Opposing Roles of Ras/Raf Oncogenes and the MEK1/ERK Signaling Module in Regulation of Expression and Adhesive Function of Surface Transglutaminase*

Tissue transglutaminase (tTG) serves as a potent and ubiquitous integrin-associated adhesion co-receptor for fibronectin on the cell surface and affects several key integrin functions. Here we report that in fibroblasts, activated H-Ras and Raf-1 oncogenes decrease biosynthesis, association with (cid:1) 1 integrins, and surface expres- sion of tTG because of down-regulation of tTG mRNA. In turn, the reduction of surface tTG inhibits adhesion of H-Ras- and Raf-1-transformed cells on fibronectin and, in particular, on its tTG-binding fragment I(6)II(1,2)I(7– 9), which does not interact directly with integrins. Analysis of Ras/Raf downstream signaling with specific pharmacological inhibitors reveals that the decrease in tTG expression is mediated by the p38 MAPK, c-Jun NH 2 - terminal kinase, and phosphatidylinositol 3-kinase pathways. In contrast, increased activation of the ERK pathway by constitutively active MEK1 stimulates tTG mRNA expression, biosynthesis, and surface expression of tTG, whereas MEK inhibitors or dominant negative MEK1 exert an opposite effect. This modulation of surface tTG by ERK signaling alters adhesion of cells on fibronectin and its fragment that binds tTG. Further-more, transient stimulation of ERK signaling in untransformed fibroblasts by adhesion

Transglutaminases are a multigene family of Ca 2ϩ -dependent enzymes that mediate covalent cross-linking of proteins by forming amide bonds between glutamines and ⑀-amino groups of lysines and mechanically stabilize tissues by formation of protein multimers resistant to enzymatic and physical degradation (1,2). Whereas most transglutaminases are restricted to one or several cell types, tissue transglutaminase (tTG) 1 is widely expressed and is particularly abundant in endothelium and smooth muscle cells (3). It localizes primarily in the cytoplasm, with a small part of its intracellular pool present in the nucleus (4). Because the cross-linking activity of tTG is induced by Ca 2ϩ and inhibited by GTP, it is mostly dormant inside the cell, but can be drastically enhanced following a rise in cytoplasmic Ca 2ϩ (5). Several intracellular targets of cross-linking by cytoplasmic tTG were identified, although physiological significance of these events has not been elucidated (6). tTG also binds and hydrolyzes GTP and mediates intracellular signaling from the ␣1B adrenergic receptor (7) to downstream targets such as phospholipase C␦ (8). Finally, tTG inhibits adenylate cyclase activity and this effect does not require its cross-linking function (9). Therefore, cytoplasmic tTG has some direct signaling functions apart from its enzymatic activity.
Importantly, some amounts of tTG are present on the cell surface (10) and in the ECM (11) where it can interact with several proteins including fibronectin (12). Unlike cytoplasmic tTG, the cell surface enzyme is constitutively active because of high Ca 2ϩ concentrations in the extracellular space and is able to cross-link fibronectin (10,13) and several other ECM proteins. Because there is no leader sequence in tTG (14), it remains essentially unknown how the protein is exported to the cell surface.
Several previous studies suggested a role of surface tTG in cell adhesion (14 -17). We recently expanded these observations by showing that tTG binds directly to the extracellular domain of ␤ 1 and ␤ 3 integrin subunits and forms ternary complexes with integrins and fibronectin on the cell surface (18). Surface tTG was found to affect a wide range of integrin functions. It stimulates adhesion, spreading, and migration of several cell types on fibronectin (18 -20), amplifies integrin-dependent activation of FAK (18), and promotes integrindependent fibronectin assembly into deoxycholate-insoluble matrix (21). This newly defined adhesive function of surface tTG is independent of its cross-linking activity (18,21,22).
tTG is a multifunctional protein involved in diverse phys-iological responses (1,16) and, therefore its expression is regulated by multiple factors. Several regulatory sites were located within the promoter region of the tTG gene (23). Retinoids act as acute stimulators of biosynthesis and crosslinking activity of tTG (24,25). Activation of the tTG gene promoter by retinoids was found to depend on the proximal element and a 30-base pair tripartite retinoid response element located 1.7 kilobases upstream of the transcription start site (26). Several inflammatory cytokines including interleukin-1␤, interleukin-6, and tumor necrosis factor-␣ trigger a sharp increase in biosynthesis of tTG during tissue injury and inflammation (27)(28)(29) by acting via NF-B and the corresponding regulatory site in the promoter (30). A large body of work demonstrated that both expression and enzymatic activity of tTG are greatly enhanced by transforming growth factor ␤ in several cell types (31). Accordingly, transforming growth factor ␤ 1 response element 5Ј-GAGTTG-GTGC-3Ј was mapped 868 base pairs upstream of the transcription start site in the promoter (32). However, very little is known about the role of these and other pathways in modulation of adhesive function of cell surface tTG. The purpose of the present study is to analyze regulatory mechanisms that control expression and localization of tTG on the surface of untransformed and oncogene-transformed cells.
In this report we show that H-Ras, Raf-1 oncogenes, and the ERK signaling cascade have opposing effects on tTG mRNA expression and, consequently, on biosynthesis of tTG, formation of integrin-tTG complexes, and surface levels of tTG. Importantly, tTG is known to exert tumor suppressor function by inhibiting cell proliferation (33)(34)(35) and its expression is frequently decreased in human tumors (36 -38). Therefore, inhibition of biosynthesis and surface expression of tTG by activated forms of H-Ras and Raf-1 oncogenes might explain the reduction of tTG in primary tumors and contribute to adhesive deficiency and anchorage-independent growth of transformed cells. Moreover, we found that down-regulation of tTG is likely a common feature of Ras and Raf family members, because activated forms of K-Ras and B-Raf oncogenes, often detected in human cancers (39), also inhibit tTG expression. At the same time, analysis of Ras/Raf downstream signaling pathways showed that constitutive activation or transient stimulation of the MEK1/ERK signaling module by adhesion or growth factors increases tTG biosynthesis and expression on the surface of untransformed cells. Finally, ERK activation is required for growth factor-dependent integrin clustering and formation of cell-matrix contacts containing ␤ 1 integrins and tTG. Opposite regulation of expression and surface localization of tTG by Ras/Raf oncogenes and MEK1/ERK signaling identified in this study is in agreement with the adhesive function of this protein, which is independent of its cross-linking activity and promotes interactions of cells with the surrounding matrix (1,2,18,21,22).

EXPERIMENTAL PROCEDURES
Reagents, cDNAs, and Antibodies-Tran 35 S-label TM (a mixture of [ 35 S]methionine and cysteine, specific activity 1175 Ci/mmol) and methionine-and cysteine-free medium were from ICN Pharmaceuticals (Irvine, CA). MEK inhibitor U0126 and p38 MAPK inhibitor SB203580 were obtained from Promega (Madison, WI). PI3K inhibitor LY294002 and JNK II inhibitor SP125600 were from Calbiochem (San Diego, CA). Recombinant acidic human fibroblast growth factor-1, epidermal growth factor, and platelet-derived growth factor-BB were from R&D Systems (Minneapolis, MN). Protein A-and Protein G-conjugated to Sepharose 6B TM were obtained from Invitrogen (Carlsbad, CA). Purified human plasma fibronectin and its proteolytic gelatin-binding fragment containing the modules I(6)II(1,2)I(7-9) were a gift from Dr. K. Ingham (American Red Cross). cDNAs encoding activated forms of H-Ras and Raf-1 oncogenes, H-Ras(V12) and ⌬Raf-22W, were kindly provided Dr. C. Der (University of North Carolina, Chapel Hill, NC).
To generate regulated expression of rasH(12V) or ⌬raf22W in NIH 3T3 fibroblasts, the GeneSwitch TM inducible expression system from Invitrogen was used. Cells were initially transfected with pSwitch plasmid and stable transfectants were selected with 300 g/ml hygromycin. Then, the resulting stable transfectants were transfected again with the original pGene plasmid or the same vector encoding rasH(12V) or ⌬raf22W and selected with 150 g/ml zeocin. Inducible expression of activated H-Ras or Raf-1 oncogenes in NIH 3T3 double transfectants was achieved by treatment with 10 ng/ml mifepristone for 36 h. The same method was employed for inducible expression of rasK(12V) and rafB1-CAAX oncogenes.
To express exogenous tTG in NIH 3T3 fibroblasts transformed with activated H-Ras or Raf-1 oncogenes, cells transfected with pSwitch and pGene plasmids encoding vector alone, rasH(12V), or ⌬raf22W were then transfected with tTG cDNA (14) in pcDNA3.1neo plasmid and the resulting transfectants were selected with 800 g/ml G-418 (neomycin).
Analysis of tTG mRNA Expression-For Northern blot hybridization, equal amounts of purified total cellular RNA (25 g/lane) were separated by electrophoresis on a 1% agarose gel containing 2.2 M formaldehyde and 1ϫ FA gel buffer (20 mM MOPS, 100 mM NaCH 3 COO, 10 mM EDTA, pH 7.0), transferred to Nylon membrane (Osmonics, Westborough, MA), and UV cross-linked to the membrane. The membrane was prehybridized by incubation with Church's buffer for 8 h at 65°C (45) and then hybridized with 32 P-labeled mouse tTG cDNA probe (ResGen, Carlsbad, CA) in Church's buffer overnight at 65°C. After hybridization, the membrane was consecutively washed for 15 min at 65°C in 2ϫ SSC ϩ 0.1% SDS, 1ϫ SSC ϩ 0.1% SDS, 0.2ϫ SSC ϩ 0.1% SDS, and 0.1ϫ SSC ϩ 0.1% SDS. To visualize tTG mRNA bands, the Storm 860 Phosphor Screen storage system (Amersham Biosciences) was used. ImageQuant 5.0 software was applied for quantitation of the intensity of tTG mRNA bands. The same blots were rehybridized with the control housekeeping gene ␤-actin to ensure equal loading.
Metabolic Labeling and Immunoprecipitation of tTG and ␤ 1 Integrin-tTG Complexes-All the procedures of metabolic labeling were performed in methionine-, cysteine-free medium. When present, methionine-, cysteine-deficient FBS was used during the labeling. Briefly, NIH 3T3 fibroblasts in serum-free medium were either kept in suspension or plated on fibronectin-coated dishes for 3 h. Cells in suspension and cells adhering on fibronectin were then labeled with 35 S for 1 h. Also, quiescent (serum-starved for 48 h) adherent NIH 3T3 fibroblasts were pretreated for 5 min with 10% FBS, 25 ng/ml fibroblast growth factor-1, 10 ng/ml platelet-derived growth factor-BB, or left untreated. Then, both quiescent and growth factor-stimulated cells were labeled with 35 S for 1 h, while growth factors were kept in the medium during the labeling. In another set of experiments, adherent NIH 3T3 cells in the presence of 10% FBS were pretreated for 12 h with 10 M U0126, 10 M SB203580, 20 M LY294002, or 20 M SP125600 before 35 S labeling for 1 h in the presence of these inhibitors. In all these cases, 200 Ci/ml Tran 35 S-label TM was used for metabolic labeling. Alternatively, adherent NIH 3T3 transfectants expressing MEK1 mutants, H-Ras(12V), or ⌬Raf-22W in the presence of 10% FBS were metabolically labeled with 50 Ci/ml Tran 35 S-label TM for 12 h.
Cell lysis in RIPA buffer and immunoprecipitation of metabolically labeled ␤ 1 integrins, tTG, and ␤ 1 integrin-tTG complexes were performed essentially as described (18 -21). 3 ϫ 10 8 cpm of protein-incorporated 35 S-labeled radioactivity equivalent to 1 mg of total cellular protein was used for each sample in immunoprecipitation experiments. 5 g/sample of mAb HM␤1-1 was used for immunoprecipitation of ␤ 1 integrin-tTG complexes (18). To reprecipitate tTG from the ␤ 1 integrin-tTG complexes, the resulting immune complexes were boiled in 1% SDS, and the eluate was reconstituted with 1% Triton X-100, 150 mM NaCl, 50 mM Tris-Cl, pH 7.5, to a final SDS concentration of 0.1% (18,19). Half of each sample was subjected to reprecipitation with 3 l of polyclonal antibody to ␤ 1A integrin cytoplasmic domain (Chemicon) and another half with 5 g of polyclonal anti-tTG antibody (18 -20). The resulting 35 S-labeled ␤ 1 integrin immune complexes and ␤ 1 integrin and tTG immunoprecipitates were analyzed by SDS-PAGE in 10% acrylamide, 0.25% bisacrylamide gels. After electrophoresis the gels were fixed and treated with Autofluor TM (Amersham Biosciences) for fluorography. Autoradiograph images were generated using Bio-Max MR single emulsion film (Eastman Kodak, Rochester, NY).
Measurement of Cell Surface tTG by Flow Cytometry-For flow cytometry, NIH 3T3 fibroblasts that were kept in suspension in the absence of serum for 4 h were initially compared with serum-starved adherent cells that were detached from fibronectin-coated dishes with EDTA immediately before the experiment. Likewise, NIH 3T3 transfectants, quiescent untransformed NIH 3T3 cells, and cells stimulated with growth factors or 10% FBS for 3 h or pretreated with pharmacological inhibitors of signaling pathways for 12 h, were detached from tissue culture dishes with EDTA and used instantly for flow cytometry. First, live non-permeabilized cells were incubated with 20 g/ml polyclonal antibody against tTG for 30 min at 4°C. After washing the cells with PBS, they were fixed with 2% paraformaldehyde in PBS, washed again, and then stained with 20 g/ml secondary fluorescein-labeled anti-rabbit IgG for 30 min at 25°C. Staining for cell surface tTG was analyzed in FACScan TM flow cytometer (BD Biosciences).
Analysis of ERK Activation-Analysis of ERK activation was performed by SDS-PAGE with 20 g of total cell lysates on 15% acrylamide, 0.25% bisacrylamide gels followed by immunoblotting with antibodies to dually phosphorylated (activated) or total ERK1/ERK2.
Immunofluorescence Microscopy-Subconfluent cultures of vectortransfected, H-Ras-and Raf-1-transformed NIH 3T3 fibroblasts on glass coverslips were used for immunofluorescence. Confluent serumstarved untransformed NIH 3T3 fibroblasts on glass coverslips were either kept quiescent or were stimulated with 10% FBS (with or without 10 M U0126) for 3 h. Live non-permeablized cells were incubated for 30 min at 37°C with a mixture of mAbs 4G3, TG100, and CUB7402 against tTG (each at 10 g/ml) and 20 g/ml mAb HM␤1-1 against mouse ␤ 1 integrin. Then, the cells were washed with PBS and fixed with 2% paraformaldehyde in PBS. A combination of rhodamine-conjugated donkey anti-mouse and fluorescein-conjugated goat anti-hamster IgGs was used as secondary antibodies. Cells were viewed using a Nikon Eclipse E800 microscope and images were generated using a Spot RT digital camera.
Adhesion and Spreading Assays-Adhesion assays were performed as described before, using 35 S-labeled cells (18 -21). Cells were preincubated for 30 min at 4°C with 20 g/ml polyclonal antibody against tTG or control non-immune rabbit IgG before plating for 1 h on plastic wells that were coated with 10 g/ml fibronectin or its 42-kDa fragment and then blocked with 2% bovine serum albumin in PBS. At least two independent experiments were performed in triplicates for each substrate and type of transfectants. Statistical differences were determined by Student's t test. For spreading assays, cells were plated for 3 h on plastic wells coated with 10 g/ml fibronectin. Adherent and spread cells were washed with PBS, fixed with 2% paraformaldehyde, and photographed.

Transformation of Fibroblasts with Activated H-Ras or Raf-1 Decreases tTG mRNA Expression and Reduces Biosynthesis and Surface
Expression of tTG-tTG serves as a potent integrin-binding co-receptor for fibronectin on the cell surface and stimulates cell-matrix adhesion and adhesion-dependent functions (2, 18 -21). tTG also functions as a tumor suppressor (6,34). Because many oncogene-transformed cells display reduced adhesion and anchorage-independent growth, we studied possible alterations of tTG expression and complex formation with ␤ 1 integrins in these cells (Fig. 1). Expression of activated (transforming) forms of H-Ras and Raf-1, H-Ras(12V), and ⌬Raf-22W, in NIH 3T3 fibroblasts leads to multiple phenotypic changes of oncogene-transformed cells including impaired adhesion (39,40). To test the expression levels of tTG mRNA, we performed Northern analysis with NIH 3T3 cells expressing vector, H-Ras(12V), or ⌬Raf-22W (Fig. 1A). Expression of trans- forming mutants of H-Ras and Raf-1 induced by treatment of cells with mifepristone for 36 h, caused, respectively, a ϳ6and ϳ3-4-fold decrease in steady state content of tTG mRNA compared with vector-expressing controls. This reduction was primarily because of transcriptional inhibition of tTG expression, because no major change in tTG mRNA stability was found upon transformation with activated H-Ras or Raf-1 (data not shown). Transformation with activated Raf-1 and, in particular, H-Ras, markedly inhibited biosynthesis of tTG (Fig. 1B). Analysis of biosynthesis and complex formation of ␤ 1 integrins and tTG by metabolic labeling and immunoprecipitation showed very little or no changes in the levels of ␤ 1 integrins synthesized by oncogene-transformed cells (Fig. 1C, upper panel). In contrast, the amounts of tTG associated with ␤ 1 integrins were significantly decreased in H-Ras-and Raf-1-transformed cells compared with vector-transfected untransformed cells (Fig. 1C, lower panel). Stable expression of activated H-Ras and Raf-1 oncogenes in NIH 3T3 fibroblasts had similar effects on the content of tTG mRNA, biosynthesis of tTG, and its complex formation with integrins (data not shown).
In agreement, flow cytometry with non-permeabilized cells revealed decreased levels of surface tTG in the H-Ras and Raf-1 transfectants compared with vector-expressing cells ( Fig. 2A). Immunostaining of adherent cells for surface tTG and ␤ 1 integrins showed their extensive co-localization at cell-matrix con-tacts on the dorsal surface of vector-transfected NIH 3T3 fibroblasts (Fig. 2B). Similar co-distribution of ␤ 1 integrins and surface tTG was found in Raf-1-transformed cells, although tTG levels were diminished compared with vector-expressing cells. Even less tTG could be visualized by immunostaining on the surface of H-Ras-transformed cells, whereas distribution of ␤ 1 integrins on these cells remained mostly diffuse. Because the expression levels of ␣ 1 , ␣ 2 , ␣ V , and ␤ 3 integrin subunits were very low or undetectable on untransformed and H-Rasand Raf-1-transformed NIH 3T3 cells (46), the majority of surface tTG in these cells was associated with ␣ 5 ␤ 1 and ␣ 3 ␤ 1 integrins (18). Furthermore, no or little difference in the surface expression levels of the ␤ 1 , ␣ 3 , and ␣ 5 integrin subunits was found in H-Ras-and Raf-1-transformed NIH 3T3 fibroblasts (data not shown). This indicates that the reduced levels of surface tTG in H-Ras-and Raf-1-transformed cells are mostly because of down-regulation of tTG mRNA and decreased biosynthesis of tTG rather than major changes in integrin expression.

FIG. 2. Expression and localization of tTG on the surface of untransformed and H-Ras-and Raf-1-transformed NIH 3T3 fibroblasts.
A, expression levels of surface tTG were determined by staining of non-permeabilized cells with polyclonal anti-tTG antibody, followed by secondary fluorescein-labeled IgG and flow cytometry. B, adherent cells were double-stained with mAbs 4G3, TG100, and CUB7402 for surface tTG (upper panels, red) and mAb HM␤1-1 for ␤ 1 integrin (middle panels, green). Merged images (yellow) are shown in the lower panels. Note extensive co-localization of tTG with ␤ 1 integrins at cell-matrix contacts and significant reduction of tTG on the surface of H-Ras-and Raf-1-transformed fibroblasts. Bar, 25 m.
bly, preincubation with anti-tTG antibody reduced adhesion of vector-expressing transfectants to the levels characteristic for H-Ras-and Raf-1-transformed cells, whereas the same treatment had very little effect on cells expressing activated H-Ras or Raf-1 oncogenes. We also examined adhesion of these transfectants on the 42-kDa fragment of fibronectin, which interacts with cell surface tTG (18,19,47). Transformation with activated H-Ras or Raf-1 caused a ϳ3-4-fold reduction in adhesion on the 42-kDa fibronectin fragment (Fig. 3B). Treatment with anti-tTG antibody greatly reduced adhesion of vector-expressing fibroblasts to the 42-kDa fragment, whereas the decrease was less prominent in the case of H-Ras-and Raf-1-transformed fibroblasts. Moreover, cells expressing activated H-Ras or Raf-1 exhibited reduced spreading on fibronectin and, unlike vector-expressing cells, were unable to spread on its 42-kDa fragment (data not shown). Collectively, these data show that down-regulation of surface tTG expression by H-Ras and Raf-1 oncogenes contributes to adhesive deficiency of transformed cells.
Expression of Exogenous tTG Increases Adhesion and Spreading of Untransformed and H-Ras-and Raf-1-transformed Cells on Fibronectin-We expressed human tTG in untransformed and H-Ras-and Raf-1-transformed mouse NIH 3T3 fibroblasts (Fig. 4). Analysis by 35 S labeling and immuno-precipitation showed increased overall levels of tTG biosynthesis in these transfectants compared with vector-transfected untransformed and H-Ras-or Raf-1-transformed NIH 3T3 fibroblasts (Fig. 4A). Increased levels of tTG biosynthesis led to elevated surface expression of tTG in these transfectants (Fig.  4B). In agreement, expression of exogenous tTG enhanced adhesion of untransformed and Raf-1-transformed fibroblasts on fibronectin, whereas very little change was seen for Ras-transformed cells (Fig. 4C). In addition, expression of exogenous tTG caused a significant increase in adhesion of all three types of transfectants on the 42-kDa fragment of fibronectin (data not shown). Finally, exogenous tTG stimulated spreading of untransformed and H-Ras-or Raf-1-transformed NIH 3T3 fibroblasts on fibronectin (Fig. 4D). Together, these results support a role for cell surface tTG as an adhesion co-receptor for fibronectin in untransformed and oncogene-transformed fibroblasts.
Effects of Different Allelic Variants of Ras and Raf Oncogenes on tTG mRNA Expression-To test whether down-regulation of tTG mRNA is a common property of different Ras and Raf isoforms, we transiently expressed activated forms of K-Ras and B-Raf oncogenes in NIH 3T3 fibroblasts. Northern blot analysis revealed a potent down-regulation of tTG mRNA by K-Ras similar to that seen with H-Ras oncogene (Fig. 5). B-Raf also decreased steady-state levels of tTG mRNA, although its inhibitory effect was less prominent compared with that of Raf-1. Therefore, different forms of Ras and Raf oncogenes are capable of inhibiting expression of tTG mRNA.

Opposing Roles of the MEK1/ERK and Other Ras/Raf-mediated Signaling Cascades in the Regulation of tTG Expression-
Because Ras and Raf regulate a number of signaling pathways, we attempted to identify downstream signaling modules that are involved in inhibition of tTG biosynthesis and surface expression (Fig. 6). The use of the pharmacological inhibitor of MEK1/2, U0126, with untransformed and H-Ras-or Raf-1transformed NIH 3T3 fibroblasts caused a decrease in steadystate levels of tTG mRNA (Fig. 6A). Similar effects on tTG mRNA expression were seen with the MEK1 inhibitor PD98059, but they could not be observed with an inactive analog of the U0126 inhibitor, U0125 (data not shown). In contrast, blocking the p38 MAPK, PI3K, or JNK signaling pathways with, respectively, SB203580, LY294002, or SP600125, led to a modest increase in the levels of tTG mRNA in all three cell types. None of these inhibitors had a significant effect on expression of ␤-actin mRNA.
Metabolic labeling and immunoprecipitation of total cellular tTG showed that blocking MEK1/ERK signaling markedly decreased biosynthesis of tTG, whereas inhibition of the p38 MAPK, PI3K, or JNK signaling pathways had an opposite effect and elevated tTG biosynthesis (Fig. 6B). Similar effects of specific signaling pathway inhibitors were observed in the case of expression of surface tTG (Fig. 6C). Whereas MEK inhibitor U0126 reduced surface levels of tTG, blocking the p38 MAPK, PI3K, or JNK signaling pathways increased surface tTG ex-pression. Therefore, the p38 MAPK, PI3K, and JNK signaling cascades can mediate Ras-dependent down-regulation of tTG expression. Surprisingly, ERK signaling has an opposite effect and stimulates expression of tTG.
Analysis of steady-state levels of tTG mRNA by quantitative Northern blot analysis showed its elevated expression in the CA-MEK1 transfectants and a reduction in the cells expressing DN-MEK1 (Fig. 7A). Our results also indicated that activation of the MEK1/ERK signaling module up-regulated tTG mRNA expression primarily by increasing the rate of mRNA transcription, because little or no change in the degradation rate of tTG mRNA was observed in response to altering ERK signaling (data not shown). Next, we evaluated biosynthesis of tTG and ␤ 1 integrin-tTG complexes by metabolic labeling and immunoprecipitation in NIH 3T3 fibroblasts expressing CA-MEK1 or DN-MEK1. In agreement with tTG mRNA expression data, we detected an increase in biosynthesis of total cellular tTG in CA-MEK1 transfectants and its reduction in cells expressing DN-MEK1, compared with vector-transfected cells (Fig. 7B). Little or no changes in the levels of ␤ 1 integrin biosynthesis were observed in both types of transfectants (Fig. 7C, upper  panel). CA-MEK1 transfectants displayed increased amounts of tTG complexed with ␤ 1 integrins, whereas reduced levels of ␤ 1 integrin-associated tTG were found in the cells expressing DN-MEK1 (Fig. 7C, lower panel). Accordingly, CA-MEK1 elevated the levels of surface tTG, whereas DN-MEK1 decreased surface expression of tTG (Fig. 7D). Together, these results show that constitutive activation of MEK1/ERK signaling increases biosynthesis and surface expression of tTG because of up-regulation of tTG mRNA, whereas interference with this pathway inhibits tTG expression.

Modulation of Surface tTG by ERK Signaling Alters Adhesion of Fibroblasts on Fibronectin-
We tested effects of the ERK signaling pathway on adhesion of NIH 3T3 fibroblasts on fibronectin and its 42-kDa fragment that binds tTG (Fig. 8). In static adhesion assays, stimulation of ERK signaling by CA-MEK1 caused a ϳ20 -25% increase in adhesion of NIH 3T3 transfectants on fibronectin (Fig. 8A). In contrast, inhibition of the ERK pathway by expression of DN-MEK1 decreased by ϳ10 -15% adhesion on fibronectin. Preincubation with anti-tTG antibody reduced adhesion of all three types of transfectants to similar levels. We also examined adhesion of these transfectants on the 42-kDa fragment of fibronectin, which interacts with cell surface tTG (18,19,47). Expression of CA-MEK1 significantly elevated, whereas DN-MEK1 reduced adhesion on the 42-kDa fibronectin fragment (Fig. 8B). Treatment with anti-tTG antibody strongly diminished adhesion of all three types of transfectants to the 42-kDa fragment. These data show that modulation of surface tTG expression by the ERK signaling pathway alters adhesion of NIH 3T3 fibroblasts on fibronectin and its 42-kDa fragment.
Transient Activation of MEK1/ERK Signaling by Adhesion or Growth Factors Increases Biosynthesis of tTG and the Amounts of ␤ 1 Integrin-tTG Complexes-We studied whether transient, rather then sustained activation of the ERK pathway, can alter biosynthesis and surface expression of tTG. For this purpose, short term effects of adhesion and growth factors on tTG were tested in untransformed NIH 3T3 fibroblasts (Fig.  9). As determined by metabolic labeling and direct immunoprecipitation of total cellular tTG, adhesion of cells on fibronectin (Fig. 9A, left panel) or treatment of quiescent adherent cells with growth factors (Fig. 9A, right panel) both increased the amounts of de novo synthesized tTG. Analysis of ␤ 1 integrin-tTG complexes by co-immunoprecipitation showed little or no changes in the levels of ␤ 1 integrins synthesized by cells in suspension or on fibronectin and in quiescent compared with growth factor-treated cells (Fig. 9B, upper panels). In contrast, the amounts of tTG associated with ␤ 1 integrins were markedly increased in adherent cells on fibronectin versus cells in suspension (Fig. 9B, lower left panel) and in growth factor-treated cells compared with quiescent cells (Fig. 9B, lower right panel). In agreement with previous reports, adhesion on fibronectin or treatment of quiescent adherent cells with growth factors activated ERK1 and ERK2 (Fig. 9C). Thus, transient activation of the ERK signaling pathway by adhesion on fibronectin or treatment with growth factors stimulates tTG biosynthesis and raises the amounts of ␤ 1 integrin-associated tTG.

ERK Activity Is Required for Expression of Surface tTG and Co-clustering of ␤ 1 Integrins and tTG at Cell-Matrix Adhesion
Contacts-The measurement of surface tTG by fluorescenceactivated cell sorter analysis with live untransformed NIH 3T3 fibroblasts demonstrated a ϳ2.5-3.5-fold increase in its amounts on adherent and growth factor-stimulated cells, compared, respectively, with cells in suspension and quiescent cells (Fig. 10A). Again, a specific inhibitor of MEK, U0126, decreased the expression of surface tTG on serum-induced adherent cells to the levels detected on quiescent cells. However, inhibitors of other signaling pathways were essentially unable to reverse the effect of growth factors on surface tTG expression ( Fig. 6C and data not shown). These results indicate that the ERK pathway has a key role in up-regulation of surface tTG in untransformed fibroblasts in response to growth factor stimulation. Immunofluorescence staining of quiescent cells revealed small dot-like clusters of ␤ 1 integrins and tTG on the dorsal surface (Fig. 10B, upper panels). Treatment with serum caused a rapid growth of dorsal cell-matrix contacts containing ␤ 1 integrins and surface tTG (Fig. 10B, middle panels), which were co-localized with nascent fibronectin fibrils (data not shown). The inhibitor of MEK, U0126, efficiently suppressed growth factor-induced co-clustering of ␤ 1 integrins and tTG on the dorsal cell surface (Fig. 10B, lower panels), whereas the pharmacological inhibitor of p38 MAPK, SB203580, was unable to do so (data not shown). Merged images revealed a significant overlap between stainings for ␤ 1 integrins and surface tTG in the case of growth factor-treated cells, whereas it was less prominent in the case of quiescent or U0126-treated cells (Fig. 10B, right panels). Together, these data show that transient activation of the ERK signaling pathway by adhesion or growth factors up-regulates expression of tTG on the cell surface as a result of increased tTG biosynthesis. Moreover, ERK signaling controls localization of ␤ 1 integrins and associated tTG on the cell surface via regulation of clustering of ␤ 1 integrin-tTG complexes at cell-matrix contacts. DISCUSSION tTG is a multifunctional protein (1, 2) that promotes cellmatrix adhesion by mediating the interaction between the extracellular domains of ␤ 1 and ␤ 3 integrin subunits and fibronectin (18). In this work we continued to explore mechanisms that regulate cell-matrix adhesion by altering expression and localization of tTG on the surface of normal and oncogene-transformed cells. The first recently identified pathway of modulating cell surface tTG is based on its matrix-dependent degradation by membrane-bound and soluble metalloprotein- ases (19). Although at this moment mechanisms of tTG externalization remain unknown, one can suggest that cells are capable of regulating expression of surface tTG by altering its transport to the surface. Because tTG binds to non-glycosylated immature ␤ 1 integrins inside the cell (18) and the interaction with fibronectin may be important for tTG externalization (51), expression of tTG on the surface may depend on post-translational association of tTG with integrins and/or fibronectin within the endoplasmic reticulum or Golgi. Our recent results show that, while altering tTG biosynthesis, H-Ras/Raf-1 oncogenes and several major signaling cascades do not change the fraction of tTG that forms complexes with ␤ 1 integrins. In turn, the amounts of tTG in complexes with ␤ 1 integrins appear to correlate closely with the surface levels of tTG. These observations suggest a participation of integrins in the "default" pathway of tTG transport to the surface that operates in both untransformed and oncogene-transformed cells.
In this work we present evidence for a novel regulatory mechanism, whereby activated H-Ras and Raf-1 oncogenes decrease surface levels of tTG because of transcriptional inhibition of its expression. Moreover, this ability to downregulate tTG is shared by several members of Ras and Raf oncogene families. An initial evaluation with pharmacological inhibitors indicates a role for the p38 MAPK, PI3K, and JNK signaling pathways in down-regulation of surface tTG in H-Ras-and Raf-1-transformed cells. Further analysis of this regulation requires identification of transcription factors and mapping regulatory elements within the tTG promoter, which are involved in transcriptional inhibition of tTG mRNA expression. A common epigenetic mechanism for gene silencing, DNA methylation, was shown to decrease the expression of the tTG gene in several lines of neoplastic human cells (52). It remains to be determined whether H-Ras and Raf-1 oncogenes cause demethylation of the tTG promoter and if so, which signaling pathways can mediate this effect. Yet, our most recent results indicate that inhibition of CpG methylation by azadeoxycytidine effectively restores tTG expression in H-Ras-transformed NIH 3T3 fibroblasts (data not shown). Surprisingly, in untransformed fibroblasts either sustained or transient activation of the key Ras/Raf-dependent downstream signaling module, MEK1/ERK, exerts an opposite effect on tTG expression. This is caused by increasing the rate of transcription of tTG mRNA, which leads to elevated biosynthesis and surface expression of tTG. Moreover, as judged by growth factor-induced co-clustering of ␤ 1 integrins and tTG, ERK activity also controls the formation of cellmatrix adhesion contacts in untransformed fibroblasts, likely via increasing RhoA activation and formation of contractile actin stress fibers. These data underscore a positive role of this pathway in the regulation of cell-matrix adhesion. Indeed, activation of the MEK1/ERK signaling cascade increases expression of ␤ 3 integrin in fibroblasts (53) and the ␣ 2 integrin subunit in epithelial cells (54). Also, previous work demonstrated that stimulation of the ERK pathway promotes ␤ 1 integrin-dependent adhesion of monocytes (55) and amplifies expression of ␤ 2 integrins and ␤ 2 integrin-mediated adhesiveness of neutrophils (56,57). Two recent studies showed that ERK activity is required for activation of ␣ IIb ␤ 3 integrin in platelets (58) and signaling pathways other than the MEK1/ERK mediate H-Ras-and Raf-1-dependent suppression of integrin activation in fibroblasts (59,60). Therefore, in untransformed fibroblasts, adhesion-and growth factor-dependent activation of the MEK1/ERK signaling module may stimulate the adhesive function of both integrins and integrin-associated surface tTG and, therefore, promote and sustain cell-matrix adhesion.
While this study was being prepared for publication, Antonyak et al. (61) reported that the Ras-ERK signaling pathway inhibits retinoic acid-induced stimulation of tTG expression in NIH 3T3 fibroblasts. The ability of Ras to down-regulate tTG reported in this work is in agreement with our results. However, these authors also showed that the ERK pathway may inhibit retinoic acid-induced expression of tTG. Because retinoic acid itself greatly stimulates tTG biosynthesis (24 -26) and, simultaneously, affects multiple features of intracellular signaling (2,6), including activation of ERKs, PI3K, and other signaling intermediates, it might be difficult to ascertain the exact role of ERK signaling in tTG expression in that system because of interference of retinoic acid with multiple signaling pathways. Moreover, concurring with our findings, this study showed that treatment of quiescent adherent fibroblasts with epidermal growth factor apparently increased tTG expression in the absence of retinoic acid (61). Therefore, without other and on quiescent cells (right panel) were taken as 100%. B, ERK activation is required for growth factor-induced co-clustering of tTG with ␤ 1 integrins at the dorsal cell-matrix adhesion contacts. Adherent quiescent cells and cells treated for 3 h with 10% FBS or with 10% FBS and 10 M U0126 were double-stained with mAbs 4G3, TG100, and GUB7402 for surface tTG (left panels, red) and mAb HM␤1-1 for ␤ 1 integrin (middle panels, green). Merged images (yellow) are shown in the right panels. Bar, 20 m. factors that affect tTG expression, both constitutive and transient stimulation of ERK signaling up-regulate tTG mRNA, induce tTG biosynthesis, and stimulate adhesive function of tTG in untransformed fibroblasts.
In contrast, increased activation of other major Ras-and Rafmediated signaling cascades in transformed fibroblasts inhibits expression of tTG by overcoming the stimulatory effect of ERK signaling. Reduced expression of tTG should relieve oncogenetransformed cells from anti-proliferative effects of this protein (2,35) and alleviate its tumor suppressor function (34). Our results with pharmacologic inhibitors indicate that several other Rasmediated signaling pathways, such as PI3K, p38 MAPK, and JNK, might be involved in down-regulation of tTG expression. Meanwhile, a recent report showed that PI3K is required for induction of tTG by retinoic acid, whereas overexpression of the constitutively active form of PI3K failed to up-regulate tTG in the absence of retinoic acid (62). Together, these findings suggest that overall regulation of tTG expression is profoundly altered by retinoids because of both direct induction of the tTG gene and indirect effects on several signaling pathways that, in turn, are capable of modulating tTG expression.
Our experiments with expression of exogenous tTG in untransformed and H-Ras-and Raf-1-transformed fibroblasts (Fig. 4) generally reiterated the role of surface tTG as adhesion co-receptor for fibronectin. However, while the effect of surface tTG on cell adhesion was prominent in untransformed cells, it was more limited in Raf-1-and absent in H-Ras-transformed fibroblasts. This indicates that Raf-1and H-Ras-transformed cells become partly or fully insensitive to tTG-dependent adhesion, likely because of functional down-regulation of other adhesion receptors caused by oncogenic transformation. One such example is a suppression of integrin activation by H-Ras and Raf-1 oncogenes (59). Apparently, surface tTG is unable to compensate for deficiency in integrin adhesive function.
Recent studies showed that overall suppression of the cellmatrix adhesion system in H-Ras-and Raf-1-transformed fibroblasts may be caused in part by deactivation of integrins (57,58) and inhibition of fibronectin biosynthesis and assembly (63). Here we found that it also includes a reduction of surface tTG as a part of ternary integrin-tTG-fibronectin complexes that promote adhesiveness (18) and fibronectin matrix formation (21). Coordinated down-regulation of these proteins on the cell surface contributes to adhesion impairment caused by H-Ras and Raf-1 oncogenes. Modulation of tTG expression by Ras, Raf, and ERK signaling emphasizes adhesive function of tTG on the surface of untransformed cells and its efficient suppression by activated oncogenes.