Activation of the Ras-ERK pathway inhibits retinoic acid-induced stimulation of tissue transglutaminase expression in NIH3T3 cells.

Retinoic acid (RA) is a potent activator of tissue transglutaminase (TGase) expression, and it was recently shown that phosphoinositide 3-kinase (PI3K) activity was required for RA to increase TGase protein levels. To better understand how RA-mediated TGase expression is regulated, we considered whether co-stimulation of NIH3T3 cells with RA and epidermal growth factor (EGF), a known activator of PI3K, would facilitate the induction or increase the levels of TGase expression. Instead of enhancing these parameters, EGF inhibited RA-induced TGase expression. Activation of the Ras-ERK pathway by EGF was sufficient to elicit this effect, since continuous Ras signaling mimicked the actions of EGF and inhibited RA-induced TGase expression, whereas blocking ERK activity in these same cells restored the ability of RA to up-regulate TGase expression. However, TGase activity is not antagonistic to EGF signaling. The mitogenic and anti-apoptotic effects of EGF were not compromised by TGase overexpression, and in fact, exogenous TGase expression promoted basal cell growth and resistance to serum deprivation-induced apoptosis. Moreover, analysis of TGase expression and GTP binding activity in a number of cell lines revealed high basal TGase GTP binding activity in tumor cell lines U87 and MDAMB231, indicating that constitutively active TGase may be a characteristic of certain cancer cells. These findings demonstrate that TGase may serve as a survival factor and RA-induced TGase expression requires the activation of PI3K but is antagonized by the Ras-ERK pathway.

Tissue transglutaminase (TGase) 1 is a member of a family of enzymes that catalyzes a calcium-dependent transamidation reaction that results in the covalent linkage of donor glutamine residues of one protein to acceptor primary amino groups of another protein or polyamine (1,2). This modification of pro-teins by TGase has been implicated in regulating numerous physiological processes including endocytosis (3), axonal growth (4), cell differentiation (5), and cell survival (6 -8). However, the most frequently reported cellular outcome associated with aberrant TGase transamidation activity has been the induction of apoptosis (8 -12), suggesting that the extent of TGase enzymatic activity may be crucial for determining the exact cellular effects induced by the TGase.
In addition to transamidation activity, TGase also undergoes a GTP binding/GTP hydrolysis cycle, a feature that distinguishes it from the other transglutaminases (2). GTP-bound TGase participates in signaling pathways that link cell surface receptors to intracellular effectors. The most studied example of this involved the up-regulation of phospholipase C activity by the ␣ 1 -adrenergic receptor (13,14), where it was shown that the ␣ 1 -adrenergic receptor stimulated the GTP binding activity of TGase. The GTP-bound TGase species, in turn, was shown to stimulate phosphoinositide lipid hydrolysis by phospholipase C. The ability of TGase to transduce signals adds another dimension to its functionality and may help explain how the same protein can be linked to several, sometimes opposing, cellular responses. This point is exemplified by the recent finding that a transamidation-defective form of TGase protected cells from heat shock-induced apoptosis (8), indicating that the GTP binding/GTPase activity of TGase likely mediated the survival advantage.
Despite advances in the characterization of TGase functions and their impact on cellular processes, the events that lead to a completely activated TGase species (a molecule having transamidation activity and capable of binding GTP) in vivo are not completely understood. One aspect of TGase regulation involves altering the expression levels of the protein. In several cell types, the TGase protein is expressed at low levels (4,7,15), and increases in TGase expression and activation can be induced following chronic stimulation with growth factors (16), chemotherapeutic agents (17), cytokines (18), and retinoids (19). Retinoic acid (RA) is the most consistent inducer of TGase expression and activation, and a mechanism by which RA upregulates TGase expression has been elucidated. RA binds to retinoic acid receptors (RARs) and stimulates their association with specific DNA motifs located in the promoter region of the TGase gene, resulting in transcription (2). Other than modifying gene transcription, RARs also activate signaling molecules. The signaling potentials of STAT3, PI3K, and ERK have been shown to increase following RA treatment (20 -22), suggesting that some RA-mediated effects may require the combined activation of certain signaling proteins and the transcriptional capability of the RARs.
For some time our laboratory has been interested in understanding the regulation and function of TGase in cells. Con-trary to findings linking TGase to apoptosis (8 -12), we have shown that RA-induced TGase expression provided a protective effect from N-(4-hydroxyphenyl)retinamide (HPR)-mediated apoptosis in HL-60 and NIH3T3 cells (7). Moreover, RA stimulation resulted in the activation of the anti-apoptotic molecule PI3K, which was necessary for the induction of TGase expression and GTP binding activity (23). Given that the regulation of TGase expression is linked to PI3K activation and that growth factors frequently activate PI3K, we considered whether cotreatment of NIH3T3 cells with growth factors and RA would potentiate the induction of TGase expression, as well as the level of transamidation or GTP binding activity, compared with RA treatment alone. Epidermal growth factor (EGF) was chosen as the co-stimulus because it is not only an effective activator of PI3K but it also up-regulates TGase activity in hepatocytes (24). However, we found that rather than enhancing the induction rate or the expression levels of TGase, EGF suppressed the ability of RA to stimulate TGase expression. Activation of the Ras-ERK pathway was sufficient to impart this inhibitory effect of EGF, since cells with continuous Ras signaling also limited RA-induced TGase expression, and blocking ERK activation in these cells restored the ability of RA to up-regulate TGase expression. These findings prompted us to question how EGF-regulated processes would be affected by exogenous TGase expression. Cells expressing TGase displayed a growth and survival advantage over control cells, and EGFmediated proliferation and resistance to serum deprivationinduced apoptosis was not compromised by the overexpression of TGase. These findings demonstrate that EGF inhibits RAinduced TGase expression in NIH3T3 cells by activating the Ras-ERK pathway, whereas limiting TGase expression does not appear to be critical for EGF to elicit its effects. When coupled with our earlier findings linking the activation of PI3K to the induction of TGase expression, we begin to appreciate the complex interplay of signaling pathways that regulate TGase in cells. Moreover, these results are consistent with the idea that EGF-coupled signaling does not counteract the upregulation of TGase as a means of reversing a TGase-mediated growth inhibitory or apoptotic activity as TGase exhibits mitogenic and survival activities rather than serving as a suppressor of cell growth. Thus, the effects of EGF receptor signaling on TGase expression are more likely the outcome of EGFdirected effects on the broad profile of RA-stimulated gene expression, in order to repress the anti-mitogenic effects of other RA-induced gene products.

EXPERIMENTAL PROCEDURES
Materials-LY294002 and PD98095 were obtained from Calbiochem, and EGF was from Invitrogen. RA, HPR, platelet-derived growth factor (PDGF), and monodansyl cadaverine were purchased from Sigma. The TGase antibody was obtained from Neomarkers; the actin antibody was from Sigma, and the phospho-ERK and ERK antibodies were from New England Biolabs. [␣-32 P]GTP was purchased from PerkinElmer Life Sciences, and all additional material was obtained from Fisher unless stated otherwise.
Cell Culture-NIH3T3 cells were grown in Dulbecco's modified Eagle's medium containing 10% calf serum and 100 units/ml penicillin, and MCF-7 cells were grown in RPMI 1640 medium containing 10% fetal bovine serum and 100 units/ml penicillin. The cell lines were maintained in a humidified atmosphere with 5% CO 2 at 37°C. For the various treatments described, the cells were grown to near confluence in medium containing 10% serum, and then medium containing 1% serum with 5 M RA (unless indicated otherwise), or 100 ng/ml EGF, Ϯ 10 M LY294002 or Ϯ 10 M PD98095, were added for the times indicated under "Results." Cells were rinsed with phosphate-buffered saline (PBS) and then lysed with cell lysis buffer (10 mM Na 2 HPO 4 , 150 mM NaCl, 1% Triton X-100, 0.5% sodium deoxycholate, 0.1% SDS, 0.004% NaF, 1 mM NaVO 4 , 25 mM ␤-glycerophosphoric acid, 100 g/ml phenylmethanesulfonyl fluoride, and 1 g/ml each aprotinin and leupeptin, pH 7.35). The lysates were clarified by centrifugation at 12,000 ϫ g for 10 min at 4°C. Protein concentrations were determined using the Bio-Rad DC protein assay.
Western Blot Analysis-Total cell lysates from each sample were combined with Laemmli sample buffer, boiled, and subjected to SDS-PAGE. The proteins were transferred to nitrocellulose filters and blocked with TBST (20 mM Tris, 137 mM NaCl, pH 7.4, and 0.02% Tween 20) containing 5% nonfat dry milk. The filters were incubated with the various primary antibodies diluted in TBST overnight at 4°C and then washed 3 times with TBST. To detect the primary antibodies, anti-mouse or -rabbit conjugated to horseradish peroxidase (Amersham Biosciences), diluted 1:5000 in TBST, was incubated with the filters for 1 h, followed by 3 washes with TBST. The protein bands were visualized on x-ray film after exposing the filters to chemiluminescence reagent (ECL, Amersham Biosciences).
Photoaffinity Labeling of TGase-Photoaffinity labeling of TGase was performed by incubating whole cell lysates with 5 Ci of [␣-32 P]GTP in 50 mM Tris-HCl, pH 7.4, 2 mM EGTA, 1 mM dithiothreitol, 20% (w/v) glycerol, 100 mM NaCl, and 500 M AMP-PNP for 10 min at room temperature. The samples were placed in an ice bath and irradiated with UV light (254 nm) for 15 min, mixed with 5ϫ Laemmli sample buffer, and boiled. SDS-PAGE was performed, followed by transfer to nitrocellulose filters and exposure on x-ray film.
Nuclear Condensation or Blebbing Assay-Cells were seeded in 6-well dishes and grown in complete medium for 2 days. The cells were then incubated in medium containing 1.5% serum Ϯ 5 M RA (unless indicated otherwise) and Ϯ 100 ng/ml EGF for 2 days. The cultures were then incubated with serum-free medium for an additional day and were fixed and stained with 4,6-diamidino-2-phenylindole (2 g/ml) for viewing by fluorescence microscopy. Apoptotic cells were identified by condensed nuclei and/or blebbing.
Cell Growth Assays-Cells were seeded at 1 ϫ 10 5 cells/well and grown in medium containing 1.5% serum Ϯ 5 M RA and Ϯ 100 ng/ml EGF. Every 2 days the medium was changed. At 0, 2, 4, and 6 days of treatment, the cells were collected and counted.

TGase Protects NIH3T3 Cells from Serum Deprivation-induced
Apoptosis-Exposure of cells to RA limits cell growth or induces a cell death response in several cell types suggesting that retinoids may be efficacious in the treatment of certain forms of cancer. More recent work has shown that exposure of GL-15 glioma cells to 1 M RA promoted cell proliferation (20), and findings from our laboratory demonstrated that RA affords protective effects to HL-60 and NIH3T3 cells from the apoptotic agent N-(4-hydroxyphenyl)retinamide (HPR) (7). We decided to test whether RA treatment could also protect NIH3T3 cells from serum deprivation-mediated cell death. Mouse fibroblast cells were chosen for this study because it was determined previously that exposure of NIH3T3 cells to RA caused the up-regulation of TGase expression, as well as stimulated the GTP binding and transamidation activities of TGase, and that the RA-induced TGase activity limited the effectiveness of HPR at inducing apoptosis (7). Consistent with these findings, Fig.  1A shows that NIH3T3 cells treated with 5 M RA for 2 days resulted in an increase in TGase protein levels (Fig. 1A, ␣TGase) and yielded a corresponding stimulation of GTP binding activity (Fig. 1A, [␣-32 P]GTP binding) as indicated by the incorporation of [␣-32 P]GTP. The extent of TGase expression and GTP binding activity stimulated at significantly higher levels of RA (50 M) was comparable with that obtained with 5 M RA, indicating that the latter level was sufficient to yield a maximum TGase response.
We then examined whether RA protected these cells from serum deprivation-induced apoptosis. The fibroblasts were stimulated with or without RA for 2 days and then were stressed by serum starvation, and the resulting apoptotic rates were compared. Cells that were not treated with RA were susceptible to apoptosis, as nearly 70% of the cells displayed nuclear condensation or blebbing, a characteristic unique to cells undergoing apoptosis (Fig. 1B, vector). To confirm that these cells were apoptotic, a second assay that detects the activation of caspases was also performed. Total cell lysates from cells that were or were not deprived of serum were subjected to Western blot analysis with an antibody that recognizes the cleaved (activated) form of caspase 3. Consistent with cells undergoing programmed cell death, cleaved caspase 3 was readily detected in the serum-deprived cells but not in the actively growing cells (data not shown). Treatment with 5 M RA resulted in a dramatic reduction in the rate of cell death compared with control cells (Fig. 1B, vector), demonstrating that RA can provide a protective effect from serum deprivationinduced apoptosis as well as HPR-mediated apoptosis (7). The increase in TGase expression caused by 5 M RA (Fig. 1B, inset, vector) can account for the pro-survival effect of RA, because simply overexpressing TGase in NIH3T3 cells (Fig. 1B, inset, WT TGase) is sufficient to inhibit apoptosis resulting from serum deprivation, and does so as effectively as treating cells with 5 M RA (Fig. 1B, WT TGase).
TGase has long been considered an apoptotic or cell differentiation factor (5, 8 -12), yet the data presented here and in our previous work (6, 7) have implicated TGase as a survival factor. We questioned whether the anti-apoptotic effect of TGase was unique to NIH3T3 cells or whether TGase might provide a protective effect to other cell lineages. To address this possibility, an analysis of TGase expression and GTP binding activity in various cell lines was performed. We reasoned that if TGase activation were detrimental to cell proliferation then it would not be detected in any actively growing cells. However, if TGase was linked to a survival response, then its activity could be enhanced in some cell types. The data presented in Fig. 2 shows that treatment with RA increased TGase expression and GTP binding activity in all of the cell lines assayed, indicating that various cell types express functional RARs and can induce TGase expression and activation similar to NIH3T3 cells. The monkey kidney cell line COS7, the breast tumor cell line MDAMB231, and the brain tumor cell line U87 all displayed substantial levels of TGase expression even prior to RA treatment. Cervical carcinoma (HeLa) cells displayed detectable TGase expression in the absence of RA treatment; however, incorporation of [␣-32 P]GTP into TGase only occurred after treatment with RA (25). In contrast, the TGase expressed in MDAMB231 and U87 cells was able to incorporate [␣-32 P]GTP in the absence of RA stimulation (Fig. 2). Thus different cell lines display varying degrees of basal TGase expression and GTP binding activity, arguing against a general role for chronic TGase GTP binding activity in limiting cell proliferation and/or promoting apoptosis.

FIG. 2. Loss of regulation of TGase GTP binding activity in certain cell lines.
Cell lines were grown for 2 days in complete media and then placed in low serum. The following day, cells were exposed to low serum Ϯ 5 M RA for 2 days and then lysed. The cell extracts were used to determine the expression level of TGase by Western blot analysis and GTP binding activity by photoaffinity labeling as outlined under "Experimental Procedures." FIG. 1. RA-induced TGase expression protects cells from serum deprivation-mediated apoptosis. A, cells were exposed to media containing low serum and different concentrations of RA for 2 days and then were lysed. The cell extracts were used to determine TGase GTP binding activities using an affinity labeling assay with radioactive GTP as outlined under "Experimental Procedures." Western blot analysis using TGase and actin antibodies was performed to assess the expression levels of each of these proteins. B, NIH3T3 cells expressing only vector (vector) or WT TGase (HA-tagged WT TGase) were maintained in low serum Ϯ RA for 2 days and then the cells were deprived of serum. A day later the cells were collected and scored for apoptosis as described under "Experimental Procedures." The inset depicts the expression levels of the RA-induced and exogenously expressed TGase in the clones prior to serum starvation.

EGF Stimulation of NIH3T3 Cells Limits the RA-induced
Expression and Activation of TGase-RA treatment has been shown to result in the up-regulation of a growing number of intracellular signaling proteins including ERK (22), STAT3 (20), and PI3K (21), and it was demonstrated recently (23) that PI3K activity was essential for RA-induced TGase expression. Each of these signaling molecules is frequently used by growth factors, such as EGF, to regulate a variety of cellular processes (26), raising the possibility that TGase expression and GTP binding activity may be positively regulated by growth factor stimulation in NIH3T3 cells. As shown in Fig. 3A, treatment of NIH3T3 cells for 3 days with EGF does not lead to an increase in TGase expression nor does EGF stimulate the GTP binding activity of TGase. Because RA-mediated TGase expression requires PI3K, together with the fact that EGF directly activates PI3K, we questioned whether the co-stimulation of cells with RA and EGF might synergize to elicit a more rapid induction of TGase expression or increase the overall level of TGase protein compared with RA stimulation alone. However, instead of increasing either of these parameters, EGF significantly diminished, but did not completely block, the ability of RA to augment the expression and GTP binding activity of the TGase (Fig. 3A). The EGF-mediated inhibition of retinoid-induced TGase expression was not dependent on the concentration of RA used to stimulate these cells, as exposure with up to 75 M RA was not able to overcome the inhibitory effects of 30 min of EGF pretreatment (Fig. 3B). It is worth noting, however, that if TGase expression was initially up-regulated by RA, followed by stimulation of the cells with EGF for 3 additional days, the growth factor was ineffective at decreasing TGase expression or GTP binding activity (Fig. 3C), despite being capable of fully activating ERK (Fig. 3D). Taken together, these findings indicate that although RA and EGF activate common signaling proteins, there must be some degree of signaling specificity since only RA stimulates TGase expression. More importantly, EGF-mediated signaling inhibits retinoid-induced TGase expression in NIH3T3 cells only when the cells are pretreated with EGF or when EGF and RA are added together, whereas EGF appears to be ineffective at down-regulating TGase expression and GTP activity when added subsequent to RA treatment.
EGF is a well established mitogenic and anti-apoptotic factor (26), whereas RA has classically been considered to be a cell differentiation factor (27). Because of the opposing functions attributed to EGF and RA, it became important to determine how these factors, alone and in combination, affect the apoptotic and cell growth rates of NIH3T3 cells. As expected, EGF caused a 50% reduction in the rate of cell death induced by serum starvation (Fig. 4A) and stimulated a 2.5-fold increase in cell number following 6 days of growth factor treatment (Fig.  4B). Exposure to 5 M RA protected the cells from serum deprivation-mediated apoptosis to an extent similar to that observed with EGF (Fig. 4A). Combining both EGF and RA did not significantly reduce the apoptotic rate any further than treating with either factor alone. This was to be expected since the protective effects of RA from stress-induced apoptosis appear to occur via the up-regulation of TGase activity (Fig. 1B), and EGF suppresses most of the RA-induced TGase expression and thereby eliminates the anti-apoptotic effect of RA. As a result, the extent of resistance from serum deprivation-induced apoptosis caused by co-stimulation with RA and EGF is similar to that obtained when treating with EGF alone. Interestingly, exposure of cells to 5 M RA, which strongly inhibits the proliferation of the breast carcinoma cell line MCF-7 (Fig.  4C) and differentiates the leukemia cell line HL-60 (28), did not block the basal or EGF-mediated growth of NIH3T3 cells (Fig. 4B). Moreover, both the untreated and EGF-stimulated cells appeared to grow slightly better in the presence of RA. These findings suggest that 5 M RA, which is capable of eliciting an anti-apoptotic effect in mouse fibroblasts (Fig.  1B), also does not inhibit the growth of these cells, regardless of whether or not RA is able to up-regulate TGase expression and activity.

FIG. 3. EGF inhibits the ability of RA to induce TGase expression and GTP binding activity.
Cells were grown to near confluence and then were placed in low serum. The following day, cells were treated singly with 5 M RA or 100 ng/ml EGF or co-stimulated with 5 M RA and 100 ng/ml EGF for 2 days and then lysed (A), pretreated with 100 ng/ml EGF for 30 min before being exposed to various doses of RA for 2 days and then lysed (B), pretreated with 5 M RA or 100 ng/ml EGF for 2 days before being exposed to medium Ϯ 5 M RA and Ϯ 100 ng/ml EGF for 3 additional days before being lysed (C), or pretreated with 5 M RA for 2 days before being stimulated with 100 ng/ml EGF for the times indicated and then lysed (D). The cell extracts were used to determine the expression level of TGase by Western blot analysis and GTP binding activity by photoaffinity labeling as outlined under "Experimental Procedures." The activation state of ERK was measured via Western blot analysis with a phospho-specific ERK antibody (P-ERK).

Ras Activity Is Sufficient to Suppress RA-induced TGase
Expression-To explore further the relationship between retinoid-induced TGase expression and the negative effects that EGF treatment has on this process, we set out to define the signaling cascade(s) used by EGF to inhibit RA-induced TGase expression. We first examined whether the suppressive effects of EGF on the RA-mediated expression and GTP binding activity of TGase were specific for EGF or if other mitogenic agents could also inhibit RA-induced TGase in these cells. For these studies, 20% calf serum was used as a general mitogenic factor since serum is essential for long term culturing of cells, and the addition of serum to serum-deprived cells stimulates the activities of several protein kinases (29). In addition, PDGF, a natural ligand for NIH3T3 cells, was also tested for its ability to block RA-mediated TGase up-regulation. Fig. 5A shows again that pretreatment of cells with EGF inhibited the ability of RA to induce TGase expression and GTP binding activity by about 90%. Cells maintained in medium containing 20% calf serum or PDGF together with RA for 2 days also resulted in a nearly complete inhibition of RA-mediated TGase expression.
The above finding implies that signal transduction pathways commonly activated by growth-promoting factors may be responsible for the inhibition of retinoid-mediated expression of TGase. The small GTP-binding proteins Ras and Cdc42 are two signaling molecules that are often activated by mitogens (30 -33) and aberrant activation of either small G-protein in fibroblasts leads to cellular transformation (34,35). NIH3T3 cells stably overexpressing dominant-active forms of Ras (Ras(G12V)) and Cdc42 (Cdc42(F28L)) are available in our laboratory (Fig. 5B), and each of these cell lines has been extensively characterized and shown to display properties related to transformation including growth in low serum, resist-FIG. 4. RA does not compromise EGF-mediated cell growth and protection from apoptosis. A, NIH3T3 cells were exposed to low serum Ϯ RA and Ϯ EGF for 2 days and then were serum-starved. The following day, the cells were scored for programmed cell death as described under "Experimental Procedures." The assay was performed 3 times, and the average percentage of cell death is shown. B, NIH3T3 cells were seeded at 1 ϫ 10 5 cells/well and grown in low serum medium Ϯ 5 M RA and Ϯ 100 ng/ml EGF for the times indicated and then counted. C, MCF-7 cells were seeded at 1 ϫ 10 5 cells/well and grown in low serum Ϯ 5 M RA and counted at the times indicated.

FIG. 5. Dominant-active Ras, but not Cdc42, can mimic the ability of EGF to inhibit RA-induced TGase expression.
A, cells maintained in low serum for 1 day were pretreated with 100 ng/ml EGF, 20% calf serum (20% CS), or 50 ng/ml PDGF for 1 h and then exposed to 5 M RA for 2 days and lysed. B, cells expressing only vector (vector), a dominant-active form of Ras (Ras G12V), and a dominant-active form of Cdc42 (Cdc42 F28L) were exposed to low serum Ϯ 5 M RA for 2 days and then lysed. The whole cell extracts from each assay were used to determine the expression levels of TGase and ERK by Western blot analysis and GTP binding activity by photoaffinity labeling as outlined under "Experimental Procedures." ance to apoptosis, and anchorage-independent growth. We have used these cell lines to determine whether activation of Ras or Cdc42 mimics the ability of EGF to inhibit RA-induced TGase expression. Control cells, as well as cells expressing Ras(G12V) and Cdc42(F28L), were placed in low serum with or without 5 M RA for 3 days, and the cells were then analyzed for TGase expression and GTP binding capability. Cells expressing Cdc42(F28L) showed an induction of TGase expression and GTP binding activity by RA that was similar to that observed in control cells (Fig. 5B). In contrast, the expression level and GTP binding activity of TGase induced by RA treatment of Ras(G12V)-expressing cells was significantly reduced compared with control cells or cells expressing Cdc42(F28L). These findings indicate that the ability of EGF to activate Ras, but not Cdc42, was sufficient for the EGF-mediated inhibition of RA-induced TGase expression.
Inhibition of ERK Activity Restores the Ability of RA to Upregulate TGase Expression in Ras-transformed Cells-Ras can regulate the activities of multiple signaling molecules, and ERK is perhaps the most well studied (26). The activation of ERK by Ras proceeds through the sequential activation of several intermediate protein kinases that have been collectively referred to as the Ras-ERK pathway. The signal transducing end point of the Ras-ERK cascade occurs when ERK becomes phosphorylated by its immediate activator, MEK, and translocates to the nucleus where it modulates the activity of transcription factors to promote cell growth (36). Given that Ras(G12V), but not Cdc42(F28L), chronically activates ERK (37,38), together with the finding that Ras(G12V)-expressing cells show a loss in the ability of RA to induce TGase expression, raises the possibility that Ras utilizes ERK to regulate negatively the retinoid-induced expression of TGase. To determine whether this is the case, we inhibited ERK activation in cells expressing dominant-active Ras using a specific chemical inhibitor that blocks the activation of MEK. As shown in Fig.  6A, incubation of cells expressing dominant-active Ras with the MEK inhibitor, PD98095, by itself did not result in increased TGase expression. However, when cells incubated with PD98095 were exposed to RA for 2 days, increases in TGase expression and GTP binding activity were observed. Parallel assays conducted on control cells and Cdc42(F28L)-expressing cells showed that PD98095 had no effect on the ability of RA to induce TGase expression or GTP binding activity (Fig. 6A). The fact that PD98095 treatment enhanced RA-induced TGase expression only in cells expressing Ras(G12V) implies that the MEK inhibitor was specific for down-regulating Ras-activated ERK and was not having a more general inhibitory effect. The level of RA-mediated TGase expression and GTP binding activity observed in the Ras(G12V)-expressing cells, under conditions where ERK activity was inhibited, was comparable with the level of TGase expression and activation induced by RA in control cells (Fig. 6B). This suggests that of the several signaling proteins regulated by Ras, activation of ERK is sufficient for the Ras-mediated inhibition of RA-induced TGase expression.
Given our previous work showing that PI3K activity was required for retinoid-stimulated TGase expression in NIH3T3 cells (23), we examined whether PI3K had a similar role in cell lines that stably expressed the dominant-active forms of Ras and Cdc42. In order to inhibit PI3K activity, we took advantage of the chemical inhibitor LY294002. Each cell line was incubated with LY294002 for 3 h prior to being exposed to RA for 2 days, and then the expression levels and GTP binding activity of TGase were compared for the different conditions. Fig.  6A shows that incubation of mouse fibroblasts expressing Cdc42(F28L) with LY294002 yielded a considerable decrease in RA-induced TGase expression and GTP binding activity. Even the already compromised ability of RA to induce the GTP binding activity of TGase in Ras(G12V)-expressing cells was further accentuated when PI3K activity was inhibited. Moreover, the ability of PD98095 treatment to restore RA-induced TGase expression in cells stably expressing Ras(G12V) was completely abrogated with the addition of LY294002 (Fig. 6B). These findings indicate that PI3K activity is indispensable for RA to induce TGase expression and activation in normal cells, as well as in both Cdc42-and Ras-transformed NIH3T3 cells.
TGase Overexpression Enhances EGF-mediated Growth and Protection from Apoptosis-Activation of the EGF receptor typically up-regulates the expression and activation of proteins FIG. 6. Inhibiting ERK activity in cells expressing dominantactive Ras restored the ability of RA to up-regulate TGase expression and GTP binding activity. A, vector only (vector), dominant-active Ras (Ras G12V), and dominant-active Cdc42 (Cdc42 F28L) expressing cells were exposed to low serum medium containing 10 M LY294002 (LY) or 10 M PD98095 (PD) Ϯ 5 M RA for 2 days and then lysed. B, cells expressing dominant-active Ras (Ras G12V) were exposed to medium containing 10 M LY294002 (LY) and 10 M PD98095 (PD) singly or combined together with (or without) 5 M RA for 2 days and then lysed. The cell extracts from each assay were used to determine TGase GTP binding activities using an affinity labeling assay with radioactive GTP as outlined under "Experimental Procedures." Western blot analysis using a TGase antibody was utilized to assess the expression levels of this protein.
that enhance cell proliferation and resistance to cell death, and blocks the expression and/or activation of proteins that limit cell growth and stimulate apoptosis (26). Given that EGF inhibits RA-induced TGase expression through activation of the Ras-ERK signaling cascade, it seemed possible that increased TGase expression and activation would in turn antagonize EGF-mediated signaling activities. To examine this possibility, we utilized NIH3T3 cells that stably overexpressed wild-type (WT) TGase or a transamidation-defective TGase mutant, TGase(C277V) (Fig. 7A). Because a role for TGase in endocytosis had been proposed previously (3), coupled with fact that down-regulation of activated EGF receptors occurs primarily via endocytosis, we first examined whether TGase might influence the half-life of the activated EGF receptor. As an assay for EGF receptor signaling activity, the magnitude and duration of ERK activation were determined following EGF stimulation, using an antibody that specifically recognizes the phosphorylated or activated form of ERK. EGF-stimulation of control cells resulted in a rapid induction of ERK activity that was maximal at 5 min and was reduced to basal levels by 20 min (Fig. 7A,  vector). The kinetics of EGF-stimulated ERK activation in cells stably expressing WT TGase or TGase(C277V) were unchanged from control cells (Fig. 7A, WT TGase and TGase(C277V)). In addition, the magnitude of ERK activity detected in cells expressing TGase was indistinguishable from control cells suggesting that TGase does not alter the signaling potential of the EGF receptor.
We also examined whether EGF-mediated cell growth and/or resistance to cell death was compromised by the expression of TGase. Consistent with our earlier observation (Fig. 5A) that TGase serves as an anti-apoptotic factor, TGase expression protected mouse fibroblasts from apoptosis because of serum starvation, just as effectively as EGF (Fig. 7B, vector and WT TGase). EGF treatment of cells overexpressing either WT or transamidation-defective TGase appeared to increase cell survival in an additive fashion, making these cells more resistant to serum deprivation-mediated cell death compared with that obtained upon EGF stimulation or TGase expression alone (Fig. 7B, vector and WT TGase). Comparing the proliferation rates in low serum for control cells and cells expressing WT TGase and the transamidation-defective TGase mutant also revealed that overexpression of WT TGase provided a growth advantage. Whereas control cells or cells expressing the transamidation-defective TGase mutant displayed limited growth, cells expressing WT TGase grew about 2.5-fold better than control cells (Fig. 7C). The addition of EGF to each of the stable cell lines caused a parallel increase in cell proliferation over the basal growth rate of each cell line, indicating that TGase did not negatively regulate EGF-mediated growth. DISCUSSION The goal of this study was to understand better how RAinduced TGase expression is regulated in cells and, once expressed, what effect TGase has on cellular processes. In mouse fibroblasts, RA enhances both the expression and activation of the GTP binding and transamidation activities of the TGase (7). It has been well documented that RA-induced increases in TGase expression are mediated through the up-regulation of the transcriptional activities of the RARs (2). Our results provide some interesting new insights into the underlying regulatory mechanisms for RA-induced TGase expression. For example, although we find that the activity of the anti-apoptotic factor PI3K is required for RA-induced TGase expression, EGF treatment of cells, which activates PI3K, causes an inhibition of the RA-mediated up-regulation of TGase. The inhibitory signal provided by EGF results from the stimulation of the Ras-ERK cascade, probably explaining an earlier report (39) that showed dominant-active Ras inhibited RA-mediated TGase expression. However, it is not yet known how the activation of PI3K nor the stimulation of the Ras-ERK cascade influences the induction of TGase expression by RA, but we speculate that these pathways lead to the activation of transcriptional regulators that act in tandem with the RARs to FIG. 7. Exogenous TGase expression does not compromise EGF-mediated effects in NIH3T3 cells. Cells stably expressing wildtype TGase (WT TGase), a transamidation-deficient TGase mutant (C277V TGase), or vector only (vector) were used in each of the following assays. A, cells were grown to near confluence and serum-starved for 1 day. The cultures were then stimulated with 100 ng/ml EGF for the times indicated and lysed. From the whole cell extracts, the activation state of ERK was measured via Western blot analysis with a phosphospecific ERK antibody (P-ERK), and the expression level of TGase was determined by Western blot analysis with an antibody for TGase. B, stable cell lines were exposed to serum-free medium Ϯ 100 ng/ml EGF for 2 days, and the cells were then collected and scored for apoptosis as described under "Experimental Procedures." C, cells were seeded at 1 ϫ 10 5 cells/well and grown in low serum Ϯ 100 ng/ml EGF for the times indicated and then counted. enhance or inhibit transcription of the TGase gene. Support for this idea comes from the identification of several putative transcription factor-binding sites in the TGase promoter region (40,41), including response elements for AP-1 as well as the nuclear factor B. The activities of both AP-1 and nuclear factor B have been shown to be up-regulated by PI3K (42,43), raising the possibility that the requirement of PI3K activity for RA-stimulated TGase expression may involve the ability of PI3K to activate one or both of these transcription factors. Additionally, it is worth noting that the ability of dominantactive Ras to abrogate RA-induced TGase expression reflects a specific signaling effect rather than a general characteristic of cellular transformation, as the expression of a transforming mutant of Cdc42 (Cdc42(F28L)) in NIH3T3 cells did not limit RA-induced TGase expression. Moreover, some human tumor cell lines (U87 and MDAMB-231) actually exhibited constitutive TGase GTP binding activity.
Thus, it seems unlikely that the negative regulatory effects elicited by EGF on RA-mediated TGase expression are the outcome of a mitogenic signaling pathway (e.g. EGF receptorsignaling) specifically targeting TGase for down-regulation. If anything, the mitogenic and survival activities exhibited by TGase under a variety of conditions would be expected to complement rather than antagonize mitogenic signaling activities. Rather, it seems more likely that the EGF receptor-Ras-ERK pathway is directed in a more general sense at RA-induced gene expression. It has been well established that RA induces the expression of genes with anti-proliferative activity that supports cell cycle arrest and cellular differentiation (4,5,28). The targeting of RA-induced gene expression by EGF would probably ensure that such anti-proliferative activities are not activated by differentiation factors like RA. TGase, which we believe is activated by RA so as to serve as an "insurance factor" against apoptosis, would not be needed by EGF signaling events that provide anti-apoptotic activity and so can be shut down with the other RA-induced gene products.
Numerous studies have associated increased TGase expression and activation in cells with the promotion of apoptosis or cell differentiation. For example, exogenous TGase expression sensitized the neuroblastoma cell line SK-N-BE to apoptosis (9), whereas blocking TGase expression in the promonocytic cell line U937, using antisense technology, reduced the effectiveness of apoptotic stimuli to induce cell death (10). Yet the results presented here argue for a role for TGase in protecting NIH3T3 cells from serum deprivation-mediated apoptosis and promoting cell proliferation. Interestingly, it was recently reported (44) that thymocytes isolated from TGase Ϫ/Ϫ mice were more susceptible to apoptosis, indicating that the protective effect of TGase was not limited to just mouse fibroblasts. How TGase can have opposing cellular effects depending on the cell type is not clear but is of particular interest to us. It may be that TGase elicits distinct cellular outcomes in different cell lineages. The precedence for proteins having opposing functions in different types of cells has been noted with c-Jun N-terminal kinase (JNK) and Myc (45)(46)(47)(48). In the case of JNK, persistent JNK activation induced apoptosis in cells of neuronal origin (45), whereas in gliomas continuous JNK signaling has been associated with transformation (46). Experiments are currently being conducted to determine whether or not aberrant TGase activity promotes or inhibits apoptosis in cells of different lineages. We are especially interested in assessing the importance of the loss of regulatable TGase GTP binding activity, exhibited by tumor cell lines U87 and MDAMB231, for cell viability.