Involvement of Phosphatidylinositol 3-Kinase, but Not RalGDS, in TC21/R-Ras2-mediated Transformation*

Oncogenic Ras and activated forms of the Ras-related protein TC21/R-Ras2 share similar abilities to alter cell proliferation. However, in contrast to Ras, we found previously that TC21 fails to activate the Raf-1 serine/threonine kinase. Thus, TC21 must utilize non-Raf effectors to regulate cell function. In this study, we determined that TC21 interacts strongly with some (RalGDS, RGL, RGL2/Rlf, AF6, and the phosphatidylinositol 3-kinase (PI3K) catalytic subunit p110δ), and weakly with other Ras·GTP-binding proteins. In addition, library screening identified novel TC21-interacting proteins. We also determined that TC21, similar to Ras, mediates activation of phospholipase Cε. We then examined if RalGDS, a RalA guanine nucleotide exchange factor, or PI3K are effectors for TC21-mediated signaling and cell proliferation in murine fibroblasts. We found that overexpression of full-length RalGDS reduced the focus forming activity of activated TC21. Furthermore, expression of activated Ras, but not TC21, enhanced GTP loading on RalA. In fact, TC21 attenuated insulin-stimulated RalA·GTP formation. In contrast, like Ras, expression of activated TC21 resulted in membrane translocation and an increase in the PI3K-dependent phosphorylation of Akt, and inhibition of PI3K activity interfered with TC21 focus formation. Finally, unlike Ras, TC21 did not activate the Rac small GTPase, indicating that Ras may not activate Rac by PI3K. Taken together, these results suggest that PI3K, but not RalGDS, is an important mediator of cell proliferation by TC21.

Although members of the Ras family share significant sequence conservation with Ras proteins, only TC21/R-Ras2 shares significant biological properties with Ras (4 -7). Although R-Ras exhibits transforming properties in NIH 3T3 cells (8,9), it may regulate cellular processes (e.g. apoptosis and integrin-mediated cellular adhesion) distinct from those regulated by Ras (10,11). R-Ras3/M-Ras is expressed exclusively in the heart and brain and promotes neuronal cell survival, and although R-Ras3 has the ability to cause weak transformation of NIH 3T3 cells, it is only a weak activator of Ras-responsive transcriptional regulatory DNA elements. No transforming activity has been described for Ral or Rap proteins. Instead, Rap1A has been shown to antagonize Ras signaling and transformation (12), presumably by competing for Ras effector targets (13), whereas Ral may facilitate Ras transformation (14). Two studies have addressed the biological function of Rheb (15,16), and whether Rheb antagonizes or mediates growth transformation remains unclear. Recent studies (17) determined that Rin, a neural specific GTPase, failed to cause growth and morphological transformation of fibroblasts. Whether Rin functions to regulate neural specific biological properties is not known. Finally, although GTPase-defective Rit can promote growth transformation of fibroblasts, it is suspected to use effectors distinct from those used by Ras (17,18). Thus, to date, only constitutively activated mutants of TC21 have exhibited the same potent transforming activity as oncogenic Ras (4, 5, 7, 19 -21). Finally, aside from Ras, only mutated and constitutively activated mutants of TC21 have been found in human cancer cells (7,22).
Several differences in biochemical properties have been described that distinguish TC21 from Ras. First, TC21 shows a preferential expression pattern in kidney, placenta, and ovary (5), compared with the ubiquitous expression pattern of Ras. Second, subcellular fractionation experiments confirm that TC21 is localized to caveolin-rich membrane regions distinct from Ras (25). In addition, TC21 is a substrate for both farnesyltransferase and geranylgeranyltransferase, whereas Ras proteins are normally modified by farnesyltransferase only (26). Third, TC21 exhibits a higher intrinsic guanine nucleotide exchange rate (5,21) compared with Ha-Ras. Finally, although Ras binds to and activates Raf in vivo, we have demonstrated previously that TC21 fails to bind and activate full-length Raf-1, A-Raf, or B-Raf in vivo (5,6). However, other studies have shown that TC21 interacts with and activates Raf-1 and B-Raf, but not A-Raf, in vitro and in vivo (20,21). The basis for these different observations with Raf is currently not known.
Only one biological property has been described that distinguishes Ras from TC21. Activated Ras, but not TC21, caused primary rodent fibroblasts to undergo senescence (27,28). Because activation of the Raf/MEK/ERK pathway was determined to be necessary and sufficient to cause senescence, it is not surprising that TC21 failed to exhibit this activity (see below).
There is considerable biological, biochemical, and genetic evidence that the Raf-1 serine/threonine kinase is a key effector of Ras-mediated transformation (29). Therefore, the striking biological similarities of Ras and TC21 are surprising in light of our observation that TC21 fails to interact with and activate the three Raf kinases (5,6). Ras causes activation of Raf-1, in part, by promoting its translocation from the cytosol to the plasma membrane (30). Raf-1, in turn, phosphorylates and activates the mitogen-activated protein kinase (MAPK) kinases (MEK1 and MEK2) which then phosphorylate and activate the p42 and p44 MAPKs (also called ERK2 and ERK1, respectively). Although TC21 fails to activate Raf, ERK activation is up-regulated in TC21-transformed NIH 3T3 cells (4), possibly as a secondary consequence of growth transformation. Ras and TC21 also activate two other MAPK cascades that lead to activation of the Jun NH 2 -terminal kinases and p38, via Raf-independent pathways (6,31,32). Thus, despite its inability to activate Raf, TC21 still causes the same biological activities that are activated by Ras.
Although it is clear that Raf-1 is an important effector of Ras function, there is growing support for the involvement of multiple effectors in mediating Ras signaling and transformation. First, additional candidate Ras effectors have been identified (33). This spectrum of functionally diverse proteins includes GEFs for the Ras-related protein Ral (RalGDS, RGL, RGL2/Rlf, and RGL3) (34,35), the p110 catalytic subunit of phosphatidylinositol 3-kinase (PI3K) family proteins (36 -38), and other putative effectors, such as phospholipase C⑀ (PLC⑀), p120, and neurofibromin Ras GAPs, Rin1, and AF6 (39 -45). Like Raf-1, these proteins show preferential binding to the active (GTPbound) form of Ras, and interaction with Ras requires an intact Ras core effector domain (amino acids 32-40) (46). Second, effector domain mutants of Ras, which have lost the ability to bind to and activate Raf-1, retain the ability to cause transformation of NIH 3T3 fibroblasts and RIE-1 epithelial cells (47)(48)(49)(50) and to induce metastatic growth of NIH 3T3 cells in vivo (51). Thus, Raf-independent effector pathways alone are sufficient to mediate some aspects of Ras-mediated transformation. Finally, because Ras, but not Raf, causes transformation of RIE-1 and other epithelial cells (52), it is clear that Ras-mediated Raf activation alone is not sufficient for transformation of some cells. Therefore, Ras causes transformation by activating both Raf-dependent and Raf-independent pathways.
In the present study, we sought to evaluate the effectors that mediate TC21-induced transformation of NIH 3T3 fibroblasts. First, we utilized yeast two-hybrid binding analyses and determined that TC21 could interact with all Ras⅐GTP-binding proteins that were evaluated. Second, we employed yeast two-hybrid and expression library screening to search for TC21-interacting proteins and identified known or novel proteins as candidate effectors of TC21. Next, we determined that TC21, like Ras, activates PLC⑀. Then, we found that although TC21 interacts with RalGDS and RGL2, TC21 failed to activate RalA. Finally, we determined that TC21 activated PI3K and that activation of PI3K is essential for TC21-mediated transformation. Thus, our results implicate PI3K, but not RalGDS, as a key effector in TC21-mediated transformation of NIH 3T3 fibroblasts.

EXPERIMENTAL PROCEDURES
DNA Constructs-For yeast two-hybrid system assays, the cDNA sequences encoding Ha-Ras(12V) or TC21(23V) were inserted downstream of and in frame with the LexA DNA binding domain of pBTM116-ADE2, which contains a TRP1 gene for tryptophan prototrophy (53). In addition, the cDNA sequences encoding of Raf-1, Raf-1-RBD, RalGDS, PI3K p110␦ (RBD, amino acids 141-291), truncated AF6 (a generous gift of Linda Van Aelst), or Rin1 (a generous gift of John Colicelli) were inserted downstream of and in frame with a nuclear localized VP16 acidic activation domain in pVP16, which contains a LEU2 gene for leucine prototrophy (40,41,53).
For expression library screening, cDNA sequences encoding wild type or mutant (72L or 46A/72L) TC21 were ligated into the BamHI site of the pGEX2T-XL bacterial expression vector (a pGEX-2T derivative encoding the A kinase-target sequence RRASV 5Ј of the BamHI site) to express glutathione S-transferase (GST)-TC21 fusion proteins. pGEX2T-XL-Ha-Ras(12V) was a generous gift of Michael White (54).
pZIP-NeoSV(x)1 retrovirus mammalian expression vector (55) cDNA sequences encoding wild type or activating mutants of Ha-Ras (12V and 61L) or TC21 (23V and 72L) were used for focus formation and reporter analyses as described previously (4). pZIP-NeoSV(x)1 encoding full-length mouse RalGDS was constructed and used for focus formation analyses. pCGN-hygro mammalian expression vector encoding mouse RalGDS was constructed by insertion of the coding sequence downstream of and in frame with the hemagglutinin (HA) epitope tag and used for reporter analyses. pMCL-mek⌬ED encodes a constitutively activated and transforming mutant of MEK2 (a generous gift of Natalie Ahn) and was described previously (56). The Gal4-Elk-1 expression vector encodes a chimeric protein consisting of the DNA binding domain of the yeast Gal4 transcription factor followed by the NH 2 -terminal transactivation domain of the Elk-1 transcription factor (a generous gift of Richard Treisman) (57). The Gal 5 -Luc plasmid contains the luciferase gene under control of a minimal promoter containing tandem Gal4 DNA-binding sequences (58). pCGN-hygro mammalian expression vectors encoding HA epitope-tagged Ha-Ras(61L) or TC21(72L) were used for RalA, Rac, and Akt activation analyses described previously (6). For PLC⑀ activation assays, the 5Јuntranslated region of PLC⑀ was removed from pCMVscript-PLC⑀FLAG (43) to generate pCMV-rPLC⑀FC. TC21(72L) cDNA was inserted into pRSV to generate pRSV-TC21(72L). pRSV-Ha-Ras(61L) expression vector was generously provided by Johannes L. Bos.
Expression Library Screening-For preparation of radiolabeled probes, GST-TC21 or GST-Ha-Ras fusion proteins were purified using procedures described previously (61), and immobilized GST-fusion proteins were labeled with [␥-32 P]ATP as described previously (53). For expression library screening, a 16-day-old mouse embryo cDNA expression library that was cloned in the EXlox expression vector (Novagen) was used as described previously (53). All positive clones were amplified and purified to tertiary clones. All positive tertiary clones were then analyzed in the same fashion for binding to GST, GST-Ha-Ras(12V), GST-TC21(72L), and GST-TC21(46A,72L) (Table II).
PLC⑀ Activation Assays-Transfection of COS-7 cells and PLC activity assay were performed as described previously (43) with minor modifications. Briefly, cells were seeded into 24-well tissue culture plates, transfected, and labeled overnight with 2 Ci of [ 3 H]inositol in inositolfree medium (special order, HyClone) in the absence of serum. Protein expression was determined in parallel wells without radioactivity by Western blot analysis using a rabbit polyclonal anti-PLC⑀ antibody generated with a protein encompassing the RA1 domain of PLC⑀, anti-Ras (BD Transduction Laboratories), and anti-TC21 (Santa Cruz Biotechnology).
To determine PLC activity, transfected cells were incubated in 300 l of fresh serum-free, inositol-free medium with 20 mM lithium chloride for 60 min. The reaction was stopped with the addition of 0.6 ml of ice-cold 4.5% perchloric acid and maintained on ice for 15 min. Samples were neutralized with 0.5 M KOH, 9 mM Borax, and the precipitated potassium perchlorate was pelleted by centrifugation. The supernatant was removed, and 5 ml of ice-cold water was added to the sample. Total inositol phosphates were separated by column chromatography using AG 1-X8 200 -400 mesh, formate form. Inositol and glycerophosphoinositol were eluted by 1 ϫ 7.5 ml of 60 mM ammonium formate, 5 mM sodium tetraborate. Total inositol phosphates were then eluted with 2.5 ml of 1.2 M ammonium formate, 0.1 M formic acid and quantitated by liquid scintillation counting. Paired Student's t test was performed, and a value of p Ͻ 0.05 was considered significant.
Elk-1 Reporter Assays-NIH 3T3 cells were transiently co-transfected with HA epitope-tagged expression constructs (pCGN-hygro) of TC21(72L), Ha-Ras (61L), or pMCL-mek⌬ED plasmid DNA and Gal4-Elk-1, Gal4-Luc, and 0.1-1.0 g of plasmid DNA encoding full-length RalGDS. Three hours post-transfection, cells were transferred to DMEM supplemented with 0.5% calf serum. 24 hours post-transfection, cell lysates were prepared as described by the manufacturer, and the luciferase activity was assayed in a luminometer, as described previously (4,58). GTPase expression was equal in each sample as determined by Western blot analysis using anti-HA antibody (Babco).
Focus Formation Transformation Assays-To determine whether RalGDS could modulate activated TC21 focus forming activity (Fig. 3), NIH 3T3 cells were co-transfected with 100 ng of TC21 (72LV), Ha-Ras (61L), or MEK⌬ED, and 0.1-1.0 g of plasmid DNA encoding fulllength RalGDS or empty vector. Cells were then maintained in DMEM supplemented with 10% calf serum. The appearance of transformed foci of cells was quantitated 14 -16 days post-transfection.
To examine the necessity of PI3K in TC21-mediated transformation (Fig. 7), NIH 3T3 cells were either transfected with pZIP-tc21(72L) and cultured in the absence or presence of 10 M LY294002 (Calbiochem) or co-transfected with an excess of pSG5 empty vector or pSG5-⌬p85 (a generous gift of Julian Downward). For the samples treated with PI3K chemical inhibitor, cells were maintained every other day in DMEM supplemented with 10% calf serum plus 10 M LY294002 (ϩ) or the equivalent volume of Me 2 SO (Ϫ). For the samples co-transfected with pSG5 or pSG5-⌬p85, cells were maintained in DMEM supplemented with 10% calf serum. The appearance of transformed foci of cells was quantitated 14 -16 days post-transfection.
RalA and Rac Activation Assays-To assess activation of RalA or Rac and GST-RalBD or GST-PAK, respectively, fusion protein probe was used to isolate specifically cellular GTP-bound RalA or Rac, respectively. GST fusion protein probe was expressed in and isolated from DH5␣ Escherichia coli transformed with pGEX-KG-RalBD (a generous gift of Doug Andres) or pGEX-2N-PAK (amino acids 70 -132) (a generous gift of Bruce Mayer) using glutathione-agarose (Sigma) as described previously (62). Glutathione-agarose-bound GST-RalBD or GST-PAK was equilibrated with Nonidet P-40 lysis buffer (see below).
NIH 3T3 cells were transiently transfected in serum-free medium with HA epitope-tagged expression constructs (pCGN-hygro) of TC21(72L) and Ha-Ras(61L). Three hours post-transfection, cells were transferred to DMEM supplemented with 0.5% calf serum for ϳ20 h. Prior to harvesting, cells were treated with 100 ng/ml insulin or mock conditions for 10 min. Cells were harvested and lysed with Nonidet P-40 lysis buffer (10 mM Tris, pH 7.4, 150 mM NaCl, 0.5% Nonidet P-40, 1 mM phenylmethylsulfonyl fluoride) and centrifuged for 10 min at 10 4 ϫ g to pellet cell debris. To precipitate RalA⅐GTP or Rac⅐GTP, ϳ100 g of cell lysate was then tumbled with GST-RalBD or GST-PAK, respectively, fusion protein conjugated to glutathione-agarose. GST fusion protein resin was then collected by centrifugation and washed two times with Nonidet P-40 lysis buffer and one time with Nonidet P-40 lysis buffer plus 0.5 M NaCl. Bound proteins were resolved by SDS-PAGE and detected by Western blot analyses with anti-RalA antibody (Transduction Laboratories). Equal amounts of GST fusion protein in each sample was confirmed by staining the polyvinylidene difluoride filter with Ponceau S. Expression of RalA or Rac and HA-TC21(72L) or HA-Ha-Ras(61L) in total cell lysates was examined by SDS-PAGE and Western blotting with anti-RalA, anti-Rac (Upstate Biotechnology, Inc.), and anti-HA antibodies (Babco), respectively. Incubation of lysate with the GST resin alone did not extract RalA or Rac nonspecifically from the cell lysate (data not shown). For Rac activation analyses, Western blot band intensities from two individual experiments were quantitated, and Rac-GTP:Total Rac ratios were calculated and expressed in arbitrary units relative to vector alone, Ϯ error.
Akt Translocation Assays-NIH 3T3 cells were transiently transfected in serum-free medium with pEGFP empty green fluorescent protein (GFP) expression vector (CLONTECH) or pEGFP-Akt(PH) (a generous gift of Tobias Meyer) (63) and a 5-fold excess of HA epitopetagged expression constructs (pCGN-hygro) of TC21(72L) or Ha-Ras(61L) or pcDNA3-p110-CAAX. Three hours post-transfection, cells were transferred to DMEM supplemented with 10% calf serum. Twenty hours post-transfection, cells were viewed in a Zeiss Axiophot fluorescence microscope (63ϫ Plan-APOCHROMAT objective) equipped with a cooled charge-coupled device camera, and digitized images were captured using MetaMorph TM 4.1.4 digital imaging software (Universal Imaging Corp.). Western blot analysis of NIH 3T3 cell lysate using either anti-GFP monoclonal antibody (CLONTECH) or phospho-specific anti-Akt antibody (New England Biolabs) (see above) confirmed that GFP-Akt(PH) was expressed at the predicted molecular weight for the fusion protein and p110-CAAX increased phospho-Akt levels (data not shown).
Akt Activation Assays-NIH 3T3 cells were transiently transfected in serum-free medium with pCGN-hygro expression vectors encoding TC21(72L) or Ha-Ras(61L). Three hours post-transfection, cells were transferred to DMEM supplemented with 0.5% calf serum for ϳ20 h. Prior to harvesting, cells were either left untreated or treated with 10 M LY294002 or the equivalent volume of Me 2 SO for 30 min at 37°C. Cells were harvested and lysed in 100 l of Nonidet P-40 lysis buffer (10 mM Tris, pH 7.4, 150 mM NaCl, 0.5% Nonidet P-40, 1 mM phenylmethylsulfonyl fluoride) and centrifuged for 10 min at 10 4 ϫ g to pellet cell debris. To assess Akt activation, 50 g of cell lysate was resolved by SDS-PAGE, and phosphorylated Akt was detected by Western blot analyses with phospho-specific anti-Akt antibody. Expression of total Akt and HA-TC21(72L) or HA-Ha-Ras(61L) in cell lysates (10 g) was examined by SDS-PAGE and Western blot analyses with anti-Akt (New England Biolabs) and anti-HA antibodies, respectively.

TC21
Interacts with Some Ras⅐GTP-binding Proteins-Because our previous analyses showed that TC21 fails to interact with full-length Raf kinases or up-regulate Raf kinase activity in TC21-transformed cells (5,64), the key downstream effectors that mediate TC21 transformation are not known. Consequently, we used two approaches to determine the non-Raf effectors of TC21. The first approach was based on the fact that Ras and TC21 share common biological properties and complete amino acid identity in their core effector domain sequences. Therefore, TC21 may interact with other known Ras⅐GTP-binding proteins. We previously employed yeast two-hybrid library screening to identify Ras-binding proteins (60). In addition to the three Raf kinases, sequences corresponding to RalGDS, RGL, and a novel PI3K isoform, p110␦ (38), were identified. Furthermore, AF6 was identified by others (41,42) as a Ras⅐GTP-binding protein. Rin1 was identified by genetic analyses as a Ras-interfering protein and found to complex directly with Ras⅐GTP in vivo (40,65). Based on these previous findings, we employed yeast two-hybrid binding analysis to determine whether these Ras⅐GTP-binding proteins also interact with TC21 ( Fig. 1 and Table I).
Similar to our observations with Ha-Ras(12V), TC21(23V) interacted with full-length RalGDS (Table I and Fig. 1), although this interaction was weaker than that observed with Ras. In addition, TC21 interacted with p110␦, with a fragment of AF6 containing the Ras-binding domain, and weakly with full-length Rin1 ( Fig. 1 and Table I). These results suggest that RalGDS, PI3K, Rin1, and AF6 may function as effectors of TC21.
In addition to determining if TC21 interacts with known Ras⅐GTP-binding proteins, we also initiated a search for novel TC21-interacting proteins by expression library and yeast twohybrid library screening (Table II). Both approaches identified RalGDS and RalGDS-related proteins, RGL and RGL2/Rlf (66,67), as TC21-interacting proteins. Similarly, Barbacid and colleagues (68) identified RalGDS as a TC21-interacting protein using yeast two-hybrid library screening. In addition, both expression library and yeast two-hybrid screens revealed novel TC21-interacting proteins that are specific for binding TC21⅐GTP and require an intact effector domain for their interaction with TC21 (Table II). Expression cloning identified a protein member of the fetuin family (␣-2-HS-glycoprotein). Referred to as pp63 or tyrosine kinase inhibitor (TKI), this secreted protein is capable of antagonizing insulin signaling (69,70). Yeast two-hybrid screening identified PTP-1B, a ubiquitously expressed protein-tyrosine phosphatase implicated in down-regulation of several intracellular signaling pathways including insulin receptor phosphorylation and signaling (71,72). Importantly, both proteins also showed GTP-dependent binding to Ras to some extent (Table II). Thus, to date, we have not identified TC21-interacting proteins that do not also bind to Ras.
In in vitro binding analyses, both PTP-1B and TKI interacted with TC21 in a GTP-dependent manner and required an intact effector domain for the interaction (Table II and data not shown). In addition, glutathione S-transferase (GST)-TC21 (72L) fusion protein and full-length PTP-1B were co-expressed in 293 human embryonal kidney cells, and using glutathioneagarose resin, GST-TC21 (72L) and PTP-1B were co-isolated from cell lysates, suggesting that TC21 interacts with fulllength PTP-1B in cells (data not shown). Interestingly, in a data base search, no canonical Ras binding domains (RBDs) similar to those found in Raf or RalGEFs were found in PTP-1B or TKI. 2 Secondary structure prediction analysis of the PTP-1B amino acid sequence did not reveal a characteristic fold of RBDs, suggesting a novel motif may exist in PTP-1B that mediates its interaction with TC21 (73)(74)(75). It is important to note that our data do not conclusively demonstrate that these two proteins (PTP-1B and TKI) are bona fide effectors of TC21 function, and it needs to be shown with endogenous proteins in vivo that the interaction with TC21 leads to a functional alteration of the effector. A more detailed analysis of the physiological significance of the interactions of TKI and PTP-1B with TC21 is currently under investigation. TC21(72L) Mediates Activation of PLC⑀-PLC⑀ is a recently identified phosphoinositide-specific PLC that cleaves membrane phosphatidylinositol 4,5-bisphosphate to generate inositol 1,4,5-trisphosphate and DAG, two second messenger molecules that mobilize intracellular Ca 2ϩ stores and activate protein kinase C. It was determined recently that oncogenic Ha-Ras stimulates PLC⑀ activity in in vitro and in vivo monkey fibroblasts (43,45), suggesting that PLC⑀ may be an important effector of Ras function. We sought to determine whether TC21, like Ras, regulates PLC⑀ activation. Similar to activated Ha-Ras, transient expression of constitutively activated TC21 in COS-7 cells caused elevated production of inositol phosphates ( Fig. 2A) in PLC⑀-expressing cell populations. Data shown in Fig. 2A are normalized for PLC⑀ expression (Fig. 2B). Expression of PLC⑀ was not affected by Ha-Ras(61L) but was increased 1.37 Ϯ 0.075 (n ϭ 3)-fold in the presence of TC21(72L) (data not shown). These data show that like Ras TC21 caused activation of PLC⑀ and suggest that PLC⑀ is a downstream target of TC21 function.  1. TC21 interaction with Ras⅐GTP-binding proteins. Yeast two-hybrid binding analyses were performed to determine whether TC21 could interact with known Ras⅐GTP-binding proteins. S. cerevisiae containing a LexA-driven chromosomal HIS3 gene and a bacterial lacZ reporter gene were co-transformed with constructs to express TC21(23V) or Ha-Ras(12V) as a fusion protein with the LexA DNA binding domain (amino acids 1-211) and Raf-1, Raf-1-RBD, RalGDS, or PI3K p110␦ϪRBD as a fusion protein with the activation domain of VP16. Similar analyses were performed with truncated AF6 or fulllength Rin1 (see Table I). Yeast strains expressing two-hybrid proteins were streaked on medium containing 5-bromo-4-chloro-3-indolyl ␤-Dgalactopyranoside (X-gal). Proteins capable of interacting stimulated growth in the absence of histidine and contained detectable ␤-galactosidase activity.

TABLE I TC21 interaction with Ras ⅐ GTP-binding proteins
Yeast two-hybrid binding analyses were performed using a yeast strain in which expression of hybrid proteins capable of interacting results in histidine prototrophy and detectable ␤-galactosidase activity. Symbols: ϩ, strong interaction; Ϯ, weak interaction; Ϫ, no interaction. (41). c Full length Rin1 (65).
b Isolated by yeast two-hybrid library screening of a mouse embryo cDNA library with TC21. 18 million yeast transformants were screened with TC21, and a carboxyl-terminal fragment of PTP-1B (amino acids 367-432) was recovered. This fragment was subsequently tested for interaction with Ras(12V), TC21(72L), and TC21 (72L/46A). ϩϩ indicates a strong interaction; and ϩ/Ϫ indicates a weak interaction; Ϫ indicates no interaction.

Overexpression Of RalGDS Inhibits TC21 Signaling and
Transformation-To assess whether RalGDS is an effector of TC21 signaling and transformation in vivo, we determined if co-expression of RalGDS potentiated or attenuated activated TC21-mediated signaling and transformation in NIH 3T3 cells. First, we examined the effect of RalGDS expression on TC21mediated Elk-1 activation, a measure of ERK MAPK activation. We observed that co-expression of full-length RalGDS did not enhance TC21(72L) activity and, instead, strongly inhibited Elk-1 activation (Fig. 3A). This inhibition was dose-dependent, observed at low DNA concentrations of RalGDS expression plasmid, and similar to the effect observed on the TC21 (23V) activating mutation (data not shown). Similarly, RalGDS also inhibited Ras-mediated Elk-1 activation (Fig. 3A). The block in TC21 or Ras signaling caused by RalGDS was specific for GTPase signaling, because co-expression of RalGDS did not cause a significant reduction of constitutively activated MEK (designated MEK⌬ED)-induced stimulation of Elk-1 (Fig.  3A).
Because co-expression of full-length Raf-1 has been shown to cooperate with wild type or oncogenic Ha-Ras and cause synergistic focus forming activity (76), we tested the effect of Ral-GDS expression on TC21-induced focus forming activity. However, analogous to our results in the signaling assay described above, expression of RalGDS significantly blocked TC21(72L)mediated focus formation (ϳ40% inhibition, Fig. 3B). These results are similar to those observed for TC21 (23V)-induced focus formation (data not shown). Co-expression of RalGDS did not greatly effect Ha-Ras(61L) focus forming activity (Fig. 3B).
Similar analyses using the isolated RBD of Raf-1 showed that these truncation mutants blocked oncogenic Ras signaling and transformation (66,77), presumably by interfering with interactions between Ras and Ras-binding proteins with comparable or lower affinity than the RBD for Ras. Additional observations showed that co-expression of RalGDS-RBD or RGL2RBD could impair Ha-Ras(61L) focus forming activity in NIH 3T3 cells (66). Similar to co-expression of the RBD of Raf-1, we found that co-expression of RalGDS-RBD or RGL2-RBD impaired TC21-mediated focus formation and activation of Elk-1 by greater than 50% (data not shown). Taken together, these results suggest that RalGDS can interact with TC21 in vivo. However, RalGDS may not be an important mediator of signaling pathways that contribute to TC21 growth transformation.
TC21(72L) Fails to Activate RalA and Inhibits Insulin-stimulated RalA Activation-Stimulation of receptor tyrosine kinases, such as insulin stimulation of the insulin receptor, has been shown to mediate an increase in GTP loading of RalA leading to activation of the RalA GTPase (78). Furthermore, previous studies (14,79) have shown that expression of oncogenic Ras or membrane-targeted RalGDS mediates RalA activation. To determine whether the interaction of TC21 with RalGDS facilitated activation of the latter, we expressed constitutively activated TC21(72L) or Ha-Ras(61L) in NIH 3T3 cells and qualitatively assessed the activation of RalA, as measured by specific interaction of endogenous RalA⅐GTP with a GST fusion protein containing the Ral binding domain (RalBD) of RalBP1. As shown in Fig. 4A, expression of Ha-Ras(61L) resulted in an increase in the levels of RalA⅐GTP; in contrast, TC21(72L) failed to elicit a similar response. Furthermore, insulin-stimulated RalA⅐GTP loading was dramatically inhibited by expression of TC21(72L) but not by Ha-Ras(61L) (Fig.  4B). Endogenous RalA and exogenous GTPase expression were approximately equal in all samples and thus cannot account for the differential effects of TC21 and Ha-Ras on RalA stimulation. Taken together, these results suggest that although both Ha-Ras and TC21 interact with RalGDS, only Ha-Ras stimulates its guanine nucleotide exchange activity for RalA. Fur- thermore, activated TC21 may form a non-productive complex with endogenous RalGDS thus preventing its activation by insulin.
TC21(72L) Activates Akt in a PI3K-dependent Manner-In addition to interacting with RalGDS, TC21 also interacts with p110␦, the catalytic subunit of PI3K ( Fig. 1 and Table I). To determine whether TC21 causes activation of PI3K, we utilized two approaches to assess the activation of Akt, a downstream target of PI3K activation. PI3K and Akt are implicated in signal transduction events associated with Ras-mediated transformation (80). PI3K mediates Akt activation by promoting translocation of Akt to the plasma membrane via the interaction of the Akt pleckstrin homology (PH) domain with the phosphorylated inositol products of PI3K, such as phosphatidylinositol 3,4,5phosphate (81,82). First, we examined whether TC21 mediates translocation of Akt to the plasma membrane. We co-expressed constitutively activated TC21(72L), Ha-Ras(61L), or PI3K (p110-CAAX) with a limiting amount of GFP or GFP-Akt(PH) in NIH 3T3 cells and qualitatively assessed PI3K activation, as measured by translocation of GFP-Akt(PH) to the plasma membrane. As shown in Fig. 5, similar to what was seen with PI3K activation (p110-CAAX), expression of either TC21(72L) or Ha-Ras(61L) resulted in an increase in membrane localization of GFP-Akt(PH) but had no effect on the distribution of GFP alone.
Because we showed that TC21 interacts with p110␦ and stimulates translocation of Akt to the plasma membrane, we next sought to determine whether TC21(72L) mediates PI3K-dependent Akt activation. We expressed constitutively activated TC21(72L) or Ha-Ras(61L) in NIH 3T3 cells and qualitatively assessed PI3K activation, as measured by phosphorylation of endogenous Akt (81). As shown in Fig. 6, expression of either TC21(72L) or Ha-Ras(61L) resulted in an increase in activated, phospho-Akt. Furthermore, activation of Akt by TC21(72L) or Ha-Ras(61L) was inhibited completely by 10 M LY294002, a specific inhibitor of PI3K (Fig. 6). Taken together, our results suggest that both Ha-Ras and TC21 stimulate PI3K-dependent activation of Akt.
PI3K Activation Is Required for TC21-mediated Transformation-It was determined previously that Ras-mediated transformation (focus formation) requires PI3K activity (80). Thus a logical candidate effector for mediating TC21 transformation is PI3K. To determine whether PI3K function is necessary for TC21-mediated focus forming activity, we used two approaches to interfere with PI3K activity. First, we transfected NIH 3T3 cells with expression vectors that encode TC21(72L) or Ha-Ras(61L), maintained the cells in the absence (Ϫ) or presence (ϩ) of 10 M LY294002, and quantitated the number of foci of transformed cells formed. Focus formation was expressed as a percent of the maximum focus forming activity of Ha-Ras(61L) or TC21(72L) in the absence of LY294002. As shown in Fig. 7A, inhibition of PI3K by LY294002 blocked TC21-mediated focus formation by ϳ50% and Ha-Ras-mediated focus formation by ϳ60%. LY294002 was not generally toxic to the background monolayer of untransformed cells (data not shown).
Second, we also used a genetic approach to confirm that PI3K function was necessary for TC21 focus forming activity. We co-expressed TC21(72L) or Ha-Ras(61L) with an excess of ⌬p85, a dominant-inhibitory mutant of the p85 regulatory subunit of PI3K that lacks the binding site for the p110 catalytic subunit (83). It was shown previously that expression of ⌬p85 interferes with focus forming activity of Ha-Ras(12V), but not the viral Src oncoprotein, and does not cause a general toxicity to NIH 3T3 fibroblasts (80). As shown in Fig. 7B, co-expression of ⌬p85 with Ha-Ras(61L) or TC21(72L) reduced focus formation by ϳ60 and 40%, respectively. This inhibitory effect of ⌬p85 was similar to that observed in cells treated with LY294002. Taken together, these data suggest that PI3K function is essential for maximal transforming potential of TC21.
TC21(72L) Fails to Activate Rac-The mechanism by which Ras mediates activation of the Rac small GTPase is poorly understood. It has been suggested that Ras mediates Rac⅐GTP up-regulation by PI3K-dependent phospholipid production (80) and subsequent activation of a Rac exchange factor (84 -86). Alternatively, RalA⅐GTP interacts with a putative GAP for Rac, suggesting Rac⅐GTP level could be regulated by Ras-mediated, RalGDS-stimulated RalA activation (87)(88)(89). Because we determined that the PI3K/Akt pathway, but not the RalGDS/ RalA pathway, is activated by TC21(72L), these uncertainties prompted us to examine if TC21 regulates Rac activation. Thus, we evaluated NIH 3T3 cells expressing constitutively activated TC21(72L) or Ha-Ras(61L) and assessed activation of Rac as measured by specific interaction of endogenous Rac⅐GTP with a GST fusion protein containing the Cdc42/Rac-interactive binding motif of PAK (GST-PAK). As shown in Fig. 8A, expression of Ha-Ras(61L) or treatment of cells with insulin prior to harvesting resulted in an increase in Rac⅐GTP; however, TC21(72L) failed to cause an increase in Rac⅐GTP. Endogenous Rac and exogenous GTPase expression were approximately equal in all samples and thus cannot account for the differential effects of TC21 and Ha-Ras on Rac stimulation. Ha-Ras(61L) and insulin treatment induced approximately 4-fold up-regulation of Rac-GTP loading as shown in Fig. 8B. Taken together, our results suggest that Ha-Ras, but not TC21, stimulate Rac activation in vivo and that PI3K may not be the major mechanism by which Ras stimulates Rac activation. DISCUSSION To date, TC21/R-Ras2 is the only member of the Ras family of small GTPases that has been found to share significant biological activities with the four Ras proteins. However, although the Raf kinases are important effectors of Ras-mediated transformation, TC21 fails to interact with and activate Raf kinases (5,6). How TC21 regulates similar cellular processes as Ras and whether TC21 utilizes other known Ras effectors to mediate transformation are unclear. In the present study, we sought to identify effectors of TC21-mediated transformation. We determined if TC21 could interact with and activate other known Ras effectors. Although TC21 interacted with RalGDS proteins, in contrast to Ras, TC21 did not promote activation of RalA. However, like Ras, TC21 did activate PI3K, and inhibition of PI3K activity blocked TC21 transforming activity. Because activated PI3K alone is not sufficient to cause transformation, we conclude that TC21 must use additional effectors other than Raf or RalGDS to cause cellular transformation. tagenesis of specific residues in this expanded effector domain has shown that residues spanning Ras amino acids 25-45 are also important for Ras recognition of Raf and other Ras-binding proteins (90). Therefore, it was not unexpected that TC21 can interact, to varying degrees, with all Ras-binding proteins evaluated in this study. Specifically, our yeast two-hybrid binding analyses showed that TC21 can interact with RalGDS and related proteins, a novel PI3K catalytic subunit isoform (p110␦), and AF6. Rin1, which interacted weakly with Ras in these assays, also showed weak interaction with TC21. Similar to the two-hybrid binding analyses, expression library screening also identified RalGDS and related proteins as possible effectors of TC21 function, showing preferential interaction with GTP-bound TC21. Barbacid and colleagues (68) also showed that TC21 can complex with endogenous RalGDS in vivo. Whether TC21 can interact with effectors that are not shared with Ras is presently not clear. Finally, our search for TC21-interacting proteins by library screening has identified two novel candidate effectors of TC21 (pp63 TKI and PTP-1B). The interaction of TC21 with full-length PTP-1B was shown in vitro by two-hybrid analysis and in cells by co-expression and co-precipitation of PTP-1B with constitutively activated TC21 GST fusion protein from mammalian cells ( Fig. 1 and Table II, and data not shown). Both candidate effectors also showed GTP-dependent binding to Ras. Thus, whether TC21 can interact with effectors that are not shared with Ras is presently not clear. It is important to note that our data do not conclusively demonstrate that these two proteins (PTP-1B and TKI) are bona fide effectors of TC21 function, and it needs to be shown with endogenous proteins in vivo that the interaction with TC21 leads to a functional alteration of the effector. Thus, we are currently analyzing the possible role(s) of these additional TC21-interacting proteins in TC21-mediated transformation.
At present, there are limited data that support a role for RalGDS in mediating Ras transformation (34). For example, co-expression of RalGDS with activated Raf-1 showed synergistic transforming activity in NIH 3T3 focus formation assays (54). These data suggest that the coordinate activation of Raf-1 and RalGDS contributes to Ras transformation. Dominant negative RalA, a target of RalGDS, was shown to inhibit Ras transforming activity in one study (14), although other studies have not observed this inhibition (54). However, in the present study, we found that co-expression of RalGDS did not enhance either activated TC21 or Ras focus forming activity. Instead, we found that overexpression of RalGDS caused an inhibition of TC21 transforming activity. Furthermore, we found that RalGDS/CAAX was growth-inhibitory when overexpressed in NIH 3T3 cells (data not shown). Finally, unlike activated Ras, we found that TC21 failed to activate RalGDS and instead interfered with insulin-simulated RalA⅐GTP loading. Therefore, the inhibition of TC21 transformation caused by RalGDS overexpression suggests that it may prevent the interaction of TC21 with other effectors that are more critical for promoting transformation of NIH 3T3 cells. Thus, similar to observations with the related R-Ras protein (66,91), TC21 can bind but not activate RalGDS. Our results indicate that RalGDS may not be a key effector for mediating TC21 growth transformation.
During the course of our studies, Rosario et al. (92) described studies evaluating the role of RalGDS in TC21 signaling and transformation. They determined that microinjection of activated TC21 caused a relocation of co-injected green fluorescent protein-tagged RalGDS to the plasma membrane of MDCK cells. Additionally, transient or stable expression of activated TC21 caused an increase in Ral⅐GTP in NIH 3T3 cells. Finally, inhibition of Ral or RalGDS function blocked TC21-induced DNA synthesis, but not morphologic transformation, when analyzed in Swiss or NIH 3T3 cells. Although experimental or cell type differences may account for some of the different observations made in our study, our analyses failed to identify an increase in Ral⅐GTP in TC21-expressing cells. We should also indicate that although we and others suggest that TC21 does not activate Raf, some groups have reported that TC21 does activate Raf (6,20,21). These differences may also be ascribed to strain type differences in NIH 3T3 cells (49) or the utilization of different expression vectors for ectopic expression of TC21 proteins (93). In addition, RalGDS may possess functions independent of its GEF activity (94). Because we did find that TC21 could interact with RalGDS, our results reported here do not exclude the possibility that TC21 may regulate functions of RalGDS independent of Ral activation.
PI3K has been shown to be a critical effector for Ras-induced transformation of NIH 3T3 cells, as well as human thyroid (95), but not RIE-1 epithelial cells (64). Our present study shows that PI3K is also activated by TC21 and a mediator of TC21 transformation. Rosario et al. (92) also determined that TC21 interacts with and activates PI3K in NIH 3T3 cells and that treatment with LY294002 resulted in inhibition of DNA synthesis and morphologic reversion of TC21-transformed NIH FIG. 8. Effect of TC21 (72L) on Rac-GTP loading. Rac activation is measured by specific interaction of Rac⅐GTP with the Rac binding domain of PAK (GST-PAK). A, TC21(72L) fails to up-regulate Rac GTP loading. NIH 3T3 cells were transiently transfected with HA-tagged expression constructs of TC21(72L) and Ha-Ras(61L). Cell lysate was then tumbled with GST-PAK fusion protein conjugated to glutathioneagarose. GST-PAK resin was then collected by centrifugation and washed. Bound proteins were resolved by SDS-PAGE and detected by Western blotting with anti-Rac antibody. Expression of Rac and HA-TC21(72L) or HA-Ha-Ras(61L) in total cell lysate was examined by SDS-PAGE and Western blotting with anti-Rac and anti-HA antibodies, respectively. Incubation of lysate with GST resin alone did not extract Rac nonspecifically from the cell lysate (data not shown). Data shown are representative of two independent experiments. B, quantitation of Rac-GTP loading. Western blot band intensities from two independent experiments were quantitated, and Rac-GTP:Total Rac ratios were calculated and expressed in arbitrary units (a.u.) relative to vector alone, Ϯ error. 3T3 cells. Thus, of the three major effectors of Ras transformation of NIH 3T3 cells, we have found that PI3K is the only one involved in transformation by both TC21 and Ras.
We showed previously (96) that Rac is required for Rasmediated transformation of NIH 3T3 cells. Ras is thought to cause activation of Rac via activation of PI3K (80, 84 -86). Because we found that TC21 also activated PI3K, we determined if TC21 can also activate Rac. Surprisingly, whereas activated Rac⅐GTP levels were elevated in Ha-Ras(61L)-transfected cells, we found no increase in TC21(72L)-transfected cells, suggesting that Rac function may not be required for TC21 activity and that PI3K may not be the link that connects Ras with Rac. Recently, we determined that activated PI3K (p110-CAAX) failed to up-regulate Rac⅐GTP loading in NIH 3T3 cells 3 and that Ras-mediated activation of Tiam1, but not PI3K, was involved in causing activation of Rac. 4 Specifically, inhibition of PI3K by LY294002 fails to block Ras-mediated Rac activation. 4 The lack of activation of Rac by TC21 is also surprising because we showed previously that activated TC21 could activate Jun NH 2 -terminal kinase in COS cells, a Racmediated activity, and that dominant negative Rac1 could block TC21-mediated transformation. It is possible that basal Rac activity is required for TC21 transformation, or alternatively, the Rac1 dominant negative may not be a specific inhibitor of Rac activation. Some of our recent analyses using the Rac1 dominant negative have found that, under conditions of sustained overexpression, nonspecific inhibition of Rac-independent activities can be seen. 5 In summary, our previous and current studies demonstrate that, despite the significant functional similarities between Ras and TC21, TC21 fails to activate two key effectors (Raf and RalGDS) important for Ras-mediated transformation. We determined that, like Ras, TC21-mediated transformation is dependent on PI3K. However, because activated PI3K alone does not cause transformation of NIH 3T3 cells (80), TC21 must utilize other effectors to mediate transformation. Because we determined that like Ras, TC21 is also an activator of PLC⑀, further study to determine whether PLC⑀-mediated second messenger activation is important for TC21 transformation is clearly necessary. Finally, we identified novel candidate effectors of TC21 in this study (PTP-1B and pp63 TKI). Whether these are physiologically relevant effectors of TC21 and contribute to TC21-mediated transformation remain to be established.