ADP-ribosylation of the G protein Rho inhibits integrin regulation of tumor cell growth.

Using a gastric derived tumor line, we investigated the involvement of β1 integrin and Rho in cell growth regulation in response to collagen. The addition of C3 exoenzyme from Clostridium botulinum to specifically ribosylate and inhibit the function of the rho gene products inhibited cellular proliferation in a dose-dependent fashion. C3 exoenzyme exhibited broad cytostatic activity toward a number of tumor lines and induced G/G accumulation, cyclin A inhibition, and pronounced alterations in cell morphology. Integrin-mediated adhesion to collagen led to the expression of the cyclin A gene whose expression could be blocked using anti-β1 integrin monoclonal antibodies. Phospholipid levels were induced upon β1 integrin-mediated adhesion to collagen, and the phospholipid induction was inhibited by either antibodies to β1 integrin or pretreatment of cells with C3 exoenzyme. Significant reduction in phospholipid levels correlated with proliferation for a panel of tumor lines deprived of adhesion to substrate. These results implicate a novel role for integrins and Rho in the regulation of tumor growth in response to matrix.

The integrins are a family of adhesion receptors that play a role in the interaction of cells with the extracellular matrix (1,2). Cellular interactions with the extracellular matrix play a role in diverse processes such as differentiation (3,4), lymphocyte activation (5), and tumor cell dissemination and metastasis (6). Collagen is a major component of the extracellular matrix that serves as a scaffold for cell binding but also plays a role in cell differentiation and growth (7).
In vitro, interaction of cells with a collagen matrix can induce arachidonic acid production in HeLa cells (8) and glandular differentiation in a human colon carcinoma line (9). Clustering of the ␣2␤1 integrin, a receptor for collagen, induces stimulation of tyrosine phosphorylation and accumulation of GTP bound Ras in a human lymphoblastic cell line (10). The interaction of cells with the extracellular matrix can also regulate growth by creating a permissive effect for the action of mitogens (11,12). Therefore, specific interaction between integrins and matrix components provide another level of growth regulation. A potential mediator of integrin signaling is the small GTP binding protein Rho. It is a member of the Ras family of GTPases that regulates the formation of actin stress fibers in response to growth factors (13). Integrins can also regulate the formation of actin stress fibers (14), which suggests a conver-gence of integrin and Rho signaling pathways.
A role in mitogen signaling has been demonstrated for Rho using Val-14 mutations, analogous to Val-12 oncogenic mutations in Ras, which induced transformation of fibroblasts (15). Amplification of Rho by transfection resulted in cells that exhibited increased tumorigenicity, higher saturation density, and reduced serum dependence (16). Mutations in Dbl, a guanine exchange factor for Rho, also induced cellular transformation (17,18). Recent reports have demonstrated the induction of phospholipid levels in untransformed fibroblasts upon adhesion to fibronectin by a mechanism involving Rho. This pathway may provide a mechanism for integrating adhesion with soluble growth factors that generate phospholipid derived second messenger signals (19,20). These findings suggest a role for Rho in the regulation of both cell growth and cell architecture. A means for elucidating the function of Rho has employed the use of the ADP-ribosylation exoenzyme C3 from Clostridium botulinum. C3 could efficiently modify Rho A, B, and C by ribosylation to alter their function, presumably by inhibiting interaction with downstream effectors (21,22). The other Rho family proteins such as Rac-1 and CDC42 are poor substrates in vitro (23,24), and therefore C3 provides a sensitive means for elucidating Rho function.
In these studies, we investigated a role for integrins and Rho in growth regulation of a tumor line. Cyclin A and cyclin D are regulators of S phase progression (25,26). They have been implicated in tumorigenesis (25,27), and cyclin A and has also been shown to serve as a link between adhesion and cell cycle progression (28). Using a gastric tumor line ST2 whose proliferation was previously shown to be regulated by adhesion (29), we have demonstrated specific ␤1 integrin-dependent regulation of cyclin A expression, cell cycle progression, and phospholipid synthesis on collagen matrix using monoclonal antibodies that block these events. We also demonstrated the involvement of Rho in this integrin-mediated process by inactivating Rho in tumor cells using C3 exoenzyme. These studies address novel mechanisms relating an integrin and Rho in mediating tumor growth in response to matrix.

MATERIALS AND METHODS
Cell Lines and Reagents-ST2 and ST7 are tumor lines established from separate patients diagnosed with gastric adenocarcinoma. A375 is derived from melanoma, MG-63 from an osteosarcoma, and MCF-7 from a mammary epithelial adenocarcinoma (ATCC, Rockville, MD). HMEC-1 is a human microvascular endothelial cell line immortalized with SV40 large T antigen (30). The astrocytoma 131-INI and glioblastoma Fogerty were gifts of Drs. I. J. Fidler and Janet Price, respectively (University of Texas, M. D. Anderson Cancer Center, Houston, TX). All cell lines were maintained in complete RPMI medium supplemented with 10% FBS, 1 2 mM L-glutamine, 1 mM sodium pyruvate, and 100 * This work was supported by National Institutes of Health Grant 62596. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
C3 Exoenzyme Purification-The exoenzyme C3 gene from C. botulinum type D strain 1873 (32) cloned into the pGEX expression vector system (Pharmacia Biotech Inc.) to generate a glutathione S-transferase fusion (pGEX2T-C3) was a gift of Dr. Larry Feig (Tufts University, Boston, MA). Purification of recombinant C3 was performed as described. 2 Essentially, a single colony of Escherichia coli (strain JM109) transformed with pGEX2T-C3 DNA was inoculated into 100 ml of LB culture containing 100 g/ml ampicillin and grown overnight at 37°C on a shaker table. This culture was then added to 900 ml of LB containing ampicillin and shaken at 37°C for 1 h. Isopropyl-␤-D-thiogalactopyranoside was then added to a concentration of 100 g/ml, and the culture was grown at 37°C with shaking for an additional 7 h. The bacteria was centrifuged at 5,000 ϫ g, and the pellet lysed in 40 ml of ice-cold PBS containing 1 mg/ml lysozyme, 1% Triton X-100, 25% sucrose, 1 mM EDTA, 5 mM 2-mercaptoethanol, and 1 mM phenylmethylsulfonyl fluoride for 30 min on ice with occasional stirring. The slurry was sonicated for 3 min, and pancreatic DNase I (Boehringer Mannheim) was added to a final concentration of 100 g/ml, stirred for an additional 20 min at 4°C, and centrifuged at 10,000 ϫ g for 10 min at 4°C.
The lysate was added to 1 ml of washed glutathione-agarose beads (Pharmacia) and mixed gently at 4°C. The beads were centrifuged at 500 ϫ g for 5 min, washed five times with 14 ml of PBS containing 1% Triton X-100, and washed three times with 14 ml of 50 mM Tris-HCl (pH 7.5), 150 mM NaCl, 2.5 mM CaCl 2 , and the beads were resuspended in 0.5 ml of the Tris buffer. 30 NIH units of purified bovine ␣-thrombin was added and incubated for 16 h at 4°C to cleave the fusion protein.
The beads were centrifuged at 1000 ϫ g for 2 min, washed twice with 1 ml of the Tris solution, and concentrated using a Centricon-3 (Amicon, Beverly, MA). The purified product ran as a single band at approximately 25 kDa on a 11.25% SDS-polyacrylamide gel (33).
Immobilization of Purified Matrix Proteins and Modification of Tissue Culture Surface-Purified rat collagen type I (Sigma) and laminin from Engelbreth Holm-Swarm sarcoma cells (Sigma) were coated onto tissue culture wells (10 g/cm 2 ) as recommended by vendor. PolyHEMA (Sigma) was dissolved in 95% ethanol and air dried onto tissue culture plastic at 5 mg/cm 2 to inhibit cell attachment and spreading. Human plasma fibronectin (Sigma) was diluted in PBS to 50 g/ml and adsorbed onto tissue culture surfaces for 4 h at 37°C in a 5% CO 2 humidified incubator. All wells were blocked with 5% bovine serum albumin in PBS for 4 h at room temperature. The wells were rinsed three times with PBS before adding cells.
Image Capture-Images were taken on a Nikon DIAPHOT-TMD inverted scope using a VI-470 CCD video camera system (Optronics Engineering, Gogeta, CA) digitized by a QuickCapture frame grabber board (Data Translation Inc., Marlboro, MA) and then stored on a magneto optical disc cartridge utilizing a Third Wave OptiDisk drive (Third Wave Computing, Inc., Austin, TX).
Proliferation Assay-Cell number was determined essentially as described (34). Briefly, cells cultured in 96-well plates were fixed with 70% ethanol at Ϫ20°C for 5 min and then stained with 0.1% crystal violet dissolved in H 2 O for 5 min at room temperature. The wells were rinsed with H 2 0 three times and aspirated, and the insoluble crystals were dissolved in 2% SDS/H 2 0. Absorbance was measured at 570 nm.
Propidium Iodide Staining-To analyze cellular DNA content, cells were fixed in ice-cold 70% ethanol for 30 min and then resuspended in PBS containing 50 g/ml propidium iodide (Sigma) and 25 g/ml RNase. After 30 min at room temperature, the cells were analyzed on an EPICS Profile (Coulter, Hialeah, FL). The Multicycle program (Phoenix Flow Systems, San Diego, CA) was used to analyze the histograms.
Northern Blot Analysis-Total RNA was extracted by the method of Huang and High (35). Briefly, after transfer of cells to microcentrifuge tubes, 2 ϫ 10 6 cells were lysed gently in 0.5 ml of solution I containing 2% SDS, 200 mM Tris-HCl, pH 7.5, and 1 mM EDTA at room temperature. 150 l of ice-cold solution II containing 42.9% (w/v) potassium acetate and 11.2% (v/v) acetic acid was added, mixed, and placed on ice for 3 min to precipitate DNA and proteins. The insoluble material was pelleted at 20,000 ϫ g for 5 min at room temperature. The supernatant was extracted with phenol/chloroform (1:1) according to standard procedures (36) and precipitated with isopropanol. Approximately 10 g of total RNA was electrophoresed on formaldehyde gels, transferred to nylon membranes (DuPont NEN) by vacuum transfer (LKB, Bromma, Sweden), and hybridized according to standard procedures (36) with 50 ng of cDNA labeled with [␣-32 P]ATP using a random prime labeling kit (Boehringer Mannheim).
Phospholipid Analysis-Cells (2 ϫ 10 5 ) were washed three times in phosphate free RPMI (ICN) containing 5% bovine serum albumin supplemented with 2 mM L-glutamine, 1 mM sodium pyruvate and then incubated in the same medium containing 100 Ci/ml [ 32 P]orthophosphoric acid (DuPont NEN) for 3 h at 37°C in a 5% CO 2 humidified incubator. The samples were cooled by placing the tissue culture plates on ice, and adherent cells were treated briefly for 2 min with 0.5% (w/v) trypsin and 0.1% EDTA diluted in PBS. After pelleting in microcentrifuge tubes, the cells were lysed in 180 l of methanol/1 N HCl (1:1). Chloroform (150 l) was added and vortexed vigorously for 10 s. The samples were centrifuged at 20,000 ϫ g for 5 min at room temperature to separate phases, and the bottom organic layer was removed. 10 l of this sample was spotted onto a flexible 1B-F silica gel plates (J. T. Baker, Phillipsburg, NJ) pretreated with 1% oxalic acid and developed in buffer composed of chloroform/methanol/29.5% ammonium hydroxide/water (9:7:0.3:1.7). Radioactive spots were visualized by autoradiography, and phosphatidylinositol phosphate and phosphatidylinositol bisphosphate were identified by comparison with phospholipid standards.

C3 Exoenzyme Inhibits Cell Proliferation-Given a role for
Rho in the regulation of cell morphology and growth, we initially investigated the effect of inhibiting Rho by culturing the cells in the presence of C3 exoenzyme from C. botulinum, which was previously demonstrated to specifically inhibit Rho by ADP-ribosylation (21,22). In Fig. 1, increasing amounts of C3 exoenzyme were added to cultures of cells plated on substrate. After 5 days in culture, cell number was determined, and relative cell number was expressed as a ratio of the number of cells cultured with C3 versus the number of cells cultured without C3. Cell proliferation was inhibited dose-dependently, with maximal inhibition at approximately 30 -50 g/ml (Fig.  1A). This demonstrated that C3 was taken up by the cells to ribosylate and inactivate Rho in situ. Using 50 g/ml, C3 not only inhibited ST2 but also exhibited cytostatic activity for a number of other cell lines as well (Fig. 1B).
C3 Catalyzed Ribosylation Induces Alterations in Cell Morphology-Cells cultured in the presence of 50 g/ml of C3 exoenzyme for 24 h exhibited pronounced alteration in morphology, showing slender processes and numerous micro-spikes relative to nontreated cells (Fig. 2). Using the same isolation procedure as for C3, a control preparation from untransformed bacteria had no effect on either morphology or proliferation, demonstrating that the effects were not due to contaminating bacterial components (not shown). These results further suggested that Rho was ribosylated by the C3 taken up by the cells from the culture medium. To demonstrate this, whole cell lysates prepared from cells cultured in the absence or the presence of 50 g/ml C3 were subjected to in vitro C3 catalyzed ribosylation using [ 32 P]NAD as a ribose donor followed by SDS electrophoresis. In Fig. 3, the resulting autoradiogram revealed that C3 specifically labeled a species that migrated with an apparent molecular mass of approximately 25 kDa on a 11.25% polyacrylamide gel; this is consistent with previously published reports of the specificity for C3 catalyzed ribosylation of Rho (23,24,37). Lysates from cells cultured in the presence of C3 (lanes 2, 4, 6, 8, 10, and 12) showed a greatly reduced level of 32 P-labeled Rho indicating prior endogenous ADP-ribosylation. C3 Induces G 0 /G 1 Accumulation and Inhibits Integrin-dependent Expression of Cyclin A-Because C3 treatment inhibited cell proliferation, we investigated its effect on cell cycle distribution. In Fig. 4, propidium iodide staining followed by flow cytometric analysis revealed accumulation of cells in G 0 /G 1 and a reduction in the number of cells in S phase after C3 treatment. In the absence of C3 treatment, 45.3% of the cells were in S phase, whereas 36.3% were in G 0 /G 1 ; in the presence of C3 exoenzyme, 33.0% of the cells were in S phase, whereas 51% were in G 0 /G 1 . G 0 /G 1 accumulation was also observed for other lines treated with C3 and is summarized in Table I. These results indicate that the inhibitory effect of C3 on proliferation (Fig. 1B) occurs at a specific point in the cell cycle to inhibit G 0 /G 1 to S phase transition.
We next investigated the effect of adhesion and C3 on cell cycle progression at the transcriptional level by analyzing the expression of cyclin A and cyclin D, two regulators of cell cycle progression that have been implicated in tumorigenesis (25,38). We first investigated the effect of integrin-mediated adhesion to collagen on the expression of cyclin A message. ST2 cells were initially deprived of adhesion for 30 h by transferring the cells to culture dishes coated with polyHEMA, a substance used to inhibit cell attachment (11) (Fig. 5A). By culturing the cells in the absence of adhesion (11,39), the effect of adhesion to specific substrates can then be analyzed. Upon transfer from polyHEMA to collagen-coated dishes, the cells expressed low levels of cyclin A (Fig. 5A, lane 1) 4 and 5). The results demonstrated that adhesion to collagen could induce the expression of cyclin A message, and its expression was dependent on ␤1 integrin-mediated adhesion.
The expression of cyclin A, which was regulated by ␤1 integrin-mediated adhesion, is also known to be regulated by growth factors (serum) (40). Also, because C3 induced G 0 /G 1 accumulation (Fig. 4 and Table I), we compared the effect of C3 treatment, serum withdrawal, and deprivation of adhesion on the expression of cyclins A and D1 (26,38). In Fig. 6A 4). The decrease in cyclin A expression in response to deprivation of adhesion, serum starvation, and Rho inactivation by C3 treatment is consistent with the effect of integrins, Rho, and mitogenic factors to regulate proliferation at similar points in the cell cycle (19,20,28,38). Although C3 inactivation of Rho is also known to induce cell rounding and detachment from substrate (41), which could potentially inhibit cyclin A expression, in this system, C3 induced morphological alterations but did not inhibit cell attachment. These results demonstrate a distinct but cooperative role for integrins, Rho, and serum factors leading to cell cycle progression.
In a similar experiment, the blots were probed with cyclin D1 (Fig. 6B), and the expected alternatively polyadenylated 4.8 and 1.7 cyclin D1 message was detected (26). In contrast to cyclin A, cyclin D1 expression was not affected by any of the treatments, which demonstrates that not all cyclin mRNA expression was affected by these different modes of treatment. Also, the lack of inhibitory effect of reduced serum on cyclin D1 was of interest because growth factors are known to regulate the expression of cyclin D1 (26). However, disregulated expres-

TABLE I
Cell cycle distribution after C3 treatment 250,000 cells in 2 ml of complete RPMI were plated in tissue culture wells either in the absence or the presence of 50 g/ml C3 exoenzyme. After 3 days in culture, the cells were briefly trypsinized, fixed in 70% ethanol, and stained with 50 g/ml PI in PBS containing 20 g/ml RNAse. After incubating for 30 min at room temperature, DNA content was determined on an EPICS Profile, and the resultant histograms analyzed using the Multicycle program (Phoenix Flow Systems, San Diego, CA) to estimate the percentage of cells in the indicated phases of the cell cycle. Chi square ( 2 ) values for the histograms generated by the Multicycle program is also shown. The average increase in the percentage of cells in G0/G1 24 h after treatment with 50 g/ml C3 treatment for seven separate experiments was 15  sion of cyclin D, by relieving requirements for G 0 to G 1 cell cycle transition, has been proposed as a mechanism contributing to tumorigenesis (38,42). These results indicate that tumor proliferation may also require integrin signaling and Rho to achieve progression into S phase.
␤1 Integrin-dependent Phospholipid Induction Is Inhibited by C3-The contribution of adhesion to endogenous phospholipid synthesis was determined for ST2 by measuring 32 P incorporation into phospholipids. In Fig. 7, ST2 cells were incubated on polyHEMA for 24 h either in the presence or the absence of C3 exoenzyme and then maintained on polyHEMA or replated on collagen-coated dishes in the presence or the absence of anti-␤1 integrin monoclonal antibodies. Phospholipid levels were then determined by 32 P incorporation, and the labeled products analyzed by thin layer chromatography. Significant induction of phospholipids including PIP and PIP 2 were seen when cells were transferred from polyHEMA (lane 1) to collagen-coated dishes and incubated (lane 2) for 3 h. Anti-␤1 integrin monoclonal antibodies could block the induction of phospholipids when plated on collagen (lane 4). The induction of phospholipids could also be inhibited if the cells were preincubated in the presence of 50 g/ml C3 exoenzyme for 24 h on polyHEMA prior to plating cells on collagen (lane 6). C3 treatment alone also slightly but consistently inhibited phospholipid levels (compare lanes 1 and 5). This suggests that a separate pathway mediated by integrin as well as Rho can lead to phospholipid induction. Both integrin-mediated adhesion as well as a functional Rho, however, were required for efficient induction of phospholipids.
Decreased Phospholipid Levels Correlate with Reduced Proliferation of Tumor Cells in the Absence of Adhesion-In Fig. 8, a panel of tumor cells were either maintained on substrate or transferred to polyHEMA-coated dishes for 24 h, and their phospholipid synthesis levels were measured using 32 P incorporation. The level of phospholipid synthesis including PIP and PIP 2 was reduced markedly in the absence of adhesion for a panel of tumor lines (Fig. 8A). We also measured the proliferation of these cells after 5 days and found that the cells plated on polyHEMA, which showed reduced phospholipid including PIP and PIP 2 levels on polyHEMA, correlated with reduced proliferation relative to cells maintained on substrate (Fig. 8B).

DISCUSSIONS
Numerous examples have demonstrated integrin signaling, including clustering of the fibronectin receptor leading to protein tyrosine phosphorylation of focal adhesion tyrosine kinase (43) and cytoplasmic alkalinization (44). In peripheral blood T cells, antibodies to integrins or matrix components have been shown to possess growth stimulatory activities (5,45).
In these studies we have shown that cyclin A and phospholipids were regulated by adhesion to collagen, and the induction was inhibited by an anti-␤1 integrin monoclonal antibody. ADP-ribosylation of Rho by C3 exoenzyme inhibited proliferation, induced G 0 /G 1 accumulation, and also inhibited integrinmediated induction of phospholipid synthesis as well as cell cycle progression as measured by the expression of cyclin A. The ability of ADP-ribosylation of Rho to inhibit integrin-mediated cell cycle progression demonstrates a convergence of integrin and Rho signaling pathways leading to positive regu- lation of tumor cell growth as a consequence of cell attachment to substratum. Cyclin A expression was also inhibited by culturing the cells under low serum (0.5%) conditions (Fig. 6A), and this correlated with G 0 /G 1 cell cycle arrest (not shown). These results demonstrate an additional level of cooperativity between integrins, Rho, and soluble serum factors leading to cell cycle progression. The generation of second messengers leading to proliferation induced by many growth factors relies on a pool of phospholipids (46) and therefore, the synthesis of phospholipids induced by integrins in tumor cells may play a pivotal role in the integration of growth factors and adhesion leading to growth stimulation.
The profound morphological alterations accompanying ADPribosylation of Rho is consistent with alterations in the level of phospholipids. Phospholipid can alter the function of cytoskeletal components including ␣-actinin (47), gelsolin (48), profilactin (49), and talin (50), and therefore, changes in morphology induced by C3 ribosylation of Rho may in part be explained by alterations in phospholipid levels induced by C3 treatment. Cells maintained in suspension overnight show reduced phospholipid synthesis and correspondingly showed similar alterations in morphology, which was evident upon reattachment to collagen coated plastic. This alteration was transient, and the cells resumed normal morphology after several hours on collagen (not shown).
These results are in agreement with those of other investigators who demonstrated that enhanced inositol lipid synthesis and platelet-derived growth factor-induced inositol lipid synthesis breakdown in response to fibronectin involved the regulation of phosphatidylinositol 4-phosphate 5-kinase by Rho (19,20). Our studies using intact cells demonstrated that integrinmediated adhesion could regulate not only PIP 2 but PIP as well. Treatment of cells with C3 also inhibited the synthesis of PIP 2 and PIP. The observed differences may be due to more complex molecular interactions in intact cells in which, for example integrin-mediated induction of phospholipids may influence downstream activation of protein kinase C, which in turn, may indirectly alter phospholipid metabolism.
Various studies have implicated a role for integrins in cancer. The mechanism by which adhesion molecules promote metastasis is not clear but may involve adhesion to platelets to escape immunologic detection (51). Integrins can mediate extravasation of cells from circulation into tissue to sequester them from immunologic detection or physical damage caused by shear forces under conditions of flow (6). An additional role for integrins in promoting metastasis may involve generation of growth promoting signals. Anchorage-independent growth is a function of the suppression of the apoptotic pathway (52) and or the activation of growth promoting pathways. The growthpromoting effect on tumors through integrins may therefore occur as a separate event in the context of mutations leading to suppression of apoptosis.
A number of tumors have been shown to overexpress growth factor receptors. Karyotype analysis of ST2 has revealed the amplification of chromosome 7p, which carries the gene for the epidermal growth factor receptor (29). Therefore, the epidermal growth factor receptor may play a relevant role in contributing to tumor proliferation of ST2, and its overexpression may facilitate rapid tumor growth under conditions of limiting soluble factors (53,54). The binding of growth factors can lead to phospholipid turnover and the generation of second messengers, which ultimately act on components that regulate cell cycle progression and cell division. Several components that respond to growth factor stimulation and that have also been implicated to play a role in tumorigenesis include cyclin A and cyclin D. Results using ST2 demonstrated that cyclin D expres-sion is disregulated and may contribute to its tumorigenicity. Cyclin D is required for exit from G 0 into G 1 , and therefore, constitutive expression of cyclin D may result in the inability of ST2 to remain in G 0 such that the cells will, in the absence of physiologic stimulation, progress to G 1 . Full progression into S phase not only requires cyclin D but also the expression of cyclin A. In contrast to cyclin D1, however, cyclin A expression in ST2 cells required not only soluble factors (serum) but also integrin-mediated adhesion. The ability of cyclin A to mediate adhesive signals has been shown in untransformed fibroblast lines (28), and therefore, integrin-mediated adhesion contributing to tumor proliferation may be a feature that may be retained upon cell transformation in some cases.