Introduction

TC21 was initially isolated during the search for ras-related genes using a polymerase chain reaction approach with degenerated oligonucleotides (Drivas et al., 1990). Sequence analysis of the TC21 protein indicates that this GTPase is closely related to members of the R-Ras subfamily, including the Drosophila Ras-2 and mammalian R-Ras and R-Ras-3. Unlike its related genes, TC21 shows high levels of transforming activity when mutations analogous to those present in ras oncogenes are introduced into mammalian cells (Chan et al., 1994; Graham et al., 1994). Perhaps more importantly, oncogenic mutations in the TC21 gene are occasionally found in tumor cell lines (Chan et al., 1994; Huang et al., 1995). The TC21 proto-oncogene is also frequently amplified in cell lines derived from breast cancer (Clark et al., 1996), further suggesting a possible implication of this GTPase in tumorigenesis.

In spite of the important role of TC21 in proliferation, many of the upstream and downstream elements of its signal transduction pathway remain to be identified. Based on the similarity in the effector domains of Ras and TC21, it was assumed that both GTPases signal through very similar, if not identical, signal transduction pathways. Indeed, initial studies showed that the TC21 oncoprotein stimulates mitogen-activated protein kinases (MAPK) and transcription from Ras-responsive promoter elements, such as AP-1 and Est-2 (Graham et al., 1994). It has been also demonstrated biochemically that TC21 is susceptible to the action of Ras GAPs (Graham et al., 1996). Moreover, activated TC21 binds to Ral GDS (Lopez-Barahona et al., 1996), a downstream element of the Ras pathway (Marshall, 1996). Other lines of evidence suggest that TC21 and Ras may also utilize distinct signaling and regulatory elements. Thus, it has been shown that while Ras is modified by farnesylation, TC21 can undergo both farnesylation and geranylation (Carboni et al., 1995). As a consequence, TC21 is not susceptible to inhibition by the farnesyl transferase inhibitors that block the growth of cancer cells harboring Ras mutations (Carboni et al., 1995). Moreover, cell fractionation experiments have suggested that TC21 and Ras are localized in distinct membrane regions (Lisanti et al., 1994). Finally, it has been reported that TC21 activated MAPK via a Raf-independent mechanism (Graham et al., 1996). It is possible therefore that the intracellular signals induced by TC21 and Ras may be overlapping but not identical.

In this work, we have conducted a series of experiments both at the biochemical and cellular level to identify the regulatory molecules lying both upstream and downstream of this oncoprotein. This research has revealed that TC21 shares most of the Ras signaling elements, including guanosine nucleotide exchange factors (GEF) and serine/threonine cascades such as those leading to the activation of MAPK and the c-Jun N-terminal kinase (JNK). In contrast to previous reports, we also provide evidence that the activation of the MAPK cascade is mediated by the physical interaction of TC21 with Raf. In addition, we have also compared the proto- and oncogenic versions of TC21 and Ras in their ability to promote cellular transformation and differentiation in rodent fibroblasts and the pheochromocytoma PC12 cell line. These results have demonstrated that both the wild type and the oncogenic TC21 proteins are capable of sustaining constitutive signals leading to cellular transformation and differentiation. Taken together, our results indicate that TC21 participates in the determination of proliferative and differentiation decisions that are very similar to those elicited by the Ras GTPase.

Results

Implication of TC21 in cell transformation and differentiation

To examine the role of TC21 proteins in cell proliferation, we first determined the transforming activity of both the wild type and the oncogenic TC21 protein in rodent fibroblasts using a focus formation assay. These experiments also included transfections with the H-ras oncogene as positive control. In agreement with previous reports (Chan et al., 1994; Graham et al., 1994), it was found that the TC21G23V mutant was as transforming as the human H-ras oncogene (Figure 1a). The TC21 proto-oncogene was also transforming, although several orders of magnitude less than its oncogenic counterpart (Figure 1a and b). The transformed phenotype of cells expressing wild type TC21 was further verified by measuring their ability to grow in an anchorage-independent manner. To this end, several foci of transformed cells were picked randomly, expanded, and cultured in semi-solid agar plates. Cells derived from both TC21- and TC21G23V-induced foci were capable of developing colonies under those conditions, a result that confirms the transforming activity of the wild type version of TC21 (data not shown). Cells overexpressing wild type TC21 protein displayed a moderate growth advantage in confluent cultures when compared with the parental cell line (Figure 1c). In addition, they showed a marked morphological change characterized by the appearance of cells with a polygonal shape that frequently contained long cytoplasmic processes (see Figure 5b). This phenotype was further accentuated in cells expressing the TC21G23V protein, as these cells achieved very high cell densities after reaching confluency (Figure 1c). Moreover, they also displayed a more dramatic morphological change characterized by the presence of fusiform cells with rounded, highly refringent cell bodies (see Figure 5c). The expression of TC21 and TC21G23V was similar in these cells, indicating that the phenotypic differences observed are due to the different activities of these proteins (Figure 1d).

Figure 1
figure 1

(a) Transforming potential of linearized DNAs encoding wild type and oncogenic TC21 proteins. (b) Giemsa staining of a representative transfection experiment conducted in the presence of genomic DNA either alone (Genomic) or in combination with linearized plasmids containing the wild type (1 μg) and oncogenic (0.5 ng) versions of TC21. (c) Growth curves of parental (circles), TC21- (squares) and TC21G23V-transformed (triangles) cells. (d) Expression of TC21 proteins in NIH3T3 cells. Equal amounts of total cellular lysates from parental cells (None) and cell clones transformed by wild type and oncogenic TC21 were submitted to anti-TC21 and anti-actin immunoblots. (e) PC12 cells were transfected with vectors (4 μg each) containing the neo selection marker (Neo) alone or in combination with wild type and oncogenic TC21 and H-Ras proteins. Alternatively, PC12 were transfected with the neo-containing vector plus a mammalian a expression plasmid encoding the mouse vav oncogene (pJC12) lacking neo sequences. After transfection, cells were split, and differentiation scored after 20 days of culture

Full size image
Figure 5
figure 2

Morphology of parental (a) and transformed NIH3T3 cells expressing wild type TC21 (b), TC21G23V (c), wild type TC21 plus cytosolic Ras GRF (d), the farnesylated catalytic domain of rat Ras GRF (d), and the mouse Vav oncoprotein (e). All pictures were taken at the same microscope magnification. Scale bar represents 250 μm

Full size image

The biological activity of TC21 and TC21G23V proteins was also studied in PC12 cells, a pheochromocytoma cell line that undergoes differentiation upon the introduction of elements of the Ras pathway (Bar-Sagi and Feramisco, 1985). To this end, these cells were transfected with mammalian expression vectors containing either the TC21, TC21G23V, H-Ras, or H-RasG12V. As a negative control, we made parallel transfections with a mammalian expression vector encoding mouse vav, an oncogene whose signal transduction pathway is independent of Ras function (Bustelo et al., 1994; Crespo et al., 1997). As illustrated in Figure 1e, the oncogenic forms of TC21 and H-ras were capable of inducing neurite formation in approximately half of the G418 resistant colonies 20 days after transfection. Interestingly, wild type TC21 also induced neurite outgrowth, although at much lower levels than the oncogenic forms of H-ras and TC21 (Figure 1e). In contrast, no significant neurite formation was observed when PC12 cells were transfected with the H-ras proto-oncogene, the pSV2neo, and the mouse vav oncogene (Figure 1e). Taken together, these results indicate that TC21 and Ras proteins induce similar phenotypic effects when expressed in distinct mammalian cell types.

TC21 shows high levels of intrinsic guanosine nucleotide exchange

The activation of GTPases during signal transduction processes requires the exchange of GDP by GTP molecules, a step catalyzed by enzymes known as guanosine nucleotide exchange factors. As a prior step to identify the upstream GEFs involved on TC21 activation, we first evaluated the intrinsic nucleotide exchange rate of this protein in vitro. To this end, a GST-TC21 fusion protein purified from baculovirus-infected Sf9 cells was loaded at 32°C with [3H]GDP and then incubated in the presence of unlabeled GTP (100 μM) for the indicated periods of time. Surprisingly, the GST-TC21 protein displayed a very rapid release of [3H]GTP in the absence of any exchange factor, showing a nearly complete loss of the radiolabeled nucleotide between 30 – 60 min (Figure 2a). This rapid exchange of nucleotides was comparable to that observed with H-Ras in the presence of the catalytic domain of Ras GRF (Shou et al., 1992) (Figure 3a). In contrast, H-Ras protein showed negligible exchange rates under the same conditions (Figure 2a). To further characterize this activity, we conducted [3H]GDP releasing assays in the absence of nucleotides or, alternatively, in the presence of either GTP or GDP. We also investigated whether this effect could be due to a high instability of the nucleotide binding site by testing the intrinsic exchange rates of TC21 at different concentrations of MgCl2. As shown in Figure 2b, a rapid release of nucleotides was observed regardless of the nucleotide and the concentration of MgCl2 present in the reaction. However, no release of bound GDP was observed in the absence of competing nucleotide (Figure 2b), demonstrating that the GDP release found on TC21 is due to the rapid exchange of guanosine nucleotides occurring in this GTPase. Titration experiments indicated that the intrinsic nucleotide exchange of TC21 occurred at very low concentrations of unlabeled nucleotide (1 – 5 μM) (Figure 2c and d) and was temperature-dependent (Figure 2d). Taken together, these results indicate that TC21, unlike H-Ras, has a very rapid turnover rate of the associated nucleotides in vitro.

Figure 2
figure 3

(a) Wild type TC21 (circles) and H-Ras (squares) were loaded with [3H]GDP and then incubated at 32°C in the presence of cold GTP for the indicated periods of time. At each time-point, nucleotide exchange rates were determined using a filter immobilization assays (Schuebel et al., 1998). (b) Nucleotide and MgCl2 dependency for TC21 intrinsic nucleotide exchange. [3H]GDP-loaded TC21 was incubated with no nucleotides (No Adds), GTP, and GDP (100 μM each) with either 1 mM (closed boxes) or 25 mM (open boxes) MgCl2. Exchange rates were determined after a 45 min incubation at 32°C. (c) Intrinsic nucleotide exchange of TC21 in the presence of increasing concentrations of GTP. [3H]DGP-loaded TC21 was submitted for 45 min to exchange assays in the presence of the indicated concentrations of cold GTP. (d) Nucleotide exchange of TC21 is temperature-dependent. [3H]GDP-loaded TC21 was submitted to exchange assays for 45 min in the presence of the indicated concentrations of cold GTP at 32°C (triangles), 25°C (squares) and 4°C (circles)

Full size image

Ras, but not Rho GEFs, are specific activators of TC21 in vitro

Since TC21 shows similarity with H-Ras in the region involved in the interaction with upstream regulators (Boriack-Sjodin et al., 1998), we decided to investigate whether the nucleotide exchange of TC21 could be enhanced by Ras GEFs. To this end, we determined the exchange rates of [3H]GDP-loaded TC21 and H-Ras in the presence or absence of a GST fusion protein containing the catalytic domain of Ras GRF, a Cdc25-related exchange factor specific for Ras and R-Ras (Gotoh et al., 1997; Shou et al., 1992). These incubations were performed at 25°C in order to reduce the high intrinsic exchange rates of TC21 (see Figure 2d). As shown in Figure 3a (left panel), Ras GRF was capable of stimulating the exchange of nucleotides on TC21 even more efficiently than on H-Ras. Ras GRF promoted also the rapid incorporation of free [35S]GTP into TC21 molecules previously loaded with unlabeled GDP molecules (Figure 3b). Titration experiments showed that the nucleotide exchange was detected at molar ratios as low as at 0.6 – 1.2 pmol of Ras GRF/60 pmol of GTPase (Figure 3c), indicating that the action of Ras GRF on TC21 is catalytic in nature. To further demonstrate the specificity of these reactions, we conducted in vitro exchange reactions with the human Dbl oncoprotein, a GEF whose catalytic activity is directed towards Cdc42 and RhoA GTPases (Cerione and Zheng, 1996). [3H]GDP release assays demonstrated that this exchange factor was not active towards TC21 (Figure 3d, left panel). As a positive control, Dbl promoted nucleotide exchange on Cdc42 (Figure 3d, right panel), confirming that this GEF was active in our assays. Thus, these results indicate that the activation of TC21 is limited to members of the Cdc25 family.

Figure 3
figure 4

(a) Ras GRF catalyzes nucleotide exchange on TC21 and H-Ras. [3H]GDP-loaded TC21 (left panel) and H-ras (right panel) were incubated in the absence (circles) or presence (squares) of GST-Ras GRF for the indicated periods of time. After each time-point, aliquots were taken in duplicate and nucleotide exchange determined by a filter immobilization technique. (b) Ras GRF facilitates the incorporation of [35S]GTP into GDP-bound TC21. TC21 was preloaded with unlabeled GDP and then incubated for the indicated periods of time with 5 μM [35S]GTP-γS in the absence (circles) or presence (squares) of GST-Ras GRF. (c) The nucleotide exchange of Ras GRF onto TC21 is concentration-dependent. [3H]GDP-loaded TC21 was incubated either alone or in combination with the indicated amounts of GST-Ras GRF protein. Values represent the percentage of GDP bound to TC21 in the presence of Ras GRF relative to the amount of GDP bound to TC21 when incubated alone for 45 min (value=100%). (d) Lack of exchange activity of Dbl towards TC21 protein. [3H]GDP-loaded TC21 (left panel) and Cdc42 (right panel) were incubated in the absence (circles) or presence (squares) of the Dbl oncoprotein for the indicated periods of time. At each time-point, the exchange activity was determined as indicated in (a)

Full size image

Ras GRF promotes activation of TC21 in vivo

To demonstrate the physiological significance of the TC21/Ras GRF interaction, we first investigated whether Ras GRF could induce the exchange of guanosine nucleotides on TC21 in vivo. Since there are not available antibodies to TC21 capable of blocking its GTPase activity, we decided to follow indirectly the activation of TC21 in vivo by measuring the levels of 32P-labeled GDP bound to TC21 in COS-7 cells after short incubations with [32P]orthophosphate (Laudanna et al., 1996). These experiments revealed that TC21 had very high levels of nucleotide exchange even in the absence of exchange factor (Figure 4a, lane 1). This can be attributed to the high levels of intrinsic exchange found in vitro (see Figure 2). This exchange was further increased upon co-expression of TC21 with Ras GRF in COS-7 cells (Figure 4a, lane 2). In agreement with our previous in vitro observations, we found that the intrinsic rates of nucleotide exchange found for H-Ras were significantly lower than those observed for TC21 under this in vivo conditions (Figure 4a, lane 3). As expected, co-transfection of Ras GRF led to the activation of nucleotide exchange on H-Ras (Figure 4a, lane 4). Western blot experiments demonstrated that TC21 and H-Ras were expressed at comparable levels in those experimental conditions (data not shown).

Figure 4
figure 5

(a) Activation of TC21 and Ras by Ras GRF in vivo. COS-7 cells were transfected with wild type AU5-TC21 (lane 1), wild type AU5-TC21 plus full length Ras GRF (lane 2), AU5-H-Ras (lane 3), and AU5-H-Ras plus full length Ras GRF (lane 4). After 48 h, cells were starved overnight, labeled with 32P-orthophosphate for 20 min, and the incorporation of labeled guanosine nucleotides onto TC21 and H-Ras determined by thin layer chromatography. The mobility of 32P-labeled GDP and the origin (OR) of the chromatography is indicated by arrows. (b) Cooperativity of Ras GRF and TC21 in cellular transformation. NIH3T3 cells were transfected with the indicated amounts of linearized mammalian expression vectors containing the catalytic domain of rat Ras GRF (without CAAX box) and the TC21 proto-oncogene. (c) Activation of TC21 and Ras in vivo by Ras GRF. Cells were transfected with the catalytic domain of rat Ras GRF (without CAAX box, 1 μg), the TC21 proto-oncogene (100 ng), or the H-ras proto-oncogene (500 ng)

Full size image

It has been previously shown that the co-expression of wild type GTPases with their specific GEFs leads to a synergism in cellular transformation (Bustelo et al., 1994; Schuebel et al., 1998). We investigated therefore whether the co-expression of the catalytic domain of Ras GRF and the wild type TC21 protein could cooperate in focus formation assays. Since the constitutively-active form of Ras GRF (farnesylated at the C-terminus) shows high levels of transforming activity (Bustelo et al., 1994), we used in these experiments a mammalian expression vector encoding the cytosolic form of Ras GRF. We have previously demonstrated that this protein lacks transforming activity unless co-expressed with wild type Ras (Bustelo et al., 1994). As shown in Figure 4b, the cytosolic form of the Ras GRF catalytic domain effectively synergized with TC21, leading to a concentration-dependent increase in the number of foci over those obtained when TC21 is transfected alone. This synergism was similar to that observed between Ras GRF and wild type H-Ras (Figure 4c). As shown in Figure 5, the TC21/Ras GRF-transformed cell lines (Figure 5d) exhibited a morphological transformation identical to those displayed by either TC21G23V – (Figure 5c) or Ras GRF-CAAX-transformed cells (Figure 5e). By contrast, TC21/Ras GRF-transformed cells were quite different in morphology from parental, untransformed cells (Figure 5a) and from fibroblasts transformed by the vav oncogene (Figure 5f), a Rac-1-specific GEF (Bustelo, 1996; Crespo et al., 1997). Collectively, these results demonstrate that Ras GRF is a bona fide exchange factor for TC21.

c-Raf-1 is a downstream element of the TC21 pathway

Activation of Ras leads to its physical association with Raf proteins, the activation of this serine/threonine kinase, and the subsequent stimulation of MEK (MAPKK) and ERK/MAPK (Marshall, 1996). Like Ras, it has been shown by others that TC21 induces the activation of the terminal elements of this route (Graham et al., 1994). However, previous reports have suggested that such activation is independent of Raf function (Graham et al., 1996). In order to identify the MAPKKK involved in the activation of the MAPK in TC21-transformed cells, we decided to re-evaluate the possible functional interaction between TC21 and the c-raf-1 proto-oncogene product. This was done by using a TC21/Raf co-immunoprecipitation assay in insect cells, a technique used before to demonstrate the interaction between Ras/Raf proteins (Morrison, 1995). Under these conditions, we found that the co-expression of c-Raf-1 and a GST-TC21 protein in Sf9 cells leads to the efficient association of these two proteins (Figure 6a, upper panel). This interaction occurred preferentially when TC21 was in the GTP-bound state (Figure 6b), indicating that c-Raf-1 binds to TC21 with the same structural constraints previously reported for the Ras/Raf interaction.

Figure 6
figure 6

(a) Association of TC21 with c-Raf-1 in baculovirus-infected Sf9 cells. Cell lysates derived from Sf9 cells infected with the indicated combinations of baculovirus were incubated with glutathione-coated beads (pull-down) or with anti-c-Raf-1 antibodies (IP). After binding and extensive washes, samples were submitted to anti-c-Raf-1 immunoblot analysis. Lower panel, expression of c-Raf-1 protein in each infection, as determined by anti-Raf immunoblot of c-Raf-1 immunoprecipitates. (b) Binding of c-Raf-1 to TC21 is nucleotide-dependent. Total cellular lysates from c-Raf-1 infected Sf9 cells were incubated with either GST or GST-TC21 in the presence of the indicated nucleotides (200 μM). Binding of c-Raf-1 to these proteins was determined by anti-c-Raf-1 immunoblot. (c) Upper panel, kinase activity of c-Raf-1 in parental, TC21G23V- and H-RasG12V-transformed cells. c-Raf-1 was immunoprecipitated from total cellular lysates of NIH3T3 and TC21 and Ras-transformed cells and its activity determined as described in Experimental procedures. Lower panel, levels of endogenous c-Raf-1 in the indicated cell lines. Equal amounts of cellular lysates from the indicated cell lines were immunoprecipitated with anti-c-Raf-1 antibodies and subjected to anti-c-Raf-1 immunoblot

Full size image

We also conducted experiments in vitro to test whether the TC21 oncoprotein activates Raf proteins in mammalian cells. To address this point, endogenous c-Raf-1 proteins were purified from quiescent parental, TC21G23V-, and RasG12V-transformed NIH3T3 cells by immunoprecipitation with specific anti-c-Raf-1 antibodies, and their kinase activities evaluated by in vitro kinase reactions using an inactive form of MEK as substrate. As shown in Figure 6c (upper panel), c-Raf-1 proteins displayed high levels of activation in both TC21G23V- and H-RasG12V-transformed cells. Immunoblot analysis confirmed that these cell lines expressed comparable levels of endogenous c-Raf-1 protein (Figure 6c, lower panel).

The coexpression of Ras and Raf proteins leads to transformation of NIH3T3 cells, (Cuadrado et al., 1993). Therefore, we reasoned that if TC21 is also an activator of c-Raf-1 in vivo, their co-expression should result in a more potent oncogenic response. To test this hypothesis, we performed focus formation assays in NIH3T3 cells using suboptimal concentrations of the TC21 expression vector either alone or with increasing concentrations of a mammalian expression vector encoding the c-raf-1 proto-oncogene (pXRB38) (Bustelo et al., 1994). As shown in Figure 7a, the overexpression of c-Raf-1 led to an increase in the transforming potential of TC21 that was already detectable at low concentrations of the co-transfected c-raf-1-encoding vector (Figure 7a). This effect appears to be only specific for TC21 and Ras proteins, as we could not detect such synergistic response upon co-transfection of wild type c-Raf-1 with other wild type GTPases such as Rac-1, RhoA, and Cdc42 (Figure 7b). Taken together, these results indicate that c-Raf-1 participates in the signal transduction pathway regulated by TC21.

Figure 7
figure 7

(a) Cooperativity of TC21 and c-Raf-1 in cellular transformation. NIH3T3 cells were transfected with 100 ng of a linearized plasmid encoding wild type TC21 either alone or in combination with the indicated amounts of a c-raf-1-expression vector. Foci were determined 15 days post-transfection. (b) Cooperativity of c-Raf-1 with GTPases of the Ras superfamily. Values represent the number of foci with each GTPase obtained in the absence (open bars) and presence (closed bars) of the co-transfected c-raf-1-encoding vector

Full size image

Activation of the JNK cascade by TC21 oncoprotein

It has been shown recently that Ras activation plays a critical role in the activation of other members of the MAPK family, including JNK. This effect appears to be mediated by the activation of members of the Rho subfamily of GTP-binding proteins (Coso et al., 1995; Minden et al., 1995). To determine whether TC21 could activate this signal transduction pathway, we performed in vitro kinase reactions with the endogenous JNK present in parental and NIH3T3 cells overexpressing either wild type or oncogenic TC21 proteins. As positive control, we used the endogenous JNK immunoprecipitated from cells transformed by the mouse Vav oncoprotein, a known activator of the JNK cascade (Bustelo, 1996; Crespo et al., 1996). These assays demonstrated that JNK was highly active in cells expressing the TC21G23V protein (Figure 8, left panel). In contrast, this kinase was significantly less active in cells transformed by the wild type TC21 protein (n=4 experiments) (Figure 8, left panel). The levels of JNK activation induced by TC21G23V were as high as those stimulated by the Vav oncoprotein (Figure 8, right panel). Similar results were obtained in exponentially-growing cells (data not shown). The protein levels of this kinase were similar in all the cell lines used (Figure 8, lower panels). Collectively, these results indicated that TC21, like Ras, is a generic activator of MAPK cascades.

figure 8

Figure 8

Discussion

We have undertaken the present experiments in order to progress into the functional characterization of TC21 and to investigate the degree of overlap of its signaling elements with those used by Ras. To this end, we first analysed the functional consequences of TC21 expression in rodent fibroblasts and PC12, two well-known cellular systems that respond to Ras-derived signals by undergoing morphological transformation and neuronal-like differentiation, respectively (Bar-Sagi and Feramisco, 1985; Santos et al., 1982). The results obtained with both systems demonstrated that TC21 induces similar functional responses at the cellular level as H-Ras. These results confirm previous observations on the transforming activity of TC21G23V (Chan et al., 1994; Graham et al., 1994) and demonstrate a lower, but significant, basal biological activity for the wild type TC21 both in rodent fibroblast and pheochromocytoma cells.

The similarity observed between Ras and TC21 at the cellular level is also seen at the biochemical level. Thus, we have shown that TC21 shares with Ras upstream regulators such as the exchange factor Ras GRF. Although we did not address in this study the structural basis for the TC21/Ras GRF interaction, the comparison of the amino acid sequence of TC21 and H-Ras indicates a high level of conservation in the phosphate-binding loop (87.5% identity), the Switch-2 (83.3% identity), and the Switch-1 (75% identity) regions, three domains important for the interaction of Ras with its specific GEFs (Boriack-Sjodin et al., 1998). Based on those sequence similarities, it is tempting to speculate that Ras GEFs will promote the exchange of nucleotides on TC21 following similar mechanisms to those used for Ras.

To date, the downstream protein kinases that act downstream of members of the R-Ras/TC21 subfamily have been poorly defined. Indeed, initial reports have shown that R-Ras can bind to c-Raf-1 and, consequently, stimulate MAPK activity and mediate c-Fos-mediated gene transcription (Cox et al., 1994). However, more recent reports have demonstrated that R-Ras lacks the ability to activate c-Raf-1, the MAPK cascade, and JNK (Marte et al., 1997). In the case of TC21, no reports are available regarding the activation of JNK. However, it has been proposed that the TC21 oncogene product activated MAPK in a c-Raf-1-independent manner (Graham et al., 1996). In contrast to those results, we have presented evidence here supporting a role for c-Raf-1 in the signaling cascade of TC21. Thus, we have demonstrated that TC21 can interact in a GTP-dependent manner with c-Raf-1. In good agreement with this physical association, we have also shown that TC21G23V-transformed cells display constitutive activation of c-Raf-1 in vivo. Moreover, we have shown that TC21 and c-Raf-1 cooperate in promoting cellular transformation, thus providing an independent confirmation for the participation of c-Raf-1 in TC21-mediated transformation. During the submission of this manuscript, Rosário et al. (1999) have reported the activation of c-Raf-1 and B-Raf by the expression of activated forms of TC21 in NIH3T3 cells, indicating that the activation of c-Raf-1 observed by us cannot be attributed to the use of a particular experimental protocol.

The utilization of a similar spectrum of protein kinases by Ras and TC21 is further emphasized by our demonstration that transformation by TC21 correlates with high levels of JNK activation. It seems therefore that, in spite of its closer similarity to R-Ras, TC21 behaves functionally more similarly to Ras than to its family counterpart. This may explain in part why TC21 displays much higher transforming activity than R-Ras in focus formation assays in spite of its structural relatedness.

An important distinction between Ras and TC21 pertains to their different rates of intrinsic nucleotide exchange rates at physiological temperatures. Since it is known that mutations that increase the nucleotide turn-over rates of Ras proteins are tumorigenic (Sigal et al., 1986), our results suggest that quiescent cells should have some inhibitory mechanism to downmodulate the high nucleotide exchange rate of TC21. One of these mechanisms could be the expression of TC21 only when it is required for specific biological purposes. For instance, it has been shown that TC21 is induced in bovine aortic endothelial cells during migration and after stimulation with αFGF, βFGF, and EGF (Kozian and Augustin, 1997). A different functional alternative to inhibit the intrinsic exchange rates of TC21 could be the use of GDP dissociation inhibitors (GDI). Similar factors have been described before for GTP-binding proteins with relatively high intrinsic exchange rates, such as Rho, or Cdc42 (Boguski and McCormick, 1993). The presence of similar factors for TC21 will allow the constitutive expression of TC21 in an inactive, GDP-bound state that will assure the rapid coupling of mitogenic and antigenic signals to the activation of TC21 and its downstream elements.

Our present results have also implications for general approaches commonly used in signal transduction. Thus, dominant negative mutants of Ras that inhibit the endogenous GEFs have been used extensively to prove the essential role of Ras in a broad variety of cellular responses. Our results showing that TC21 and Ras share the same GEFs indicate that the drastic effect of Ras dominant negative mutants observed in tissue culture experiments may be due to the dual interference of the Ras and TC21 pathways. It will be important therefore to develop monoclonal antibodies to TC21 that could block its GTPase activity in vivo to conduct careful analysis of the activation state of TC21 during the signal transduction pathways of mitogenic and antigenic receptors previously thought to work through the direct stimulation of the Ras pathway.

Materials and methods

Antibodies

Rabbit polyclonal antisera to GST-TC21 were provided by Dr M Barbacid (Centro Nacional de Biotecnología, Madrid, Spain). Antisera to c-Raf-1, Ras GRF and JNK-1 were from Santa Cruz Biotechnology. Antibodies to actin and the AU5-epitope were from Sigma and Babco, respectively.

Plasmids and cell transfections

Plasmids for focus formation assays included those containing the H-ras proto-oncogene (pbcN1) (Bustelo et al., 1994) and oncogene (pAL8) (Bustelo et al., 1994), TC21 proto-oncogene (pTC21) and oncogene (pTC21Val23) (Lopez-Barahona et al., 1996), the human c-raf-1 proto-oncogene (pXRB38) (Bustelo et al., 1994), the catalytic domain (residues 798 – 1244) of rat Ras GRF with (pXRB27) or without a C-terminal CAAX box (pXRB26) (Bustelo et al., 1994), and the mouse vav oncogene (pJC12) (Coppola et al., 1991). For in vivo exchange assays, AU5-tagged versions of wild type TC21 and H-Ras were cloned in the pCEFL vector. Vectors containing Rho subfamily members have been described elsewhere (Schuebel et al., 1998). Transfections of NIH3T3, PC12, and COS-7 cells were performed using the calcium-phosphate (van der Eb and Graham, 1980), liposome (FuGENE, Roche Molecular Biochemicals), and the DEAE-dextrane methods (Crespo et al., 1997), respectively.

In vitro kinase assays and GDP/GTP exchange techniques

Immunocomplex kinase assays were performed in the presence of either a MEK defective kinase (for c-Raf assays) or a home-made GST-ATF-2 fusion protein (for JNK-1 assays) according to established protocols (Coso et al., 1995; Suen et al., 1995). In vitro [3H]GDP release and [35S]GTP-γS incorporation assays were performed as indicated (Schuebel et al., 1998). For in vivo exchange assays, transfected COS-7 cells were cultured for 48 h in complete media, followed by an incubation in serum-free media for 12 h. Cells were then incubated in phosphate-free media supplemented with 500 μCi.ml−1 of [32P]orthophosphate (Amersham) for 20 min. Cells were then lysed and the GTP-binding proteins immunoprecipitated with anti-AU5 antibodies (Crespo et al., 1997). After extensive washes, the nucleotides remaining bound to the immunoprecipitated GTPases were released by heating and analysed by polyethyleneimine thin-layer chromatography (Crespo et al., 1997).