RalGDS functions in Ras- and cAMP-mediated growth stimulation.

Thyroid-stimulating hormone stimulates proliferation through both the cAMP-dependent protein kinase and Ras but not through Raf-1 and mitogen-activated and extracellular signal-related kinase kinase. We now report that thyroid-stimulating hormone represses mitogen-activated protein kinase activity and that microinjection of an effector domain mutant Ha-Ras protein, Ras(12V,37G), defective in Raf-1 binding and mitogen-activated protein kinase activation, stimulates DNA synthesis in quiescent and thyroid-stimulating hormone-treated thyrocytes. A yeast two-hybrid screen identified RalGDS as a Ras(12V,37G) binding protein and therefore a potential effector of Ras in these cells. Associations between Ras and RalGDS were observed in extracts prepared from thyroid cells. Microinjection of a mutant RalA(28N) protein thought to sequester RalGDS family members reduced DNA synthesis stimulated by Ras as well as cAMP-mediated DNA synthesis in two cell lines which respond to cAMP with mitogenesis. These results support the idea that RalGDS may be an effector of Ras in cAMP-mediated growth stimulation.

The elucidation of the critical role played by Ras in growth control initiated an intensive effort to identify Ras binding effector molecules. The first Ras binding protein identified was p120GAP (1), which functions both as an effector and downregulator of Ras. The best characterized Ras effector is the cytoplasmic serine/threonine protein kinase Raf-1 (2,3). Direct interaction between the N-terminal domain of Raf-1 and the effector loop of Ras was first demonstrated using the yeast two-hybrid system (4,5), while later studies reported the coprecipitation of Ras and Raf (6,7). A number of other proteins bind directly to GTP-bound Ras, including the p110 catalytic subunit of phosphatidylinositol 3-kinase (8), AF6 (9), Rin-1 (10), NF1 (11), MEKK1 (12), protein kinase C (13), a herpesvirus-encoded protein (14), and a guanine nucleotide dissocia-tion stimulator for the Ras-related protein Ral (RalGDS) and a closely related protein (RGL) (15)(16)(17)(18). While interaction of Ras with Raf-1, p110 (8), and RalGDS (19) results in increased enzyme activity, the biological outcomes of these other Rasprotein interactions are less understood. TSH 1 -stimulated DNA synthesis requires both the cAMP-dependent protein kinase and Ras (20). Although Ras-dependent, mitogenic signaling induced by TSH does not require Raf-1 or MEK (21), suggesting that Ras utilizes alternate effectors in the presence of cAMP. To identify such effectors, an extensive two-hybrid screen of a thyroid cell cDNA library was performed with an effector domain mutant of Ras which does not bind Raf-1 (22) but stimulates DNA synthesis in thyrocytes. This screen repeatedly identified RalGDS as a potential Ras effector. Consistent with such a role, interaction between Ras and RalGDS was detected in extracts prepared from thyroid cells. Further, microinjection of a mutant RalA(28N) protein thought to sequester RalGDS (19,23) specifically reduced cAMP-mediated DNA synthesis in both rat thyrocytes and Schwann cells as well as DNA synthesis stimulated by Ras. These results are consistent with others (19,(23)(24)(25)(26) which suggest that RalGDS functions as an effector of Ras in some cellular signaling pathways. Our results suggest that RalGDS may be an important component of cAMP-stimulated mitogenesis.

EXPERIMENTAL PROCEDURES
Cell Culture, DNA Synthesis, and Microinjection-The conditions under which Wistar rat thyrocytes (WRT) were propagated have been described previously (27). Thyrocytes were rendered quiescent by incubation in TSH-and serum-free medium containing 0.3% bovine serum albumin and 0.5 g/ml insulin for 48 h. Secondary rat Schwann cells were propagated in Dulbecco's modified Eagle's medium supplemented with glial-derived growth factor and forskolin as described for human cells (28). One week prior to injection, Schwann cells were plated on coverslips in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum. Proteins were coinjected with rabbit IgG into the cytoplasm. Injected cells were identified by staining with an fluorescein isothiocyanate-labeled anti-rabbit antibody, and DNA synthesis was assessed by bromodeoxyuridine (BrdUrd) incorporation as described previously (20).
Expression and Purification of GST Fusion Proteins-cDNA corresponding to the C-terminal 98 amino acids of RalGDSb was kindly provided by Dr. F. Hofer. (University of California, Berkeley). The C-terminal domain of RalGDSb isolated in the two-hybrid screen was cloned into pGEX2T (Pharmacia Biotech Inc.). pGEX-Ha-Ras(12V), Ras(12G), and Ras(12V,37G) were provided by Dr. M. White (University of Texas Southwestern Medical Center, Dallas). Vectors encoding RalA, RalA(72L), and RalA(28N) fused to GST were kindly provided by Dr. L. A. Feig (Tufts University). The GST fusion proteins were ex-pressed and purified as described by the manufacturer. For injection studies, glutathione-Sepharose beads containing the fusion proteins were incubated with 1.5 units of thrombin for 16 h at 4°C. Following centrifugation, the supernatant was incubated with 20 l of p-aminobenzamidine-agarose beads for 30 min at 4°C. Following centrifugation, the purified proteins were exchanged into injection buffer (20 mM Tris, pH 7.5, 20 mM NaCl, 2 mM MgCl 2 , 0.1 mM EDTA), concentrated, and aliquots frozen at Ϫ80°C.
Construction of Thyroid Cell Library and Yeast Two-Hybrid Screen-Total cellular RNA was prepared from WRT cells using guanidinium hydrochloride and mRNA isolated using a Pharmacia oligo(dT) purification kit (Pharmacia Biotech Inc.). cDNA was synthesized using a Great Lengths cDNA synthesis kit (Clontech) and cloned into EcoRIdigested yeast two-hybrid vector pGADGH (Clontech) using EcoRI adapters. Mutant Ha-Ras(12V,37G) in the yeast two-hybrid vectors pGBT9 and pBTM116 was kindly provided by Dr. M. H. Wigler, Cold Spring Harbor Laboratory. The genotype of the Saccharomyces cerevisiae reporter strain HF7c is MATa ura3 his3 ade2 lys2 trp1 gal80 LYS2::GAL1-GAL1-HIS3-URA3::GAL4-CYC1-lacZ (30). The genotype of the L40 reporter strain is MATa trp1 leu2 his3 LYS2::lexA-HIS3-URA3::lexA-lacZ (5). Approximately 2 ϫ 10 6 clones were screened with both Ras-LexA and Ras-Gal4 constructs. Positive clones were identified on the basis of their ability to grow in the absence of histidine and to stimulate ␤-galactosidase expression.
Generation of RalGDS Monoclonal Antibodies-Mice were immunized following a standard regimen with a bacterially produced protein encoding a portion of RalGDSa as described previously (34). Spleens were fused to P3-X63-Ag8.653 cells to generate hybridomas following standard protocols (44).
cAMP Response Element-regulated Gene Expression-Quiescent WRT cells stably transfected with a cAMP response element-regulated lacZ gene were injected with the RalGDS Ras binding domain together with rabbit IgG and subsequently stimulated with TSH (10 mU/ml) or 8-Br-cAMP (1 mM) for 6 h. Following fixation, the cells were stained with an fluorescein isothiocyanate-anti-rabbit antibody (to identify injected cells) and 5-bromo-4-chloro-3-indoyl ␤-D-galactoside (to assess ␤-galactosidase expression) and scored as described previously (20).

Ras(12V,37G) Stimulates DNA Synthesis-Since
Raf-1 is apparently not required for the mitogenic effects of TSH, we assessed whether microinjection of Ras(12V,37G), an effector domain mutant defective in binding Raf-1, was sufficient to stimulate DNA synthesis. Data shown in Table I demonstrate that injection of Ras(12V,37G) stimulated DNA synthesis in quiescent WRT cells. Although Ras(12V,37G) encoded an activating mutation, it appeared to be less active than Ras(12V) in these assays but similar in activity to cellular Ras(12G). Ras(12G,37G) also stimulated DNA synthesis, although this single mutant was less active than both its activated counter-  part and cellular Ras (data not shown). Unlike the Ras proteins, activated forms of Rho and Rac as well as GST alone (data not shown) failed to stimulate DNA synthesis, although they were purified in parallel with the Ras proteins. Like cellular Ras (21), Ras(12V,37G) retained its mitogenic activity in the presence of cAMP stimulated by TSH, consistent with its effects being independent of Raf-1.
Effects of TSH on MAPK Activity-In some cells, cAMP inhibits Raf activation without similar effects on MAPK activity (31,32). Since TSH-stimulated DNA synthesis appears both Raf-1-and MEK-independent, the effects of TSH on MAPK activity were examined. Treatment with TSH for 5-120 min (Fig. 1A, lanes 2-6) failed to stimulate MAPK activity, although 12-O-tetradecanoylphorbol-13-acetate stimulated a dramatic increase in activity (lane 7). Similar to the effects of TSH, which elevates cAMP, forskolin and 8-Br-cAMP also failed to increase MAPK activity in these cells (data not shown). These results are in good agreement with those published earlier in canine thyrocytes (33). Not only did TSH fail to increase MAPK activity, pretreatment with TSH repressed serum-stimulated (Fig. 1B) and IGF-1-stimulated activity (data not shown). This data further supports our hypothesis that TSH acts through a novel pathway which is Ras-dependent but independent of Raf, MEK, and MAPK.
Identification of RalGDS as a Ras(37G) Binding Protein-To identify potential Ras effectors expressed in thyroid cells, a yeast two-hybrid screen was performed to identify Ras binding proteins. Ras(12V,37G) was used in this screen in order to identify proteins other than Raf-1 which bind to Ras. Approximately 2 ϫ 10 6 clones were screened and a total of 4 positive clones were obtained that permitted growth under selective conditions and which expressed ␤-galactosidase (Fig. 2). The cDNA inserts from all 4 clones encoded a single sequence identical to the C-terminal Ras binding domain (RBD) of Ral-GDSb (15-18, 34).

FIG. 3. Effects of the RalGDS RBD on Ras and cAMP-mediated signaling.
A, Ras(12V) (0.2 mg/ml) and Ras(12V37G) (1.5 mg/ml) (Ras37G) were injected alone or together with the RalGDS RBD (20 mg/ml) (RBD) into quiescent thyrocytes and DNA synthesis was examined. More than 300 cells were examined for each condition. B, the RalGDS RBD (20 mg/ml) was injected into quiescent thyrocytes which were subsequently stimulated with TSH (1 mU/ml), cholera toxin (CT) (10 g/ml), or 8-Br-cAMP (8Br) (1 mM) and DNA synthesis examined. More than 200 cells were examined for each condition. None indicates no treatment or the level of BrdUrd incorporation in quiescent cells. C, the RalGDS RBD (20 mg/ml) was injected into quiescent thyrocytes stably transfected with a cAMP response element-regulated lacZ gene. Following stimulation with TSH (1 mU/ml) or 8-Br-cAMP (8Br) (1 mM) for 6 h, the cells were fixed and stained with 5-bromo-4-chloro-3-indoyl ␤-D-galactoside. More than 200 cells were examined for each condition. All injection solutions contained rabbit IgG (6 mg/ml) used to identify injected cells. those obtained following injection of a dominant interfering mutant Ras(17N) protein (20).
To assess whether RalGDS binds to Ras in thyroid cells, in vitro binding experiments were performed. Lysates from exponentially growing thyroid cells were incubated with immobilized GST fusion proteins encoding activated forms of Ras and Rho, washed, and associated proteins were analyzed by Western blotting with a RalGDS monoclonal antibody (Fig. 4A) or mouse IgG (Fig. 4B). RalGDS was detected only in lysates incubated with GTP-bound Ras and probed with the RalGDS antibody. Neither GDP-bound Ras nor GTP-bound Rho interacted detectably with RalGDS. Probing with anti-GST (Fig. 4C) and anti-Ras (Fig. 4D) antibodies revealed that the abundance of the immobilized GST fusion proteins was similar.
Ral Activity Is Required for Ras-and cAMP-stimulated DNA Synthesis-To assess whether RalGDS was required for Rasstimulated DNA synthesis, a dominant interfering mutant RalA(28N) protein (19,23) was utilized. Similar to dominant interfering Ras(17N), this mutant Ral protein exhibits a preferential affinity for GDP (36) and should therefore act to sequester molecules which bind to the GDP-bound form of RalA, including RalGDS and related molecules such as RGL (16). Injection of RalA(28N) protein reduced DNA synthesis stimulated by cellular, activated, and the effector domain Ras proteins in thyroid cells (Table II). Ras(12G) and Ras(12V,37G) appeared more sensitive to repression by RalA(28N) than was Ras(12V). The reason for this difference remains to be determined.
Because TSH-stimulated DNA synthesis is Raf-1-, MEK-, and MAPK-independent, we assessed whether RalA(28N) would reduce it. Microinjection of RalA(28N) markedly reduced TSH-stimulated (Fig. 5, a and b), cholera toxin-stimulated, and 8-Br-cAMP-stimulated DNA synthesis, all mitogens which act through elevations in cAMP, to near background levels (Table  III). In contrast, DNA synthesis stimulated by IGF-1 (Fig. 5, c  and d) or fetal calf serum (FCS) was not reduced by injection of RalA (28N). Similarly, injection of RalA(28N) did not reduce FCS-stimulated DNA synthesis in NIH3T3 or REF52 fibroblasts, pathways which are partially Ras-dependent but independent of cAMP (data not shown).
These results suggested that Ral activity is critical for cAMP-mediated DNA synthesis. To examine this further, we extended our analysis to proliferating rat Schwann cells, which respond to cAMP with mitogenesis (28,37,38). Injection of RalA(28N) profoundly reduced DNA synthesis stimulated by cholera toxin and 8-Br-cAMP in these cells (Table III). Because RalA(28N) is thought to act by sequestering Ral exchange factors, including RalGDS (39), the repressive effects of RalA(28N) on cAMP-and Ras-mediated DNA synthesis suggest that a member or members of the RalGDS family play an essential role in Ras-dependent, cAMP-mediated mitogenic signaling pathways. DISCUSSION TSH stimulates proliferation through a cAMP-dependent pathway which requires Ras activity (20). Although Ras-dependent, TSH does not utilize Raf-1 or MEK in mitogenic signaling (21). We now demonstrate that TSH fails to stimulate MAPK activity and represses serum-stimulated activity. Consistent with these results, microinjection of a Ras effector domain mutant Ras(12V,37G), defective in Raf-1 interaction, stimulated DNA synthesis in quiescent and TSH-treated thyrocytes. Based on these results, this mutant Ras protein was   (28N) reduces Ras-stimulated DNA synthesis Column headings are the same as described in Table I. The results shown are from a single experiment derived from a set of 6 -8 experiments which gave similar results. used in a yeast two-hybrid screen of a thyroid cell library. This screen identified RalGDS as a Ras(12V,37G) binding protein and therefore a potential effector of Ras. Similar results were recently reported by White et al., who found that Ras(12V,37G) binds RalGDS expressed in a fibroblast library (18). RalGDS was first isolated from a mouse cell library using sequences derived from yeast Ras guanine nucleotide dissociation stimulator proteins as probes (34). Although its repeated identification as a Ras binding protein (15)(16)(17) suggests that RalGDS acts in Ras-dependent signaling pathways, its biological roles remain unknown. Kikuchi et al. reported that protein kinase A enhanced Ras association with RalGDS by decreasing Ras-Raf-1 interaction (24). Because TSH-stimulated DNA synthesis is both protein kinase A-and Ras-dependent, we examined whether RalGDS played a role in mitogenic signaling initiated by TSH.
In vitro binding data revealed that Ras binds to RalGDS in thyroid cell extracts. Consistently, microinjection of the Ral-GDS RBD reduced Ras-stimulated DNA synthesis, suggesting that this domain also binds to Ras in living cells. As expected, injection of the RalGDS RBD also reduced TSH-and cAMPstimulated DNA synthesis in these cells, confirming the importance of Ras in this signaling pathway. A mutant RalA(28N) protein exhibiting a preferential affinity for GDP (36) analogous to the dominant interfering Ras(17N) mutant was used to examine whether RalGDS or a related activity was required for TSH-stimulated mitogenesis. RalA(28N) has been used previously to document a requirement for Ral activity in Src-stimulated phospholipase D activation (23) and in Ras-stimulated transformation (19). Microinjection of RalA(28N) dramatically reduced DNA synthesis stimulated by TSH, cholera toxin, and 8-Br-cAMP, all mitogens which act through elevations in cAMP. The level of DNA synthesis in quiescent thyrocytes varies between 0 and 30% depending upon how long the cells have been in culture. In most experiments, injection of RalA(28N) very nearly abolished DNA synthesis stimulated by agents which act through cAMP. In contrast, IGF-1-and FCSstimulated DNA synthesis was little affected by injection of RalA(28N). These results suggest that a greater proportion of cAMP-mediated mitogenic signaling proceeds through a Ral-GDS-sensitive pathway in these cells than does signaling stimulated by IGF-1 or FCS. This would be expected given that IGF-1 and FCS do not elevate cAMP and therefore can transduce Ras-dependent signals through Raf-1 and the MAPK cascade. Unlike TSH, which fails to stimulate MAPK activity in either rat (Fig. 2) or canine thyrocytes (33), both IGF-1 and FCS increased MAPK activity in thyroid cells. Pretreatment with TSH abolished this activation, demonstrating that TSH interferes with signaling through the Raf-MAPK cascade and suggesting that TSH channels its Ras-mediated signals to a novel pathway involving RalGDS or a related molecule.
To determine whether the apparent requirement for Ral activity extended to other cells in which cAMP stimulates mi-togenesis, RalA(28N) was injected into secondary rat Schwann cells. In the presence of serum, these cells cycle only very slowly. Addition of cholera toxin, 8-Br-cAMP, or forskolin stimulates a dramatic increase in DNA synthesis (28,37,38). Similar to the effects observed in rat thyrocytes, injection of RalA(28N) profoundly reduced both cholera toxin-and 8-Br-cAMP-stimulated DNA synthesis in secondary Schwann cells. These results support the hypothesis that RalGDS plays an important role in cAMP-stimulated mitogenesis.
Although in vitro RalGDS catalyzes nucleotide exchange on Ral, its effectors in vivo are unknown. By analogy with Ras, it is possible that recruitment of RalGDS to the plasma membrane through association with Ras activates Ral. However, microinjection of RalA or a GTPase-defective mutant RalA-(72L) protein failed to stimulate DNA synthesis in quiescent thyrocytes. This may indicate that other factors, i.e. cAMP or Ras, are required for Ral to exert its mitogenic effects. This would be consistent with recent results demonstrating that RalA(72L) failed to stimulate phospholipase D activity, although it potentiated this activity in src-transformed cells (23), and that it failed to stimulate focus formation although it enhanced Ras-mediated transformation (19). If Ral is the effector of RalGDS, then other small G proteins are likely to function in this signaling pathway. Three laboratories have identified Ral binding proteins which exhibit characteristics expected for a Ral effector and cdc42GAP activity (40 -42). It is also conceivable that there are targets of RalGDS other than RalA and RalB. In fibroblasts, RalGDS but not Ral collaborated with Raf in focus-forming assays (18). In further support of this idea, microinjection of RalA protein repressed TSH-stimulated DNA synthesis, perhaps indicating that RalA interferes with Ral-GDS-mediated signaling to another effector. Attempts to isolate full length RalGDS protein for microinjection studies have not yet been successful. Since thyrocytes arrest in G 1 following intranuclear DNA injections and exhibit very low transfection efficiencies in transient assays, it will be necessary to generate stable lines overexpressing RalGDS and Ral in order to determine whether they exert similar effects on thyrocyte growth control.
In addition to Ha-and K-Ras, RalGDS also binds Rap1A (19), R-Ras (43), and TC21 (35), although binding to these molecules does not stimulate RalGDS activity in vitro (19). Several lines of evidence support the idea that TSH transduces its mitogenic signals through Ras, rather than one of these other proteins, to RalGDS or a related protein. First, purified Ras protein binds to RalGDS in thyroid cell extracts. Second, microinjection of the RalGDS RBD reduced DNA synthesis stimulated by Ras and TSH, suggesting that this domain binds to Ras in living cells. Third, similar to the effects of TSH, overexpression of Ras stimulates DNA synthesis in thyroid cells. It remains to be determined whether any of the other small G proteins which bind RalGDS exhibit similar effects. Additional support for signaling from Ras to RalGDS has been reported. In Xenopus oocytes, expression of the Ras-interacting domain of RGL interfered with Ras-mediated maturation (25). In NIH3T3 cells, this domain reduced Ras-mediated transformation (26), although in another study, expression of the RBD of RalGDS reportedly had no inhibitory effect on Ras-or TC21-stimulated transformation (35). Last, the observation that protein kinase A stimulates Ras-RalGDS association (24) further supports the hypothesis that Ras is capable of signaling through RalGDS and that this signaling pathway might predominate in the presence of cAMP.