INTRODUCTION

An increase in the tyrosine phosphorylation of the non-receptor protein tyrosine kinase p125^fak (1, 2) and the cytoskeletal associated protein paxillin (3, 4) has recently been identified as an early event in the action of diverse signaling molecules that mediate cell growth and differentiation (5), including mitogenic neuropeptides (6-8), growth factors such as PDGF¹ (9, 10), bioactive lipids (11-15), extracellular matrix proteins (16-20), transforming variants of pp60^src (17, 21), and the potent mitogenic toxin PMT (22). p125^fak lacks SH2 and SH3 domains but associates with other proteins including Src, paxillin, and p130^cas (1, 2, 23). Paxillin is a multidomain protein that may function as an adaptor capable of complex formation with p125^fak, Crk, Src, and vinculin (3, 4, 23). Gene disruption experiments indicate a critical role of p125^fak in embryonic development and cell locomotion (24). The increases in p125^fak and paxillin tyrosine phosphorylation are accompanied by profound alterations in the organization of the actin cytoskeleton and in the assembly of focal adhesions (9, 13, 15, 25, 26), the distinct areas of the plasma membrane where p125^fak and paxillin are localized (1, 2, 27). The small G protein p21^rho, a member of the Ras superfamily of small GTP-binding proteins, has been implicated in the mitogen-stimulated formation of focal adhesions and actin stress fibers as well as in the tyrosine phosphorylation of p125^fak and paxillin (22, 25, 28-30). These findings suggest the existence of a distinct signal transduction pathway in which p21^rho is upstream of cytoskeletal reorganization and tyrosine phosphorylation of focal adhesion proteins (5).

The mechanism of action of bacterial toxins has provided novel insights into the control of cellular regulatory processes, including signal transduction and cell proliferation. For example, the Clostridium botulinum C3 exoenzyme and the enterotoxins A and B from Clostridium difficile which selectively inactivate members of the Rho subfamily, have provided useful tools to evaluate the role of these small G proteins in signal transduction and cytoskeletal organization (31-33). In contrast to these clostridial toxins, CNF toxins produced by some pathogenic strains of Escherichia coli (34) and DNT from Bordetella bronchiseptica (35) directly target and activate p21^rho (36, 37). CNF and DNT induce actin reorganization and multinucleation in several cell types (37-40) but their effects on signal transduction pathways, specially on p125^fak and paxillin tyrosine phosphorylation have not been examined.

Here we report that CNF1 and DNT stimulate tyrosine phosphorylation of p125^fak and paxillin and induce a concomitant increase in the formation of actin stress fibers and in the assembly of focal adhesion plaques in Swiss 3T3 cells. Microinjection of C3 exoenzyme prevented both the cytoskeletal responses and the increase in tyrosine phosphorylation of focal adhesion proteins. In contrast to most other stimuli that promote tyrosine phosphorylation of p125^fak and paxillin (6-15, 22), CNF1 and DNT do not activate phospholipase C-mediated events including inositol phosphate production, Ca²⁺ mobilization, and PKC activation. In addition, CNF1 and DNT both stimulated reinitiation of DNA synthesis but neither induce activation of p42^mapk (ERK2). Thus, our results demonstrating that CNF1 and DNT induce a specific subset of molecular and cytoskeletal responses in Swiss 3T3 cells provide novel evidence for the existence of a signal transduction pathway that links p21^rho activation to tyrosine phosphorylation of focal adhesion proteins.

EXPERIMENTAL PROCEDURES

Stock cultures of Swiss 3T3 fibroblasts were maintained in DMEM supplemented with 10% fetal bovine serum in a humidified atmosphere containing 10% CO₂ and 90% air at 37 °C. For experimental purposes, cells were plated either in 30-mm Nunc Petri dishes at 10⁵ cells/dish, or in 90-mm dishes at 6 × 10⁵ cells/dish, in DMEM containing 10% fetal bovine serum and used after 6-8 days when the cells were confluent and quiescent (41).

Quiescent cultures of Swiss 3T3 cells were washed twice with DMEM and incubated with DMEM/Waymouth's medium 1:1 (v/v) containing [³H]thymidine (1 µCi/ml) and various factors as indicated. After 40 h, the cultures were washed twice with ice-cold phosphate-buffered saline and incubated in 5% trichloroacetic acid for 30 min at 4 °C. Trichloroacetic acid was then removed and the cultures were washed twice with ethanol and extracted in 1 ml of 2% Na₂CO₃, 0.1 M NaOH, 1% SDS. Incorporation of [³H]thymidine was determined by scintillation counting in 6 ml of scintillation fluid. The proportion of cells in the G₀/G₁, S, G₂, and M phases of the cell cycle were determined by fluorescence-activated cell sorter analysis as described previously (42).

Quiescent cultures of Swiss 3T3 cells (1-2 × 10⁶) were washed twice with DMEM, treated with CNF1, DNT, or PMT or other factors in 10 ml of DMEM/Waymouth (1:1, v/v) for the times indicated and lysed at 4 °C in 1 ml of a lysis buffer solution containing 10 mM Tris/HCl, pH 7.6, 5 mM EDTA, 50 mM NaCl, 30 mM sodium pyrophosphate, 50 mM sodium fluoride, 100 µM Na₃VO₄, and 1% Triton X-100. Proteins were immunoprecipitated at 4 °C overnight with agarose-linked mAbs directed against phosphotyrosine, paxillin, or p125^fak as indicated. Immunoprecipitates were washed three times with lysis buffer and extracted for 10 min at 95 °C in 2 × SDS-PAGE sample buffer (200 mM Tris-HCl, 6% SDS, 2 mM EDTA, 4% 2-mercaptoethanol, 10% glycerol, pH 6.8) and analyzed by SDS-PAGE.

Treatment of quiescent cultures of cells with factors, cell lysis, and immunoprecipitations were performed as described above. After separation by SDS-PAGE, proteins were transferred to Immobilon membranes (43). Membranes were blocked using 5% nonfat dried milk in PBS, pH 7.2, and incubated for 3-5 h at 22 °C with PY72 or 4G10 anti-Tyr(P) mAbs (1 µg/ml). Immunoreactive bands were visualized using ¹²⁵I-labeled sheep anti-mouse IgG followed by autoradiography. Autoradiograms were scanned using an LKB Ultrascan XL internal integrator. The values expressed represent percentages of the maximum increase in tyrosine phosphorylation above control values.

Quiescent and confluent cultures of Swiss 3T3 cells in 30-mm dishes were washed twice in phosphate-free DMEM and incubated at 37 °C with this medium containing 50 µCi/ml carrier-free [³²P]P_i. After 12 h, various factors were added for the indicated times. The cells were subsequently lysed and the lysates were immunoprecipitated with specific anti-80K/MARCKS antibody (44).

Cultures of Swiss 3T3 cells in 33-mm dishes were labeled for 16-18 h in 1 ml of DMEM/Waymouth's medium (1:1) containing 10 µCi of [2-³H]inositol. Additions were made to the cells as described for each experiment and LiCl was added to a final concentration of 20 mM for the last 30 min of the incubation. Inositol phosphates were extracted by replacing the medium with 1 ml of ice-cold 3% HClO₄. After 20 min at 4 °C the extract was neutralized with 0.5 M KOH containing 25 mM HEPES, 5 mM EDTA, and 0.01% phenol red. Precipitated KClO₄ was removed by centrifugation. Analysis of total inositol phosphates was by anion-exchange column chromatography (45). Samples were diluted to 10 ml with water and then loaded onto 1 ml of Dowex AG1-X8 (100-200 mesh, HCOO form) in Bio-Rad Econo-columns. After 3 washes with 3 × 10 ml of H₂O and 2 × 10 ml of 60 mM NH₄COOH, 5 mM Na₂B₄O₇ the inositol phosphates were eluted with 7 ml of 1 M NH₄COOH, 0.1 M HCOOH. An aliquot (1 ml) of eluate was counted in 10 ml of scintillation fluid.

To determine the intracellular concentration of Ca²⁺, [Ca²⁺]_i, 5 × 10⁵ cells were subcultured with 10 ml of DMEM containing 10% fetal bovine serum in 90-mm Nunc dishes. After 6-8 days, when the cells were confluent and quiescent, the medium was replaced with 5 ml of DMEM/Waymouth's medium (1:1). The cells were pretreated as indicated, and then 5 µl of 1 mM fura-2/AM (46-48) was added directly to the medium. After a further 10 min the cells were washed twice at 37 °C with 3 ml of electrolyte solution, comprising 140 mM NaCl, 5 mM KCl, 1.8 mM CaCl₂, 0.9 mM MgCl₂, 25 mM glucose, 16 mM HEPES, 6 mM Tris, pH 7.2, and the amino acid content of DMEM. The cells were then gently removed from the dish by scraping into 2 ml of the electrolyte solution. The cell suspension was transferred to a 1-cm² quartz cuvette and stirred at 37 °C for 4 min prior to any additions. Fluorescence was measured in a Perkin-Elmer LS-5 fluorometer with an excitation wavelength of 510 nm. [Ca²⁺]_i was calculated from the maximum and minimum fluorescence of the fura-2, as described (46, 49).

Quiescent Swiss 3T3 cells were washed twice with DMEM and incubated for the indicated times in DMEM/Waymouth's medium (1:1, v/v) at 37 °C with CNF1, DNT, or PMT. For actin staining, cells were washed once with PBS, fixed in 4% paraformaldehyde in PBS for 10 min at room temperature, and permeabilized with PBS containing 0.2% Triton X-100 for 8 min at room temperature. The cells were then incubated with FITC-conjugated phalloidin (0.25 µg/ml) in PBS for 30 min at room temperature and visualized using a Zeiss Axiophot immunofluorescence microscope. In experiments in which quiescent Swiss 3T3 cells were labeled with both FITC-conjugated phalloidin and anti-vinculin antibody, the cells were fixed and permeabilized as described above and then stained with a mixture of FITC-conjugated phalloidin (0.25 µg/ml) and anti-vinculin antibody (dilution 1:100) for 30 min at room temperature. Cells were subsequently washed three times in PBS and then incubated with Cy3-labeled rabbit anti-mouse IgG as a secondary antibody at a dilution of 1:100 for another 30 min at room temperature.

For microinjection experiments Swiss 3T3 cells were plated in 33-mm Nunc Petri dishes at 10⁵ cells/dish in DMEM containing 10% fetal bovine serum and used after 6-8 days when the cells were confluent and quiescent. Approximately 50 cells were microinjected with 100 µg/ml recombinant C3 exoenzyme. The efficiency of injection was determined by co-injecting guinea pig immunoglobulin at 0.5 mg/ml followed by staining with FITC-linked goat anti-guinea pig IgG. A single batch of C3 transferase was used in all experiments described.

Confocal imaging was performed using a Bio-Rad MRC 600 laser scanning head fitted to a Nikon Optiphot microscope. A 60X N.A./1.4 planapochromat oil immersion lens (Nikon) was used for all imaging. FITC-conjugated phalloidin and Cy3 anti-mouse IgG fluorochromes were excited at 488 and 568 nm, respectively, using a krypton/argon mixed gas laser (Bio-Rad). Two filter blocks were used, K1 and K2. K1 is a double dichromic filter enabling excitation at the wavelengths of 488 and 568 nm, whereas the K2 filter is a 560-nm dichromic combined with 522-nm green emission and 585-nm red emission filters. Images were collected using the Kalman filter. Care was taken to ensure that the FITC-conjugated phalloidin channel was sufficiently bright relative to the fluorescein signal to minimize the contribution of bleed-through from the green channel into the red channel (approximately 10%). Correction of images for bleed-through and other processing was carried out using COMOS and SOM programs (Bio-Rad) run on a Compaq Deskpro 66M 486 computer (66 MHz). Data are presented as projections of sequential optical sections. For Z-series, optical sections were recorded at 0.5-µm intervals. Final images were photographed directly from the VDU screen.

Activation of p42^mapk and p44^mapk was determined by the appearance of slower migrating forms in gel electrophoresis due to phosphorylation of specific threonine and tyrosine residues (50). Lysates from quiescent cells in 33-mm dishes prepared as above were subjected to SDS-PAGE and transferred to Immobilon membranes. Membranes were blocked using 5% nonfat dried milk in phosphate-buffered saline. Rabbit polyclonal antibodies raised against COOH-terminal peptides (EETARFQPGYRS for p42^mapk and IFQETARFQPGAPE for the p44^mapk) were used at 1/1000 dilution, and ¹²⁵I-protein A was used to visualize immunoreactive bands.

Strain pISS392 is E. coli DH5 containing plasmid pISS392 which consists of a 3.5-kilobase fragment encoding only CNF1 inserted into the vector pGEM3. Control strain pGEM3 is E. coli DH5 containing a plasmid derived from pISS392 by deletion of a 3.5-kilobase insert. It is pGEM3 with a deletion in the multiple cloning site. Strain DNT1 is E. coli XL1-Blue containing plasmid pDNT1 which consists of a 5-kilobase fragment encoding the DNT gene (B. bronchiseptica B58) inserted into the vector pBluescript. Control strain pBlue is E. coli XL1-Blue with plasmid pBluescript (51). Strain pTOX2 is E. coli XL1-Blue with plasmid pTOX2 which consists of the vector pBluescript with an approximately 4-kilobase pair insert encoding PMT. Control strain pBlue is E. coli XL1-Blue with plasmid pBluescript here nominated PC12.

The E. coli strains were grown overnight from a fresh colony. They were lysed in the lysis buffer (50 µg/ml lysozyme, 5 mM EDTA, 0.1% toluene, 0.05 M Tris-Cl, pH 7.2, 0.1 M NaCl) treated with RNase and DNase at 10 µg/ml each to remove nucleic acid and filtered through a 0.2-µm nitrocellulose filter.

Bombesin, cytochalasin D, FITC-conjugated phalloidin, monoclonal anti-vinculin mAb, Cy3-linked anti-mouse IgG, guinea pig IgG, and FITC-conjugated goat anti-guinea pig IgG were obtained from Sigma. Recombinant PDGF (BB homodimer), ¹²⁵I-sheep anti-mouse IgG (15 mCi/mg), carrier-free [³²P]P_i, and [2-³H]inositol (18.8 Ci/mmol, 1 Ci = 37 GBq) were all supplied by Amersham Corp., United Kingdom. Agarose-linked anti-Tyr(P) mAb was purchased from Oncogene Science Inc., New York. 4G10 anti-Tyr(P) mAb was from UBI, Lake Placid, NY. mAb 2A7 directed against p125^fak was from TCS Biologicals Ltd., Buckingham, United Kingdom. p125^fak immunoblotting was performed with mAb from Transduction Laboratories, Lexington, KY. The anti-Tyr(P) mAb PY72 was obtained from the hybridoma development unit, Imperial Cancer Research Fund. Clone pISS392 was kindly supplied by Dr. A. Caprioli, Instituto Superiore di Sanita, Rome, Italy. All other reagents were of the highest grade commercially available. The C3 C. botulinum exoenzyme was a gift from Dr. N. Morii and Professor S. Narumiya, Department of Pharmacology, Kyoto University Faculty of Medicine, Sakyo-ku 606, Japan.

RESULTS

To examine the effects of CNF1 on tyrosine phosphorylation, quiescent cultures of Swiss 3T3 cells were treated with the control bacterial lysate (pGEM3) or with lysate containing CNF1 for 6 h. Cell extracts were immunoprecipitated with anti-Tyr(P) mAb followed by Western blotting with an anti-Tyr(P) mAb. As shown in Fig. 1A, CNF1 markedly stimulated the tyrosine phosphorylation of a group of bands migrating with an apparent M_r of 110,000-130,000 and 70,000-80,000.

The pattern of tyrosine-phosphorylated proteins induced by CNF1 is similar to that stimulated by bombesin, bioactive lipids, and PMT in Swiss 3T3 cells (8, 13, 15, 22). The cytosolic tyrosine kinase p125^fak and the adaptor protein paxillin have been identified as prominent tyrosine-phosphorylated proteins in Swiss 3T3 cells treated with these agents. To determine whether these cellular proteins were also substrates for CNF1-induced tyrosine phosphorylation, extracts from Swiss 3T3 cells, incubated with the control bacterial lysate (pGEM3) or CNF1 for 6 h, were immunoprecipitated with mAbs that recognize either p125^fak or paxillin, and the immunoprecipitates were analyzed by Western blotting with anti-Tyr(P) mAb. As shown in Fig. 1A, CNF1 markedly stimulated tyrosine phosphorylation of p125^fak and paxillin in Swiss 3T3 cells. The control bacterial lysate (pGEM3) did not stimulate tyrosine phosphorylation of these bands or of any other protein (data not shown).

Neuropeptides and lysophosphatidic acid induce tyrosine phosphorylation of p125^fak and paxillin within minutes of treatment (6, 8, 13), whereas there is a lag period of 1 h between the addition of PMT and a detectable increase in tyrosine phosphorylation of these proteins (22). As shown in Fig. 1B, CNF1-stimulated tyrosine phosphorylation of p125^fak and paxillin occurred after a lag period of 2 h, reached a maximum after 8 h of treatment and declined to 56% ± 9.5 (mean ± S.E.) of the maximum after 24 h. This is in contrast to the tyrosine phosphorylation of p125^fak and paxillin stimulated by PMT which remains maximal after 24 h of treatment (data not shown). Immunoblotting with p125^fak mAb of anti-p125^fak immunoprecipitates prepared in parallel with those used for the assays of tyrosine phosphorylation verified that similar amounts of p125^fak were recovered after different times of CNF1 treatment (Fig. 1B).

Next we examined whether the lag period in CNF1 action reflects a requirement for internalization and activation of the toxin during transit through endosomal/lysosomal compartments. The lysosomotrophic agent methylamine, a membrane-permeant weak base known to inhibit lysosomal processing, completely blocked tyrosine phosphorylation of the M_r 110,000-130,000 and M_r 70,000-80,000 bands in response to CNF1 (Fig. 1C). Methylamine also prevented the increase in tyrosine phosphorylation of p125^fak in response to CNF1 (Fig. 1C). The inhibitory effect of methylamine was selective because it did not prevent the increase in tyrosine phosphorylation induced by bombesin in parallel cultures (data not shown). It is also known that the entry of many bacterial toxins into the cell cytoplasm is temperature-dependent. Treatment of Swiss 3T3 cells with CNF1 for 6 h at 22 °C failed to induce tyrosine phosphorylation (Fig. 1C). In contrast, bombesin stimulated tyrosine phosphorylation of M_r 110,000-130,000 and M_r 70,000-80,000 bands in parallel cultures incubated at 22 °C (data not shown). Therefore, CNF1 appears to enter cells via an endosomal/lysosomal pathway where it is processed and released into the cytosol in an active form to stimulate tyrosine phosphorylation of p125^fak and paxillin.

Given the localization of p125^fak and paxillin to focal contacts which form at the end of actin stress fibers we examined the kinetics of actin cytoskeleton reorganization and focal adhesion assembly induced by CNF1. As shown in Fig. 2, CNF1 treatment of Swiss 3T3 cells induced the formation of new actin stress fibers after a 2-h lag period. Actin reorganization reached a maximum after 8 h of exposure to CNF1. Interestingly, after 24 h of CNF1 treatment, the number of actin stress fibers decreased and actin appeared to accumulate in a cortical fashion. Vinculin staining showed the recruitment of this protein to focal adhesions formed at the ends of actin stress fibers. The number of focal adhesions also reached a maximum after 8 h and decreased after 24 h treatment (Fig. 2). The control bacterial lysate did not induce actin stress fiber formation or focal adhesion assembly (2-24 h, data not shown).

The time-dependent effects of CNF1 on actin stress fiber formation and focal adhesion assembly depicted in Fig. 2 prompted us to examine whether the integrity of the actin cytoskeleton was necessary for CNF1-induced tyrosine phosphorylation. Quiescent cultures of Swiss 3T3 cells were pretreated with 1.2 µM cytochalasin D for 1 h and then stimulated with CNF1 for 6 h. As shown in Fig. 3, cytochalasin D prevented the assembly of focal adhesions (upper panel C) and completely blocked tyrosine phosphorylation of p125^fak (middle panel) and paxillin (data not shown) in response to CNF1.

Recent data from our laboratory have shown that PDGF, at a high concentration (30 ng/ml), completely abolishes bombesin-, lysophosphatidic acid-, and PMT-induced actin stress fiber formation and focal adhesions assembly. Similarly, the marked increase in focal adhesion assembly and actin stress fibers induced by CNF1 was prevented by the addition of PDGF at 30 ng/ml (Fig. 3, upper panel D, and results not shown). Since CNF1 stimulated p125^fak tyrosine phosphorylation by a mechanism dependent on the integrity of the actin cytoskeleton, we examined whether CNF1-stimulated tyrosine phosphorylation of p125^fak could be affected by high concentrations of PDGF. As shown in Fig. 3 (middle panel), PDGF (30 ng/ml) markedly reduced CNF1-stimulated tyrosine phosphorylation of p125^fak.

The rho gene product, p21^rho, has been implicated in the assembly of focal adhesions and in tyrosine phosphorylation of p125^fak and paxillin (9, 25, 29, 30). It has been suggested that CNF1 directly targets and activates p21^rho (36, 52). To investigate the role of p21^rho in the CNF1-stimulated tyrosine phosphorylation of focal adhesion-associated proteins we utilized the C. botulinum C3 exoenzyme which ADP-ribosylates Asn⁴¹ of p21^rho and thereby prevents its function (53-55). Recombinant C3 exoenzyme and pure guinea pig IgG were comicroinjected into confluent and quiescent Swiss 3T3 cells, and the cultures were further treated with CNF1 for 4 h. Cells were then fixed, permeabilized, and stained for tyrosine-phosphorylated proteins which are predominantly localized at the focal contacts in CNF1-treated cells (Fig. 3, lower panel B). The tyrosine phosphorylation of focal adhesion proteins in response to CNF1 was profoundly inhibited in cells microinjected with C3 exoenzyme (Fig. 3, lower panel D). Microinjection itself did not interfere with CNF1-induced tyrosine phosphorylation of focal adhesion proteins (Fig. 3, lower panels A and B).

PKC activation is a potential signal transduction pathway leading to increased tyrosine phosphorylation of p125^fak and paxillin (8). As the effect of CNF1 on PKC activation was unknown, we tested whether CNF1 induces phosphorylation of 80K/MARCKS, a major PKC substrate in Swiss 3T3 cells. Cells labeled with [³²P]P_i were treated with CNF1 for 6 h and lysed. The lysates were incubated with an anti-80K/MARCKS rabbit polyclonal antibody and the immunoprecipitates were analyzed by SDS-PAGE. As shown in Fig. 4A, CNF1 did not stimulate 80K/MARCKS phosphorylation. In contrast, addition of a bacterial lysate containing PMT to parallel cultures induced 80K/MARCKS phosphorylation, in agreement with previous results (56).

To substantiate that CNF1 does not activate a phospholipase C-dependent pathway that leads to the production of the intracellular second messengers diacylglycerol and Ins(1,4,5)P₃, we determined the effect of CNF1 on the production of total inositol phosphates. Quiescent cultures of Swiss 3T3 cells labeled with [2-³H]inositol were incubated with the control bacterial lysate (pGEM3) or with lysate containing CNF1 for 6 h. Fig. 4B shows that CNF1 did not stimulate a detectable increase in total inositol phosphate production. In contrast, PMT potently induced the formation of inositol phosphates in parallel cultures of Swiss 3T3 cells.

Prolonged incubation of cells with PMT depletes Ca²⁺ from intracellular stores and thereby prevents Ca²⁺ release by subsequent addition of bombesin (57). To examine whether CNF1 reduces the Ins(1,4,5)P₃-sensitive Ca²⁺ pool in Swiss 3T3 cells, quiescent cultures of these cells were treated with CNF1 or PMT for 4 h, loaded with fura-2-tetraacetoxymethyl ester, and stimulated with 10 nM bombesin. As shown in Fig. 4C, treatment with CNF1 did not reduce the Ca²⁺ mobilization induced by subsequent addition of bombesin. In contrast, PMT in parallel cultures caused a dramatic reduction of bombesin-mediated Ca²⁺ mobilization. We conclude that CNF1 stimulates tyrosine phosphorylation through a phospholipase C- and PKC-independent pathway.

Several mitogens stimulate rapid and transient activation of p42^mapk and p44^mapk through various signal transduction pathways including Ca²⁺ mobilization and PKC activation (58, 59). A significant property of MAPKs activation is the requirement for both Thr and Tyr phosphorylation within its protein kinase subdomain VIII, resulting in a mobility shift in SDS-PAGE gels. To determine if CNF1 induces p42^mapk and p44^mapk activation, lysates of Swiss 3T3 stimulated for various times with control bacterial lysate (pGEM3), CNF1, or PMT were analyzed by Western blotting with anti-p42^mapk/p44^mapk antibody. Only PMT induced p42^mapk and p44^mapk activation as judged by the appearance of slower migrating forms of p42^mapk and p44^mapk (Fig. 4D). We verified that CNF1 failed to stimulate p42^mapk and p44^mapk activation after various times of incubation (2-6 h).

The striking differences between the early events induced by CNF1 and PMT in Swiss 3T3 cells prompted us to examine the effect of CNF1 on DNA synthesis in these cells. Quiescent cultures of Swiss 3T3 cells were transferred to serum-free medium containing various concentrations of the control bacterial lysate (pGEM3) or CNF1. Cumulative [³H]thymidine incorporation was measured after 40 h. As shown in Fig. 5A, CNF1 induced DNA synthesis in a dose-dependent manner whereas the control bacterial lysate (pGEM3) did not induce any measurable increase in [³H]thymidine incorporation in parallel cell cultures. Insulin potentiated the mitogenic effect of CNF1. The maximum levels of DNA synthesis induced by the combination of CNF1 and insulin were similar to those stimulated by medium containing 10% (v/v) fetal bovine serum. Methylamine (10 mM) profoundly inhibited the stimulation of DNA synthesis induced by CNF1 and insulin. Fluorescence-activated cell sorter analysis confirmed that CNF1 stimulated DNA synthesis and induced multinucleation in Swiss 3T3 cells, as described in other cell types (data not shown).

Recently, DNT from B. bronchiseptica has been show to directly target and activate p21^rho (37). As shown in Fig. 5B, DNT also promotes DNA synthesis in Swiss 3T3 cells. Consequently, we examined whether DNT stimulates a pattern of early events comparable to that induced by CNF1. As shown in Fig. 6, treatment of Swiss 3T3 cells with a bacterial lysate containing DNT stimulated tyrosine phosphorylation of p125^fak and paxillin (upper panel) with similar kinetics to stress fiber formation (data not shown) and focal adhesion assembly (lower panel). The observed lag period suggests that internalization and processing of the toxin in the endosomal/lysosomal compartment is required. Indeed, addition of 10 mM methylamine to the culture medium or reduction of the temperature to 22 °C inhibited DNT-stimulated tyrosine phosphorylation (data not shown).

Next, we tested whether the integrity of the cytoskeleton was necessary for DNT-stimulated tyrosine phosphorylation. As shown in Fig. 7, cytochalasin D or PDGF blocked DNT-stimulated stress fiber formation (data not shown), focal adhesion assembly (upper panel), and tyrosine phosphorylation of p125^fak (middle panel). In addition, microinjection of C3 exoenzyme profoundly inhibited tyrosine phosphorylation of focal adhesion proteins in response to DNT (lower panel).

CNF1 stimulates p125^fak and paxillin tyrosine phosphorylation and stimulate DNA synthesis without promoting phospholipase C-mediated hydrolysis of PtdIns(4,5)P₂ or PKC activation. As shown in Fig. 8, DNT failed to stimulate an increase in the phosphorylation of 80K/MARCKS or to promote a detectable increase in total inositol phosphate production. In addition, DNT, like CNF1, did not induce p42^mapk activation. All these responses were induced by PMT in parallel cultures of Swiss 3T3 cells.

DISCUSSION

The results presented here show for the first time that CNF1 and DNT, both intracellularly acting bacterial toxins, induce tyrosine phosphorylation of multiple substrates in Swiss 3T3 cells. In this study, we identified two substrates, p125^fak and paxillin, which were tyrosine phosphorylated in response to CNF1 and DNT. A coordinate increase in tyrosine phosphorylation of p125^fak and paxillin is induced by a variety of molecules that regulate cell growth and differentiation (6-9, 11-20, 22). Our results suggest that p125^fak and paxillin tyrosine phosphorylation could also play a role in the signaling pathways stimulated by CNF1 and DNT.

Recent findings have raised the possibility that tyrosine phosphorylation of p125^fak and paxillin, actin stress fiber formation, focal adhesion assembly, and p21^rho function may lie in a novel signal transduction pathway (5, 22, 28, 30) that regulates cell motility and cell proliferation (24, 60). Here we used CNF1 and DNT to examine further the connections predicted by this pathway. We found that CNF1 and DNT stimulated tyrosine phosphorylation of p125^fak and paxillin, actin stress fiber formation, and focal adhesion assembly with similar kinetics. Furthermore, pretreatment of quiescent Swiss 3T3 cells with cytochalasin D, at concentrations that completely disrupted the actin cytoskeleton and the focal adhesion plaques, abolished tyrosine phosphorylation of p125^fak stimulated by both toxins. These results indicate that integrity of the cytoskeleton is essential for CNF1- and DNT-induced tyrosine phosphorylation. This conclusion was substantiated by experiments using PDGF at a high concentration (30 ng/ml), which disrupted actin stress fibers and focal adhesion assembly in response to CNF1 and DNT. At this concentration PDGF also profoundly decreased CNF1- and DNT-induced tyrosine phosphorylation of p125^fak, revealing a novel cross-talk between these toxins and PDGF. Microinjection of C. botulinum C3 exoenzyme, which ADP-ribosylates and inactivates p21^rho function, prevented tyrosine phosphorylation of focal adhesion proteins in response to either CNF1 or DNT. Thus, our results with CNF1 and DNT provide a novel line of evidence for the existence of a pathway in which p21^rho is upstream of cytoskeletal reorganization and tyrosine phosphorylation of p125 and paxillin.

Most extracellular stimuli that promote actin reorganization and tyrosine phosphorylation of p125^fak and paxillin also stimulate phospholipase C-mediated hydrolysis of PtdIns(4,5)P₂ leading to Ins(1,4,5)P₃-mediated Ca²⁺ mobilization and to the activation of diacylglycerol-dependent isoforms of PKC. Previous studies indicated that neither Ca²⁺ mobilization nor PKC activation mediate tyrosine phosphorylation of p125^fak and paxillin in response to bombesin, lysophosphatidic acid, or PMT (6, 13, 22). However, activated phospholipase C not only generates Ins(1,4,5)P₃ and diacylglycerol but also decreases PtdIns(4,5)P₂ in the plasma membrane. PtdIns(4,5)P₂ binds to several actin-binding proteins and vinculin (61-64) and thereby could modulate the organization of the actin cytoskeleton (65). In addition, p21^rho activates phosphatidylinositol-4-phosphate-5-OH kinase and elevates PtdIns(4,5)P₂ synthesis (66). These results raise the possibility that PtdIns(4,5)P₂ turnover may play a role in promoting cytoskeletal reorganization and thus, tyrosine phosphorylation of p125^fak and paxillin.

Our results demonstrate that CNF1 and DNT failed to stimulate the formation of inositol phosphates in Swiss 3T3 cells, suggesting that these toxins do not stimulate phospholipase C-mediated PtdIns(4,5)P₂ hydrolysis. This conclusion was substantiated by the inability of CNF1 and DNT to deplete the Ins(1,4,5)P₃-sensitive pool of Ca²⁺ and by the failure of these toxins to stimulate the phosphorylation of 80K/MARCKS, a prominent substrate of diacylglycerol-activated PKC isoforms expressed in Swiss 3T3 cells. In contrast, all these responses were induced by PMT in parallel cultures. Thus, our results demonstrate that p21^rho-dependent actin cytoskeleton reorganization, focal adhesion assembly, and tyrosine phosphorylation of p125^fak and paxillin can by dissociated from hydrolysis of PtdIns(4,5)P₂ in CNF1- and DNT-treated Swiss 3T3 cells.

The findings presented here have another interesting implication. A wide range of extracellular signals including growth factors and mitogenic neuropeptides activate one or more members of the family of the highly conserved serine/threonine mitogen-activated protein (MAP) kinases (ERKs) (58, 59, 67). MAP kinases, important intermediates in signal transduction pathways leading to mitogenesis or differentiation, are activated predominantly via PKC in Swiss 3T3 cells stimulated by bombesin. Once activated p42^mapk (ERK2) and p44^mapk (ERK1) phosphorylate an array of cellular proteins including protein kinases, transcription factors, and proteins involved in the regulation of cell growth (68-71). Because CNF1 and DNT stimulate DNA synthesis in Swiss 3T3 cells, we investigated the ability of these toxins to induce activation of p42^mapk and p44^mapk. Our results demonstrated that unlike PMT, CNF1 and DNT did not stimulate p42^mapk in these cells. These results suggest that CNF1 and DNT can signal entry into S phase without activating p42^mapk and p44^mapk in Swiss 3T3 cells.

The signaling events triggered by CNF and DNT appear to be very similar which suggest that both toxins trigger the same initial event, and indeed there is evidence that CNF and DNT each directly target and modify Rho (36, 37). CNF and DNT are homologous over a region of 100 amino acids near the C terminus (72) which in DNT but not in CNF contains a putative P-loop nucleotide-binding site. We have identified a lysine residue within this motif that is essential for the action of DNT, which suggests that the motif, and by inference this region, is involved in the catalytic function of DNT (51). Thus both CNF and DNT activate Rho but possibly through a different mechanism.

In conclusion, our results with CNF1 and DNT provide further evidence for a p21^rho-dependent pathway leading to tyrosine phosphorylation of p125^fak and paxillin. In particular, actin stress fiber formation, focal adhesion assembly, and tyrosine phosphorylation of p125^fak were dissociated from PtdIns(4,5)P₂ hydrolysis in CNF1- and DNT-treated cells. Furthermore, CNF1 and DNT induce DNA synthesis in the absence of detectable activation of p42^mapk and p44^mapk providing additional evidence for a novel p21^rho-dependent signaling pathway that leads to entry into the S phase of the cell cycle.