown that the transposon mutant used in the earlier studies had a disruption in the gene encoding a component of its type III secretion system; thus, this mutant would have pleiotropic effects on the expression of other type III secreted factors in addition to ExoS (16). Two recent studies have implicated a role for the type III secretion apparatus in the delivery of ExoS into eukaryotic cells. Olson and co-workers (17) have shown that incubation of cultured cells with strains of P. aeruginosa expressing ExoS resulted in the ADP-ribosylation of Ras, whereas Forsberg and co-workers (18) have demonstrated that ExoS is cytotoxic to cultured cells when delivered by the type III secretion system of the heterologous host Yersinia.
In this study, we examined the biochemical aspects of the ADP-ribosylation of Ras by ExoS. We demonstrate that ExoS ADP-ribosylates c-Ha-Ras at more than one site, with Arg-41 being the preferred site of ADP-ribosylation. In addition, we show that infection of PC-12 cells by a strain of P. aeruginosa that produces ExoS inhibits nerve growth factor (NGF)-stimulated neurite outgrowth, which is the first demonstration that ADP-ribosylation by ExoS disrupts a Ras-mediated signal transduction pathway.  (19). P. aeruginosa 388 and P. aeruginosa 388⌬ExoS were cultured as described previously (16).

Materials
Purification of His-tagged Ras Proteins-Ras proteins were expressed in E. coli and purified as described previously (20). Briefly, protease inhibitors (phenylmethylsulfonyl fluoride and aprotinin) were added to the concentrated cell suspensions, which were disrupted with a French press. The lysate was centrifuged (30,000 ϫ g for 20 min) and filtered (0.45-m filter), and His-tagged Ras proteins were purified by Ni 2ϩ affinity chromatography (20). His-tagged Ras proteins were eluted with elution buffer containing 3 M GTP and 10 mM MgCl 2 , diluted to 40% glycerol, and stored at Ϫ20°C. These proteins were termed Ras⌬CAAX.
ADP-ribosylation of Ras⌬CAAX by ExoS-Reaction mixtures (25 l) contained 0.2 M sodium acetate (pH 6.0), with the indicated amount of [adenylate phosphate-32 P]NAD, and Ras⌬CAAX in the presence and absence of FAS and/or ExoS or ⌬N222 (a catalytic deletion peptide of ExoS) (20) at the indicated concentrations. Reactions were performed for 1 h at room temperature and stopped with 0.5 volume of gel loading buffer containing ␤-mercaptoethanol and boiling. Samples were analyzed by SDS-PAGE followed by autoradiography.
Construction of M13mp18Ras⌬CAAX and Site-directed Mutagenesis of Ras⌬CAAX-Ras⌬CAAX was engineered by deletion polymerase chain reaction mutagenesis of pet16b c-Ha-Ras using primers containing EcoRI and BamHI restriction sites at the 5Ј-and 3Ј-ends of the gene, respectively. The resulting polymerase chain reaction product was digested with XbaI and BamHI and ligated into the pET15b vector. The EcoRI-BamHI fragment, encoding the Ras⌬CAAX gene and the T7 promoter, was then cloned into M13mp18. M13mp18Ras⌬CAAX was used as the template for site-directed mutagenesis. Mutagenesis was performed essentially following the instructions of Amersham Pharmacia Biotech. DNA primers were constructed to be 10 base pairs complementary to the DNA template flanking the mutation of interest and encoded a Arg-to-Ala substitution (GC to CG). The presence of the mutation was confirmed by sequence analysis. The mutants were designated RnA, where arginine at residue n of Ras was changed to alanine.
ADP-ribosylation of in Vitro Transcribed/Translated M13ras by ExoS-M13mp18Ras⌬CAAX (M13ras) or mutated M13mp18Ras ⌬CAAX replicative form DNA was subjected to coupled in vitro transcription/translation (Promega). Transcription/translation reactions (12 l) contained 0.5 g of replicative form DNA and [ 35 S]Met (specific activity of 98 Ci/mmol). An aliquot of the transcription/translation reaction (2 l) was added to 4 volumes of gel loading buffer containing ␤-mercaptoethanol and boiled for 5 min. This sample was subjected to SDS-PAGE followed by autoradiography to monitor the electrophoretic mobility of intrinsically labeled [ 35 S]Met-Ras⌬CAAX. The transcription/translation mixture was also subjected to ADP-ribosylation by ExoS in reaction mixtures (25 l) containing 4 l of the transcription/ translation mixture and 0.6 M sodium acetate (pH 6.0) in the presence and absence of 0.1 M FAS, 960 M NAD, and either 12 or 1.2 nM ExoS as indicated. Reactions were stopped at the indicated times by adding 2 volumes of gel loading buffer containing ␤-mercaptoethanol and boiling for 5 min. Samples were subjected to SDS-PAGE followed by autoradiography. The electrophoretic mobility of this sample was compared with the electrophoretic mobility of intrinsically labeled GTP Dissociation Kinetics of Wild-type Ras and Ras⌬CAAXR41K-The rate of GDP/GTP dissociation of wild-type Ras⌬CAAX and Ras⌬CAAXR41K was measured essentially as described previously (21). Briefly, 2 M Ras⌬CAAX was incubated in 4 mM EDTA, 50 mM Tris (pH 7.6), and 5 mM dithiothreitol for 10 min at 37°C, and then 50 M [␣-32 P]GTP (specific activity of 1.4 Ci/mmol) and 10 mM MgCl 2 were added. After 30 min at 37°C, a 40-fold excess of unlabeled GTP was added. At defined times, 25-l aliquots were removed, filtered through nitrocellulose filters, and washed. Filter-bound radioactivity was determined by scintillation counting.
PC-12 Cell Culture-PC-12 cells were passaged in Dulbecco's modified Eagle's medium supplemented with 10% heat-inactivated horse serum and 5% fetal bovine serum. Twenty-four hours prior to an experiment, ϳ4 ϫ 10 3 cells were seeded per well in a 12-well culture dish coated with poly-L-lysine. For differentiation, the culture medium was changed to Dulbecco's modified Eagle's medium containing 1% horse serum and 100 ng/ml NGF (Boehringer Mannheim).

ADP-ribosylation of Ras⌬CAAX by ExoS-ExoS
ADP-ribosylates several eukaryotic proteins in vitro, including vimentin, IgG, and several members of the Ras superfamily (10,13). In this study, the ADP-ribosylation of c-Ha-Ras by ExoS was examined. To facilitate expression and purification, a deletion peptide of Ras (termed Ras⌬CAAX) was expressed as a Histagged protein in E. coli. Ras⌬CAAX possesses the complete GTP-binding domain and effector functions of Ras, but lacks the four carboxyl-terminal residues, which comprises the farnesylation sequence necessary for membrane targeting. ADPribosylation of Ras⌬CAAX by ExoS showed an absolute requirement for FAS and NAD (Fig. 1). ADP-ribosylated Ras⌬CAAX possessed a slower electrophoretic mobility as determined by SDS-PAGE relative to non-ADP-ribosylated Ras⌬CAAX. Other experiments showed that at saturation, ExoS had ADP-ribosylated Ras⌬CAAX at a stoichiometry of ϳ2 mol of ADP-ribose bound per mol of Ras⌬CAAX (Table I). This suggested that ExoS ADP-ribosylated Ras⌬CAAX at potentially two sites. Similar results were obtained when ExoS ADPribosylated full-length c-Ha-Ras (data not shown).
ADP-ribosylation of in Vitro Transcribed/Translated Ras⌬CAAX by ExoS-Biochemical characterization of eukaryotic proteins that had been ADP-ribosylated by ExoS indicated that arginine was the site of ADP-ribosylation (10). Since the presence of two potential sites for ADP-ribosylation within Ras could complicate a biochemical determination of the preferred site for ADP-ribosylation, a molecular approach was used to determine if ExoS ADP-ribosylated Ras at a preferred arginine residue. Briefly, the gene encoding Ras⌬CAAX was subcloned downstream of a T7 promoter, which was then subcloned into M13mp18 (M13ras). The replicative form of M13ras was isolated and subjected to in vitro transcription/translation using  Incubation of in vitro translated [ 35 S]Met-Ras⌬CAAX with exogenous NAD, FAS, and ExoS resulted in the detection of two electrophoretic mobility shifted forms of Ras⌬CAAX. This was consistent with ExoS ADP-ribosylating Ras at two distinct sites. Ras⌬CAAX showed little electrophoretic mobility shifting in reactions lacking NAD, whereas two electrophoretic mobility shifted forms of Ras⌬CAAX were observed in reactions lacking FAS. This indicated that the reticulocyte lysates used for in vitro transcription/translation of Ras⌬CAAX contained a small amount of NAD and saturating amounts of FAS relative to the concentrations used in the ADP-ribosyltransferase assay.
ADP-ribosylation of Ras⌬CAAX Containing Single Arginine Mutations by ExoS-M13ras single-stranded DNA was subjected to site-directed mutagenesis in which 12 individual Argto-Ala mutations (each mutant was designated RnA (Arg at residue n of Ras mutated to an Ala) were introduced into the ras open reading frame. M13ras replicative form DNA, encoding the wild type or an individual Arg-to-Ala mutation, was subjected to in vitro transcription/translation. In vitro translated [ 35 S]Met-Ras⌬CAAX was ADP-ribosylated by ExoS at a concentration of ExoS such that a single preferred site for ADP-ribosylation would be detected. Wild-type [ 35 S]Met-Ras⌬CAAX and 11 of the individual Arg-to-Ala mutants of Ras⌬CAAX, with the exception of Ras⌬CAAXR41A, showed an electrophoretic mobility shift upon ADP-ribosylation by ExoS. In contrast, Ras⌬CAAXR41A did not show an electrophoretic mobility shift upon incubation with ExoS. These data suggested that Arg-41 was the preferred site of ADP-ribosylation by ExoS (Fig. 3).
ADP-ribosylation of in vitro translated [ 35 S]Met-Ras⌬CAAX (the wild type and RnA mutants) was also assayed at higher concentrations of ExoS such that multiple sites within Ras would be ADP-ribosylated (Fig. 4). With the exception of Ras⌬CAAXR41A, ExoS ADP-ribosylated wild-type Ras⌬CAAX and the RnA mutants, with the appearance of two sequential electrophoretic mobility shifts, suggestive of the presence of double ADP-ribosylated Ras. In contrast, ExoS ADP-ribosylated Ras⌬CAAXR41A to only a single electrophoretic mobility shift, which was consistent with the presence of only a single site for ADP-ribosylation. These data also indicated that the second site of ADP-ribosylation was not at a specific arginine residue and that more than one arginine could be ADP-ribosylated at the second site. Since a third ADP-ribosylation event was not observed, it appeared that ADP-ribosylation at the second site excluded a third ADP-ribosylation event.
ADP-ribosylation of Several Arg-41 Mutants of Ras by ExoS-To support the identification of Arg-41 as the preferred site for ADP-ribosylation by ExoS, several more conservative site-directed mutations were engineered at residue 41, lysine (Ras⌬CAAXR41K) and glutamine (Ras⌬CAAXR41Q). As observed for Ras⌬CAAXR41A, low concentrations of ExoS did not ADP-ribosylate either Ras⌬CAAXR41K or Ras⌬CAAXR41Q (Fig. 5). A second positive control, Ras⌬CAAXG12V, was ADPribosylated by ExoS as measured by the generation of electrophoretic mobility shifted Ras.
ADP-ribosylation of Ras⌬CAAX and Ras⌬CAAXR41K by ExoS-To address ExoS-mediated ADP-ribosylation of Ras at Arg-41 in more detail, Ras⌬CAAX and Ras⌬CAAXR41K were analyzed as targets for ADP-ribosylation. Bacterially produced Ras⌬CAAX and Ras⌬CAAXR41K were expressed and purified to similar levels and reacted with polyclonal antisera to Ras by Western blotting (Fig. 6). To determine whether Ras⌬CAAX and Ras⌬CAAXR41K were functional, the GTP dissociation kinetics were measured. Both Ras⌬CAAX and Ras⌬CAAXR41K exhibited similar GTP dissociation kinetics (Table I), which indicated that the R41K mutation did not alter the global functional properties of Ras. Others have reported that mutations at Arg-41 did not alter the GTP dissociation kinetics (21).
Next, the ADP-ribosylation of Ras⌬CAAX and Ras⌬- b Proteins were subjected to ADP-ribosylation as described in the legend to Fig. 7. Results of two representative experiments are shown. The absolute stoichiometry of ADP-ribosylated Ras was determined by measuring the moles of ADP-ribose incorporated into Ras by subjecting the gel band corresponding to Ras to scintillation counting and dividing this value by the concentration of Ras present in the gel band. The absolute concentration of Ras was determined by subjecting an aliquot of acid-hydrolyzed Ras to amino acid composition analysis.
c Proteins were subjected to ADP-ribosylation as described in the legend to Fig. 8. Results of two representative experiments are shown.

FIG. 2. ADP-ribosylation of in vitro transcribed/translated Ras⌬CAAX requires the addition of exogenous NAD and ExoS.
In vitro transcribed/translated [ 35 S]Met-Ras⌬CAAX was subjected to ADP-ribosylation by ExoS in the presence (ϩ) or absence (Ϫ) of the indicated reagents for 10 or 120 min at room temperature. ExoS (12 nM) was added to reaction mixtures as indicated. Reaction mixtures were analyzed by SDS-PAGE followed by autoradiography. Electrophoretic migrations are indicated as Ras⌬CAAX (arrow) and ADPribosylated Ras⌬CAAX (*, single-shifted Ras⌬CAAX; **, double-shifted Ras⌬CAAX). CAAXR41K by ExoS was studied. At saturation, ExoS ADPribosylated wild-type Ras⌬CAAX to a stoichiometry of ϳ2 mol of ADP-ribose incorporated per mol of Ras (Table I), whereas ExoS ADP-ribosylated Ras⌬CAAXR41K to a stoichiometry of 0.7 mol of ADP-ribose incorporated per mol of Ras. Examination of the Coomassie Blue-stained gel and autoradiogram showed that, relative to non-ADP-ribosylated Ras, ADP-ribosylated Ras⌬CAAX migrated with an altered electrophoretic mobility on SDS-PAGE, whereas ADP-ribosylated Ras⌬CAAX-R41K did not have an altered electrophoretic mobility on SDS-PAGE (Fig. 7). The fact that ADP-ribosylated Ras⌬CAAXR41K did not have an altered electrophoretic mobility in this experiment is due to fact that this experiment was performed using a 12% polyacrylamide gel, whereas the second ADP-ribosylation event was observed upon separation on a 15% polyacrylamide gel, as seen in Fig. 4. Under linear velocity conditions, ExoS ADP-ribosylated wild-type Ras⌬CAAX at a faster velocity than Ras⌬CAAXR41K. A representative graph depicting the velocity of the ADP-ribosylation of wild-type Ras⌬CAAX and Ras⌬CAAXR41K by ExoS is shown in Fig. 8. The average of three independent experiments indicated that ExoS ADP-ribosylated wild-type Ras⌬CAAX at 5-fold greater velocity than Ras⌬CAAXR41K (Table I).
Infection of PC-12 Cells with P. aeruginosa Expressing ExoS Inhibits NGF-stimulated Neurite Formation-Arg-41 is located next to the switch 1 region of Ras (residues 30 -39), a domain that interacts with downstream effectors (Fig. 9) (22). Thus, ADP-ribosylation of Ras at Arg-41 could potentially disrupt eukaryotic signal transduction by inhibiting the interaction of Ras with downstream effectors. The ability of ExoS to disrupt eukaryotic signal transduction was determined by analysis of NGF-stimulated neurite outgrowth in PC-12 cells. NGF-stimulated neurite outgrowth in PC-12 cells follows a Ras-mediated signal transduction pathway, although there are also Ras-independent components to the pathway (23). Since ExoS is secreted from P. aeruginosa via a type III secretion pathway (16), ExoS was delivered into PC-12 cells by infection with P. aeruginosa. PC-12 cells were incubated alone, with a strain of P. aeruginosa that produced ExoS (388), or with a strain of P. aeruginosa that was genetically deleted for exoS (388⌬ExoS). After 2 h, PC-12 cells were washed to remove unbound bacteria and incubated alone or with NGF. At 24 h post-infection, PC-12 cells were examined by phase microscopy. PC-12 cells that were incubated in the absence of bacteria and NGF showed a rounded morphology, typical of unstimulated cells (Fig. 10). In contrast, PC-12 cells that had been incubated alone or with a strain of P. aeruginosa that lacks exoS (388⌬ExoS) and then incubated with NGF had undergone a morphological change, where the cells had spread and exhibited neurite outgrowth. PC-12 cells that had been incubated with the ExoS-expressing strain of P. aeruginosa followed by incubation with NGF showed a rounded morphology, similar to the morphology of non-infected, non-NGF-stimulated cells. These data indicated that infection of PC-12 cells with an ExoS-producing strain of P. aeruginosa inhibited NGF-stimulated signal transduction leading to neurite outgrowth. In other experiments, we observed that transient expression of ExoS from a cytomegalovirus promoter also inhibited NGF-mediated neurite outgrowth in PC-12 cells. 2 Together, these data indicated that inhibition of NGF-mediated neurite outgrowth was due to the ADP-ribosyltransferase activity of ExoS and not a general cytotoxicity of ExoS or P. aeruginosa for PC-12 cells. DISCUSSION Earlier studies predicted that ExoS ADP-ribosylates eukaryotic proteins at arginine residues (10,12). We initially pursued a biochemical analysis of the site of ADP-ribosylated Ras. However, we found that, during proteolysis, a considerable amount of radiolabel was released as ADP-ribose and that the radiolabel that remained peptide-associated was present in several peptides. 2 This, in addition to the observation that ExoS ADPribosylated Ras to a stoichiometry of ϳ2 (Table I), prompted a molecular approach for the characterization of the site of ADPribosylation. The subsequent determination that ExoS ADPribosylates Ras at multiple sites is consistent with our inability to identify a specific arginine as the preferred site of ADP-ribosylation by biochemical approaches. Characterization of in vitro transcribed/translated forms of Ras that each possessed one of 12 individual Arg-to Ala substitutions identified Arg-41 as the preferred site for ADP-ribosylation since the Arg-41 mutants of Ras were not ADP-ribosylated by low concentrations of ExoS. Also, examination of the velocities of ADP-ribosylation of Ras by ExoS showed that Arg-41 was essentially completely ADP-ribosylated prior to the appearance of the second ADP-ribosylation event, which was consistent with Arg-41 being the high affinity site of ADP-ribosylation within Ras. 2 Analysis of the single Arg mutants of Ras did not identify a specific Arg as the second site for ADP-ribosylation since all of the single Arg mutants of Ras, with the exception of Ras R41A, could be double ADP-ribosylated. These data indicated that one of several arginine residues could constitute the second site of ADP-ribosylation within Ras. Since a third ADP-ribosylation event was not observed, it appeared that the ADP-ribosylation at one arginine residue excluded a third ADP-ribosylation event. This suggests that the arginines that are targeted for the second ADP-ribosylation event are located in close proximity in the three-dimensional structure of Ras. These data also showed that the ADP-ribosylation at Arg-41 and that at the second arginine were independent events and not the result of sequen-2 A. K. Ganesan and J. T. Barbieri, unpublished data. tial, ordered ADP-ribosylation reactions since the Arg-41 mutants could be ADP-ribosylated at the second arginine residue. There is precedence for the ADP-ribosylation of proteins at two sites. Aktories and co-workers (24) recently showed that an endogenous eukaryotic ADP-ribosyltransferase modifies actin at Arg-95 and Arg-372.
The ability of ExoS to ADP-ribosylate Ras at multiple arginine residues assists in addressing an earlier report by Coburn et al. (12) that described the ADP-ribosylation of several members of the Ras superfamily by ExoS and the subsequent prediction that ExoS ADP-ribosylated Ras at Arg-123 (13). Our data could be consistent with Arg-123 being one of the secondary sites of ADP-ribosylation within Ras.
Examination of the x-ray crystallographic structure of Ras (25) predicts that Arg-41 lies adjacent to the switch 1 domain, which has been shown to function in the interactions with downstream effectors, particularly Raf (22). Alignment of the primary amino acid sequences of members of the Ras superfamily of small molecular weight GTP-binding proteins showed that Arg-41 is not a highly conserved residue. Thus, it appears that ExoS can potentially ADP-ribosylate the Arg-41 homologue of Ras, Ral, and Rap, but that members of the Rab or Rho family will not be targeted for ADP-ribosylation by ExoS at the Arg-41 homologue. However, the observation that ExoS may ADP-ribosylate Ras at a second arginine residue poses the possibility that other members of the Ras superfamily that lack the Arg-41 homologue, including Rho and Rab, may be targeted for ADP-ribosylation by ExoS at an arginine residue that corresponds to the second site of ADP-ribosylation within Ras. The ADP-ribosylation of other Ras superfamily members by ExoS is currently under investigation.
Neurite outgrowth in PC-12 cells is a multistep process that is initiated by NGF binding to the Trk membrane tyrosine kinase. NGF binding stimulates Trk kinase activity, which results in the phosphorylation of several effector molecules that activate Ras. This pathway has both Ras-dependent and -independent components. Several studies have suggested that Ras is a dominant element in the transmission of NGF signals from Trk, including microinjection of antibodies to Ras, which interferes with the signaling pathway (26), or introduction of dominant inhibitory Ras alleles into PC-12 cells, which inhibits NGF-induced neurite outgrowth. In addition, transfection of constitutively active Ras, in the absence of NGF stimulation, is sufficient to induce neurite outgrowth (26). The Ras switch 1 domain is required for this signal transduction process since point mutations within the switch 1 regions of constitutively active RasV12 interfere with neurite outgrowth (21). The fact that ExoS preferentially modifies Ras at a region adjacent to the switch 1 domain and that infection of PC-12 cells with strains of P. aeruginosa expressing ExoS results in an inhibition of neurite outgrowth in PC-12 cells is consistent with the model that ExoS inhibits Ras-mediated signal transduction in vivo. However, since ExoS has the potential to modify a wide variety of targets in vitro, it is conceivable that ExoS can disrupt PC-12 cell signal transduction by the ADP-ribosylation of components of this signal transduction pathway. Future studies will focus on whether ExoS disrupts PC-12 cell signal transduction via modification of other components involved in PC-12 cell signal transduction.