Mas Oncogene Signaling and Transformation Require the Small GTP-Binding Protein Rac

Abstract

The Mas oncogene encodes a novel G-protein-coupled receptor that was identified originally as a transforming protein when overexpressed in NIH 3T3 cells. The mechanism and signaling pathways that mediate Mas transformation have not been determined. We observed that the foci of transformed NIH 3T3 cells caused by Mas were similar to those caused by activated Rho and Rac proteins. Therefore, we determined if Mas signaling and transformation are mediated through activation of a specific Rho family protein. First, we observed that, like activated Rac1, Mas cooperated with activated Raf and caused synergistic transformation of NIH 3T3 cells. Second, both Mas- and Rac1-transformed NIH 3T3 cells retained actin stress fibers and showed enhanced membrane ruffling. Third, like Rac, Mas induced lamellipodium formation in porcine aortic endothelial cells. Fourth, Mas and Rac1 strongly activated the JNK and p38, but not ERK, mitogen-activated protein kinases. Fifth, Mas and Rac1 stimulated transcription from common DNA promoter elements: NF-κB, serum response factor (SRF), Jun/ATF-2, and the cyclin D1 promoter. Finally, Mas transformation and some of Mas signaling (SRF and cyclin D1 but not NF-κB activation) were blocked by dominant negative Rac1. Taken together, these observations suggest that Mas transformation is mediated in part by activation of Rac-dependent signaling pathways. Thus, Rho family proteins are common mediators of transformation by a diverse variety of oncogene proteins that include Ras, Dbl family, and G-protein-coupled oncogene proteins.

G-protein-coupled receptors comprise a large family of cell surface receptors which mediate the actions of a diverse array of extracellular ligands, including hormones, neurotransmitters, phospholipids, odorants, photons, and purine nucleotides (see reference 76 for a review). G-protein-coupled receptors share a conserved predicted tertiary structure containing seven transmembrane domains. Intracellular signaling by these receptors is mediated by one or more members of the heterotrimeric G protein family. The G protein α subunit cycles between an inactive GDP-bound form and an active GTP-bound form, where the inactive α subunit is bound to the receptor and to a βγ heterodimer. Upon ligand stimulation, the receptor stimulates GDP-GTP exchange to promote the formation of the GTP-bound α subunit, which then dissociates from both the receptor and the βγ dimer. Both the Gα and βγ subunits then mediate activation of downstream effectors, including activation (Gα_s) or inhibition (Gα_i) of adenylyl cyclase or activation of phospholipase C (Gα_q or Gα₁₁). βγ dimers mediate a diverse array of effector functions, including activation of the Ras signal transduction pathway (76).

In addition to mediating a spectrum of normal physiological responses that include neurotransmission, metabolism, growth, and differentiation (76), there is also emerging evidence for the involvement of aberrant G-protein-coupled receptor function in cellular transformation and oncogenesis (20). For example, active mutants of the α1B-adrenergic receptor have been shown to cause transformation of NIH 3T3 cells (2). The serotonin 5HT1b receptor and the M₁, M₃, and M₅ subtypes of the acetyl cholinergic receptors were found to cause agonist-dependent transformation of NIH 3T3 cells (27, 38). Similarly, deregulated expression of Gα subunits (e.g., Gα₁₂, Gα₁₃, or Gα_q) has also been demonstrated to cause transformation of rodent fibroblasts (11, 19, 48, 78, 84). Although these Gα subunits are known to activate specific signaling pathways which may contribute to mitogenesis, whether involvement of these signaling pathways promotes the transforming actions of G-protein-coupled receptors has not been established.

The Mas oncogene was originally identified by its ability to render NIH 3T3 cells tumorigenic in nude mice (86). Further studies showed that Mas could promote the growth of rodent fibroblasts in serum-free medium (3). The predicted tertiary structure of Mas indicates that it functions as a G-protein-coupled receptor. Although Mas was once thought to be an angiotensin II receptor, recent studies argue against this possibility (3, 10). Thus, the ligand for Mas is presently unknown. Expression of Mas leads to activation of phospholipase C, indicating that Mas couples to the Gα_q/11 family of heterotrimeric G proteins (29, 54, 62). However, the signal transduction pathways activated by Mas, and those that cause cellular transformation, remain unknown.

Among the diverse signaling pathways that mediate G-protein-coupled receptor function are pathways that lead to activation of Rho family proteins. Rho proteins constitute a major branch of the Ras superfamily of small GTPases. To date, at least 11 distinct mammalian Rho family proteins have been identified: Rac1, Rac2, RhoA, RhoB, RhoC, RhoD, RhoE, RhoG, TC10, TTF, and CDC42Hs (41). Like Ras and Gα subunits, Rho family proteins also function as GDP-GTP binary switches (6). Rho proteins are activated by guanine nucleotide exchange factors (GEFs; Dbl family proteins) which stimulate formation of the active, GTP-bound Rho (83). Conversely, Rho proteins are inactivated by GTPase-activating proteins which stimulate GTP hydrolysis and formation of the inactive, GDP-complexed Rho. Additionally, Rho guanine nucleotide dissociation inhibitors inhibit dissociation of GDP and activation of Rho proteins.

Rho family proteins are regulators of diverse cellular processes (41, 69, 74). First, specific Rho family proteins regulate the organization of the actin cytoskeleton. CDC42Hs stimulates the formation of filopodia, whereas Rac1 induces lamellipodium formation and membrane ruffling, and RhoA causes the formation of actin stress fibers and focal adhesions (58, 70, 71). Second, Rho family proteins are regulators of gene expression. Both Rac1 and CDC42Hs are activators of the c-Jun NH₂-terminal kinases (JNKs; also known as stress-activated protein kinases) and p38 kinase. JNKs activate the c-Jun and ATF-2 nuclear transcription factors (16, 57, 60), whereas p38 activates ATF-2 (68). For example, a noncanonical AP-1 binding site in the c-jun promoter is bound and activated by heterodimers composed of c-Jun and ATF-2, and this site is stimulated preferentially by JNK-activating signals (4, 33, 77). Rac1, RhoA, and CDC42Hs activate the serum response factor (SRF), which together with ternary complex factors (TCFs) such as the Elk-1 transcription factor stimulates transcription from serum response elements present in the promoter of c-fos and other genes (34, 35). TCFs are activated by the p42 and p44 extracellular signal regulated kinases (ERKs), which together with JNKs and p38 comprise three distinct members of the mitogen-activated protein kinase (MAPK) family of proteins. Additionally, Rac1, RhoA, and CDC42 are activators of NF-κB (61, 73). NF-κB binding sites are present in a wide variety of promoters, including those that regulate the expression of genes that promote antiapoptotic responses (5).

Third, Rho family proteins are also regulators of cell proliferation. Rac1, RhoA, and CDC42 function is required for cell cycle progression and increased expression of the cell cycle regulator cyclin D1 (60, 81). Fourth, constitutive activation of RhoA, RhoB, Rac1, and CDC42Hs has been shown to cause tumorigenic transformation of rodent fibroblasts (43, 64–67), to promote invasion by T-cell lymphoma cells (Rac1) (28), and to promote increased motility and invasiveness (28, 56) of T-47D breast carcinoma cells (Rac1 and CDC42) (40). Finally, Rho protein function is necessary for the transforming actions of Ras and Dbl family oncogenes (42–44, 64, 66, 67). However, what aspect of Rho function contributes to cellular transformation remains to be resolved.

There is some evidence that Rho proteins are activated by heterotrimeric G proteins and G-protein-coupled receptors. First, the muscarinic and angiotensin II receptors, and Gα₁₂, Gα₁₃, and Gα_q, activate JNK, possibly via activation of Rac or CDC42Hs (14, 15, 32, 63, 88). Second, Rho proteins mediate the effect of G-protein-coupled receptors on the actin cytoskeleton in Swiss 3T3 cells. Lysophosphatidic acid activates RhoA-mediated induction of stress fiber and focal adhesions. Bombesin activates Rac1, leading to lamellipodium formation, whereas bradykinin activates CDC42Hs, leading to filopodium formation (46, 58, 70, 71). Finally, microinjection of constitutively activated mutants of Gα₁₂ or Gα₁₃ induced stress fiber and focal adhesion formation in Swiss 3T3 cells (7, 36). However, despite these connections, whether the transforming actions of G-protein-coupled receptors are mediated through activation of specific Rho family proteins has not been established.

Since G-protein-coupled receptors can cause activation of Rho family proteins, we have addressed the possibility that specific Rho family proteins also contribute to Mas transforming activity. We observed that Mas expression in NIH 3T3 cells caused a transformed phenotype that was similar to that seen with NIH 3T3 cells transformed by constitutively activated Rac1 and RhoA proteins and distinct from the transformed phenotype caused by activated Ras. Furthermore, our microinjection and immunofluorescence analyses showed that Mas induced lamellipodia similar to those induced by constitutively activated Rac1. We also observed that Mas caused activation of many of the same signal transduction pathways as activated Rac1: Mas upregulated the activity of JNK, p38, c-Jun, SRF, cyclin D1, and NF-κB. In contrast to Rac1, Mas activated the ERK/Elk-1 signaling pathway. Finally, dominant negative Rac1 blocked Mas transformation and some signaling. We conclude that Mas transformation is mediated in part by activation of Rac1 or a Rac-related protein. However, activation of Rac-independent pathways may also be important for Mas transformation.

MATERIALS AND METHODS

Molecular constructs.

Mas expression constructs were generated by subcloning the BamHI/NsiI fragment from the pM22 construct (genomic sequence) (87) into the BamHI site of the pZIP-NeoSV(x)1 retrovirus vector, where expression is under the control of a Moloney long terminal repeat (LTR) promoter, or the BamHI site of pCDNA3 (Invitrogen), where expression is under the control of the cytomegalovirus promoter. pAX142-mas was generated by converting the 5′ and 3′ BamHI sites into blunt ends with T4 DNA polymerase (GIBCO-BRL) and subcloning into the SmaI site of the pAX142 mammalian expression vector, where expression is under the control of the elongation factor 1α promoter (82). pCGN-mas was generated by PCR-mediated DNA amplification to create a 5′ BamHI site in frame following the ATG start codon of a hemagglutinin (HA) epitope tag, where expression is under the control of the cytomegalovirus promoter in the pCGN-hyg mammalian expression vector (75). pZIP-mas (cDNA sequence) was generated by digesting the StuI/PstI fragment of mas from the cDNA clone pM242 and subcloning it into the BamHI site of pZIP-NeoSV(x)1 (87). The pZIP-NeoSV(x)1 constructs encoding Ras(61L), H-Ras(WT), H-Ras(17N), Rac1(61L), Rac1(WT), Rac1(17N), RhoA(63L), RhoA(WT), RhoA(19N), Raf(340D), and pCGN-hyg constructs encoding Ras(61L) and Rac1(61L) have been described previously (43, 81). pAX142 constructs encoding Ras(61L), Rac1(61L), Rac1(WT), Rac1(17N), and RhoA(63L) were generated by converting the 5′ and 3′ BamHI sites from the BamHI fragments from the respective pZIP construct into blunt ends and subcloning the fragments into the SmaI site of pAX142.

Cell culture and transformation assays.

NIH 3T3 cells were grown in Dulbecco’s modified Eagle medium (DMEM) supplemented with 10% calf serum (GIBCO-BRL). Cells were transfected by the calcium phosphate precipitation technique as described previously (12). For the establishment of stable cell lines, transfected cultures were maintained in growth medium supplemented with 400 μg of G418 (Geneticin; GIBCO-BRL) per ml, and multiple (>100) G418-resistant colonies were pooled and used for the analyses described. For focus formation assays, 60-mm-diameter dishes were transfected and maintained in growth medium for 3 weeks and then stained with 0.4% crystal violet, and the number of transformed foci was quantitated by visual inspection.

Immunofluorescence analyses.

Actin and vinculin staining was performed as described previously (42). Briefly, cells were plated on glass coverslips, and two series of stainings were performed. Actin was stained with either tetramethyl rhodamine isothiocyanate (RITC)-phalloidin or fluorescein isothiocyanate (FITC)-phalloidin (Molecular Probes). The focal adhesion protein vinculin was stained with antivinculin monoclonal antibody 7f9 (a gift from Alexeu Belkin) (26) followed by FITC-conjugated goat anti-mouse immunoglobulin G (Jackson Immunoresearch Laboratories) or RITC-conjugated donkey anti-mouse immunoglobulin G (Chemicon International) (9).

Analysis of lamellipodium formation.

Analysis of lamellipodium formation was performed as described previously (81). Briefly, porcine aortic endothelial (PAE) cells were coinjected in the nucleus with pCDNA3-mas (100 μg/ml) and the Green lantern plasmid (Bethesda Research Laboratories) encoding the green fluorescent protein (25 μg/ml). Rac1(12V), CDC42(12V), and RhoA(12V) expression plasmids were microinjected in the nucleus at a concentration of 50 μg/ml. Subsequently, cells were starved in serum-free growth medium for 6 h, fixed, and processed as described previously (81).

Immunoprecipitation and in vitro MAPK assays.

ERK, JNK, and p38 kinase assays were performed as described previously (13, 81). Briefly, Cos-7 cells were transfected by using the Lipofectamine reagent (GIBCO-BRL) as described by the manufacturer. Cells were transfected with 1 μg of either pCGN-hyg, pCGN-ras(61L), pCGN-rac1(61L), or pCGN-mas plasmid DNA along with 1 μg of plasmid DNA encoding HA epitope-tagged ERK2 (provided by Michael Weber), FLAG epitope-tagged JNK1, or FLAG epitope-tagged p38 (provided by Michael Karin). Thirty-six hours after transfection, cells were serum starved in DMEM supplemented with 0.5% fetal bovine serum for 12 to 16 h. Cells were collected in 750 μl of lysis buffer containing protease and phosphatase inhibitors (80). Protein concentration was determined by using the Bio-Rad protein assay with bovine serum albumin as a standard, and 250 μg of lysate was used for immunoprecipitation. Epitope-tagged kinases were immunoprecipitated with either anti-HA (BabCo) or anti-FLAG (Kodak Eastman) antibodies and protein A/G-agarose beads (Santa Cruz Biotechnology).

Kinase assays were performed as described previously with myelin basic protein and glutathione S-transferase (GST)-conjugated c-Jun [GST–c-Jun(1-79)] and GST–ATF-2(1-254) as substrates for the ERK2, JNK1, and p38 kinase assays, respectively. Reactions were stopped with 2× Laemmli sample buffer and resolved by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE). Proteins were transferred to Immobilon membranes (Millipore), and the membranes were then exposed to film. Incorporated radioactivity was quantitated on a PhosphorImager (Molecular Dynamics). For standardization of transfection efficiency, 25 μg of total cell protein was analyzed by Western blot analysis as described previously (13, 81).

Transient expression reporter assays.

Transcriptional activation of luciferase gene expression constructs was performed as described previously (30). Briefly, NIH 3T3 cells were transfected with the indicated plasmid DNAs by the calcium phosphate coprecipitation method (12). The growth medium was then replaced with DMEM supplemented with 0.5% calf serum (GIBCO-BRL). Forty-eight hours after transfection, cells were lysed in 300 μl of luciferase lysis buffer (Amersham), and 25 to 50 μl of lysate was analyzed by using enhanced chemiluminescence reagents and a Monolight 2010 luminometer (Analytical Luminescence).

The reporter constructs Gal4–Elk-1 (30), 5X Gal4-Luc (72), (SREm)₂-Luc (81), cyclin D1-Luc (1), and HIV-Luc (25) have been described previously. Gal4–Elk-1 encodes a chimeric protein that contains the Gal4 DNA binding domain together with the transcriptional activation domain of Elk-1. The (SREm)₂-Luc construct contains the luciferase gene where expression is controlled by a minimal promoter with a mutated serum response element from the c-fos promoter. HIV-Luc contains the luciferase gene where expression is controlled by a minimum promoter and tandem copies of the NF-κB binding sites from the human immunodeficiency virus (HIV) LTR promoter. The cyclin D1-Luc construct consists of the luciferase gene where expression is controlled by sequences from −963 of the human cyclin D1. The Jun₂Luc reporter plasmid contains three tandem copies of the Jun/ATF-2 DNA binding motif present in the c-jun promoter introduced into HindIII and SalI sites in the minimal c-fos promoter in the Δ56dEFos-luciferase reporter plasmid using the following oligonucleotides: 5′ AGC TAG CAT TAC CTC ATC CC 3′ (top strand) and 5′TCG AGG GAT GAG GTA ATG CT 3′ (24).

RESULTS

Overexpression of Mas in NIH 3T3 cells has been shown to promote cell growth in serum-free growth medium and tumor formation when introduced into nude mice (3, 87). We observed that the transformed foci induced by coexpression of Mas were very distinct from those caused by activated Ras or Src. Instead, they were similar to those caused by constitutively activated mutants of Rac1 or RhoA as well those caused by Dbl family oncoproteins such as Dbl or Vav (42). As shown in Fig. 1A, in a primary focus assay, activated Ras(61L) induced large foci which contained highly refractile spindle-shaped cells, while activated RhoA(63L) or Mas induced foci which contained densely packed nonrefractile cells. However, the appearance of a Mas-induced focus was not identical to that of a RhoA-induced focus: foci caused by RhoA were large and diffuse, while the foci caused by Mas were more punctate. Furthermore, Mas focus-forming activity was much greater than that seen with either activated Rho or Rac. In particular, activated Rac1 mutants (12V, 61L, and 115I) do not show focus-forming activity in a primary NIH 3T3 focus formation assay. Instead, activating mutants of Rac1 form foci only after they are selected, thus enriching the cell population for cells which may express higher levels of the Rac protein (43).

FIG. 1.

Mas-transformed cells exhibit a transformed phenotype similar to those of Rac- and Rho-transformed cells. (A) Transformed foci from NIH 3T3 cultures transfected with pZIP-ras(61L), pZIP-rhoA(63L), and pZIP-mas. (B) Morphology of NIH 3T3 cells stably transfected with pZIP-NeoSV(x)1, pZIP-ras(61L), pZIP-rac1(115I), and pZIP-mas. Multiple (>100) G418-resistant colonies were pooled to establish the cell lines used for these analyses.

We next compared the transformed morphology of Mas-expressing cells with that of cells transformed by activated Ras, Rac1, or RhoA. For these analyses, NIH 3T3 cells were stably transfected with plasmid DNAs encoding either Ras(61L), RhoA(63L), Rac1(115I), or Mas or the empty pZIP vector control, and multiple (>100) G418-resistant colonies were then pooled. As we have described previously, Rac1(115I)- and RhoA(63L)-expressing cells retained the nonrefractile and adherent characteristics of nontransformed NIH 3T3 cells, while Ras(61L)-expressing cells were spindle shaped, less adherent, and highly refractile (Fig. 1B) (42, 43). We observed that Mas-transformed cells exhibited a morphology that was most similar to that of Rac- or Rho-transformed cells. Although Mas-transformed cells were slightly more refractile in appearance and exhibited a more elongated morphology than that seen with the Rac1- or RhoA-transformed cells, the appearance of Mas-transformed cells was still very distinct from that of Ras-transformed cells.

Coexpression of Mas with Raf causes synergistic enhancement of transforming activity.

We and others have shown that coexpression of activated Rho family proteins or GEFs with activated Raf causes a synergistic enhancement of focus-forming activity (42–44, 65, 66). Therefore, we determined if Mas could also cooperate with activated Raf. For these analyses, we used the weakly activated Raf(340D) mutant protein (21). As we have shown previously, transfection of expression vectors encoding either activated Rac1(115I) or Raf(340D) alone induced very few or no foci in NIH 3T3 cells (Fig. 2). However, the coexpression of activated Rac1 with Raf(340D) caused a greater than 30-fold enhancement of focus-forming activity. Similarly, coexpression of Mas with Raf(340D) caused synergistic enhancement of focus-forming activity (greater than threefold above additive). These results indicate that Mas, like Rac1, can cooperate with activated Raf-1 and cause synergistic focus-forming activity.

FIG. 2.

Like Rac1, Mas cooperates with Raf(340D) and causes synergistic focus-forming activity. NIH 3T3 cells were cotransfected with pZIP expression plasmids encoding the indicated proteins. One hundred nanograms of pZIP-mas cDNA and 1 μg of all other DNAs were transfected per 60-mm-diameter dish. The data are shown as mean ± standard error for triplicate plates and are representative of at least three separate experiments.

In addition to a synergistic enhancement of focus-forming activity, coexpression of activated Raf enhanced the size of foci and altered the appearance of cells within the foci induced by Rac1, RhoA, and Mas. The morphologies of cells within foci induced by coexpression of activated Raf with either Mas or activated Rac1 were similar to each other but distinct from those induced by coexpression of activated Raf and RhoA. The Raf-RhoA-induced foci contained elongated, refractile cells that were similar in appearance to cells in Ras-induced foci. In contrast, the Raf-Rac1- and Raf-Mas-induced foci contained cells that had a refractile appearance but lacked the elongated shape seen with Ras or Raf plus RhoA foci (data not shown).

Mas and Rac cause similar changes in actin cytoskeletal arrangement.

Immunofluorescence analysis was done on NIH 3T3 cells which stably expressed activated Ras, Rac1, and Mas to visualize actin stress fibers or focal adhesions (Fig. 3A). As we have reported previously, cells transformed by activated RhoA(63L) exhibited enhanced stress fibers and focal adhesions, whereas cells transformed by activated Rac1(115I) retained stress fibers and focal adhesions but also exhibited membrane ruffling (42). In contrast, Ras(61L)-transformed cells showed a loss of stress fibers and focal adhesions but also exhibited membrane ruffling. The actin cytoskeletal arrangement of Mas-transformed cells most closely resembled that of Rac-transformed cells. Mas-transformed cells retained stress fibers and focal adhesions and exhibited increased membrane ruffling.

FIG. 3.

Mas and Rac1 cause similar changes in the organization of the actin cytoskeleton. (A) Mas- and Rac-transformed cells retain stress fibers and show enhanced membrane ruffling. Stably transfected NIH 3T3 cell lines expressing the indicated proteins were subject to immunofluorescence analysis as described in Materials and Methods. Shown are actin stress fibers and focal adhesions stained with RITC-phalloidin and FITC-antivinculin antibodies, respectively. Membrane ruffles are indicated by arrowheads. (B) Like Rac1(12V), Mas induces membrane ruffles in PAE cells. PAE cells were microinjected with expression constructs encoding either Mas along with GFP (a and b) or Myc epitope-tagged Rac1(12V) (c and d) as described in Materials and Methods. Cells were serum starved following injection, and actin was stained with RITC-phalloidin (a and c). Microinjected cells were identified by expression of GFP (for Mas) (b) or anti-Myc antiserum (for Rac1) (d).

Our immunofluorescence analysis of Mas-transformed cells suggested that Mas activated Rac proteins. To address this further, we determined if microinjection of Mas could induce the same changes in actin cytoskeletal organization as activated Rac1. For these analyses, we microinjected expression vectors encoding Mas, Rac1(12V), RhoA(14V), and CDC42Hs(12V) into PAE cells. We observed similar results with Rac1(12V) as with the 61L and 115I activating mutants. As described previously, RhoA induced stress fiber formation (70) and CDC42Hs(12V) induced filopodium formation (data not shown), whereas Mas induced lamellipodium formation similar to that induced by Rac1(12V) (Fig. 3B). These results suggest that Mas activates Rac1 or a Rac-related protein rather than RhoA or CDC42.

Mas and Rac stimulate common signal transduction pathways.

If Mas transforms NIH 3T3 cells by activation of Rac or a Rac-related protein, then Mas should also stimulate the same signal transduction pathways as activated Rac. To evaluate this possibility, we used two approaches. First, we and others have found that expression of activating mutants of Rac1 and CDC42Hs increases the kinase activity of the MAPK family members JNK and p38 but not ERKs, while RhoA does not activate any of these MAPKs when assayed in NIH 3T3 or Cos-7 cells (16, 57, 60). Therefore, if Mas activates Rac1 or a Rac-related protein, we would predict that Mas would also increase JNK and p38, but not ERK, kinase activity. Second, activated Rac1 has been shown to stimulate transcription from a variety of promoter elements that include a mutated version of the serum response element of the c-fos promoter that no longer binds TCFs, the c-Jun/ATF-2-responsive element of the c-jun promoter, the NF-κB-responsive element of the HIV LTR promoter, and the cyclin D1 promoter (35, 61, 73, 81). Therefore, we determined if Mas also stimulated transcription from these promoter elements and, if so, whether dominant negative Rac1(17N) could selectively block activation of these promoters.

For analysis of MAPK activation, we cotransfected Cos-7 cells with plasmid DNA encoding Ras(61L), Rac1(61L), or Mas, together with plasmid DNA encoding epitope-tagged p38, JNK1, or ERK2/p42^MAPK. The epitope-tagged kinases were immunoprecipitated, and kinase activity was determined as described previously (13, 81). As shown in Fig. 4, Mas caused a 3.3-fold increase in JNK kinase activity, similar to the 4.2-fold increase induced by Ras(61L), whereas Rac1(61L) caused a 13.6-fold activation of JNK. Similarly, Mas caused a 6.5-fold increase in p38 kinase activity, which was comparable to the 5.8- and 6.8-fold increases caused by Ras(61L) and Rac1(61L), respectively. In contrast to Ras(61L), Rac1(61L) and Mas did not cause a significant stimulation of ERK2 activity (Fig. 4C). However, we did observe that Mas caused a reproducible but weak activation of ERK2 (less than 1.5-fold) that was never seen with Rac1. Thus, Mas may stimulate a Rac-independent pathway leading to activation of ERK2.

FIG. 4.

Mas and Rac are strong activators of p38 and JNK but not ERK. (A) Activation of ERK2 by Mas. Cos-7 cells were transfected with either pCGN-hyg (vector), pCGN-ras(61L), pCGN-rac1(61L), or pCGN-mas along with an HA epitope-tagged ERK2 expression vector. Immunocomplex kinase assays with myelin basic protein (MBP) as a substrate were performed following immunoprecipitation of HA-ERK2 (top panel). Fold activation (Act) (middle panel) of ERK was determined by PhosphorImager analysis and expressed relative to phosphorylation levels in vector-transfected cells. Twenty-five micrograms of lysate was resolved by SDS-PAGE, transferred to an Immobilon membrane, and subsequently probed with anti-HA antibody to ensure equivalent expression levels of HA-ERK2 (bottom panel). (B) Activation of JNK1 by Mas. Cos-7 cells were transfected as for panel A but with FLAG epitope-tagged JNK1. JNK1 kinase activity was determined with GST–c-Jun(1-79) as a substrate (top panel). Fold activation (middle panel) and JNK1 levels (bottom panel) were determined as described for panel A. Data in panels A and B are representative of at least three independent experiments in Cos-7 and NIH 3T3 cells, using pCGN and pCDNA3 expression plasmids. (C) Activation of p38 by Mas. Cos-7 cells were transfected as described for panels A and B but with FLAG epitope-tagged p38. p38 kinase activity was determined with GST–ATF-2(1-254) as a substrate (top panel); fold activation (middle panel) and p38 expression levels (bottom panel) were determined as described above. Data are representative of two separate experiments.

We next determined if, like Rac, Mas could stimulate transcription from reporter constructs where luciferase expression was controlled by minimal promoters containing recognition sites for SRF, c-Jun/ATF-2, and NF-κB. For these analyses, we cotransfected plasmid DNAs encoding Mas, activated Rac1, RhoA, and Ras along with the indicated reporter constructs. Mas, like activated Rac1, stimulated transcription from these Rac-responsive elements (Fig. 5A to C). Both Rac1(61L) and Mas caused a greater than 100-fold stimulation of SRF, 5-fold or greater activation of NF-κB, and 10-fold or greater stimulation of c-Jun/ATF-2. Additionally, we showed recently that activated Rac1, as well as RhoA, stimulated transcription from the cyclin D1 promoter (81). Like Rac1, Mas stimulated transcription from the cyclin D1 promoter (Fig. 5D). Thus, Mas and Rac1 activate signaling pathways that stimulate transcription from common DNA promoter elements.

FIG. 5.

Mas and Rac stimulate transcription from common promoter elements. NIH 3T3 cells were transfected with pAX142 (vector), pAX142-ras(61L), pAX142-rac1(61L), pAX142-rhoA(63L), or pAX142-mas along with luciferase gene reporter constructs for SRF transcriptional activity (A), NF-κB transcriptional activity (B), c-Jun/ATF-2 transcriptional activity (C), and cyclin D1 expression (D). Data shown are representative of at least three independent experiments using both pAX142 and pCDNA3 mammalian expression constructs.

We also determined if Mas could activate the Elk-1 transcription factor by cotransfection of a plasmid encoding the Gal4 DNA binding domain fused to the transcription activation domain of Elk-1 (Gal4–Elk-1) together with a second plasmid where luciferase gene expression is controlled by a minimum promoter that contains tandem Gal4 DNA binding sequences (5XGal4-Luc). In contrast to Rac1, Mas caused a slight (30-fold) activation of Elk-1 activity, while Ras(61L) caused a 180-fold increase in activity (Fig. 6). These results along with the slight increase in ERK2 activation observed in the kinase assay (Fig. 4C) indicate that Mas can also activate Rac-independent signaling pathways which may also contribute to Mas transformation.

FIG. 6.

Mas, but not Rac1, caused activation of Elk-1. NIH 3T3 cells were transfected with pAX142 (vector), pAX142-ras(61L), pAX142-rac1(61L), pAX142-rhoA(63L), or pAX142-mas along with Gal4–Elk-1 and the Gal4-responsive 5XGal4-Luc construct. Data shown are representative of at least three independent experiments using both pAX142 and pCDNA3 mammalian expression constructs.

To establish if activation of these signaling pathways by Mas was dependent on Rac1 function, we determined if Mas stimulation of transcription from SRF- or NF-κB-responsive promoters, or the cyclin D1 promoter, could be blocked by coexpression of dominant negative Rac1 (Fig. 7). For these analyses, we performed transient expression assays where Mas was expressed either alone or together with Rac1(WT) or Rac1 (17N). We observed that coexpression of Rac1(17N) decreased the ability of Mas to stimulate transcription from both the SRF and the cyclin D1 promoter reporter plasmids by 70% (Fig. 7A and B). Coexpression of Rac1(WT) had no effect or caused a slight enhancement of stimulation. Thus, signaling by Mas to SRF and the cyclin D1 promoter is dependent on Rac function. In contrast to the requirement for Rac1 for SRF and cyclin D1 activation, coexpression of dominant negative Rac1 did not inhibit Mas stimulation of transcription from the NF-κB reporter plasmid (Fig. 7C). Thus, although Rac1 has been shown to be an activator of NF-κB (61, 73), Mas activation of NF-κB does not appear to be dependent on Rac function.

FIG. 7.

Dominant negative Rac1 blocks Mas signaling. NIH 3T3 cells were cotransfected with pAX142-mas and either pAX142-rac1(17N) or pAX142-rac1(WT) along with luciferase gene reporter constructs for SRF (A), cyclin D1 (B), or NF-κB (C) expression. Data are expressed as the mean of the percentage of the activation in the Mas-plus-vector samples ± standard deviation of duplicate samples and are representative of at least two independent experiments.

Mas requires Rac, Rho, and Ras for transformation.

The data described above support the hypothesis that Mas activates Rac1. To address the role of Rac1 function in Mas transformation, we determined if Rac1 function is required for transformation by Mas. For these analyses, NIH 3T3 cells were transfected with the Mas expression plasmid either alone or together with expression plasmids encoding wild-type or dominant negative Rac1. Coexpression of dominant negative, but not wild-type, Rac1 caused a 60% reduction in Mas-induced foci (Fig. 8). Together with our actin cytoskeletal and signaling analyses, these results suggest that Mas transformation and signaling are mediated by activation of Rac or a Rac-related protein. Finally, we found that coexpression of dominant negative Ras and RhoA also impaired Mas focus-forming activity (Fig. 8). These results suggest that RhoA and Ras function may be required for Mas transformation.

FIG. 8.

Dominant negative Rac, RhoA, and Ras block Mas transformation. NIH 3T3 cells were transfected with pZIP-mas and the indicated wild-type (WT) and dominant negative Rho family proteins, and the focus formation assay was performed as described in Materials and Methods. Data are expressed as the mean of the percentage of the total number of foci in the Mas-plus-vector dishes ± standard error and are the average of six separate experiments performed in duplicate or triplicate.

DISCUSSION

The Mas oncogene was identified originally as a gene that encodes a novel G-protein-coupled receptor that caused tumorigenic transformation of NIH 3T3 cells (87). However, the signaling pathways that mediate Mas transforming activity have not been determined. We observed that Mas caused the appearance of transformed foci of cells that were similar to those caused by constitutively activated Rho family proteins. We determined that Mas and Rac1 induced similar changes in the actin cytoskeleton that included the induction of lamellipodia and membrane ruffles. Furthermore, like Rac1, Mas caused strong activation of the MAPK family members JNK and p38 but not ERK. Mas and Rac1 also stimulated transcription from the same DNA promoter elements, including NF-κB, SRF, c-Jun/ATF-2, and the cyclin D1 promoter. Finally, Mas transformation and some signaling were impaired by coexpression of the Rac1(17N) dominant negative protein. Taken together, these results strongly suggest that Mas transformation is promoted by activation of Rac or a Rac-related protein. However, since we did observe quantitative (potency of focus-forming activity) and qualitative (activation of ERK and Elk-1 and Rac-independent activation of NF-κB) differences between Mas and Rac1, we also suggest that Mas transformation may be mediated by Rac-independent signaling pathways as well.

Our first indication that Mas transformation may be mediated by the activation of Rho family proteins came from the appearance of the foci of transformed cells caused by Mas in the NIH 3T3 focus formation assay and the morphology of Mas-transformed cells. We and others observed that constitutively activated mutants of Rac1, RhoA, and RhoB caused the appearance of transformed foci that were very distinct from those caused by oncogenic Ras, activated tyrosine or serine/threonine kinases, or transcription factors (43, 64–67, 81). RhoA- or Rac-induced foci retain a nonrefractile appearance and consist of tightly packed clusters of cells, whereas Ras-induced foci are large and well spread and contain highly refractile spindle-shaped cells. Dbl family proteins, such as Dbl and Vav, that function as Rho GEFs cause the appearance of Rho-like foci (42). Similarly, foci induced by constitutively activated mutants of Gα₁₂ and G₁₃, which are activators of Rho, induce Rho-like foci (11, 37, 78, 79, 84, 85). Furthermore, effector domain mutants of Ras that no longer bind to or activate the Raf-1 serine/threonine kinase [e.g., Ras(12V, 40C)] but retain Rho-dependent functions also induce Rho-like foci (44). Finally, the morphology of cells stably expressing activated Rac1, RhoA, or Rho family activators, such as Dbl, Vav, Gα₁₂, or Gα₁₃, is distinct from that of cells expressing Ras. These cells retain a flat nonrefractile appearance which is unlike the highly refractile appearance of cells transformed by activated Ras (42–44). Thus, we have observed that the induction of Rho-like foci in NIH 3T3 cells and a lack of significant morphological transformation have been reliable indicators that transformation may involve activation of Rho family proteins.

In addition to the appearance of Rac-like foci, further evidence that Mas transformation is mediated by activation of a specific Rho family protein was provided by the cooperative transforming activity observed when Mas was coexpressed with activated Raf. We and others observed that coexpression of activated Raf-1 together with activated Rac1 or RhoA caused a synergistic enhancement of focus-forming activity (43, 64, 65, 67). Additionally, coexpression of activated Raf-1 with Rho activators such as activated Dbl family proteins, Gα₁₂, Gα₁₃, and Ras effector domain mutants which are defective in Raf binding causes synergistic enhancement of focus-forming activity (11, 42, 44). This cooperation between Raf and Rho family proteins is believed to reflect the fact that full Ras transforming activity is mediated by the coordinate activation of Raf and a Raf-independent pathway(s) that leads to activation of Rho family proteins.

Key evidence that Mas specifically activated Rac or a Rac-related protein was provided by our analyses of the actin cytoskeletal arrangement in NIH 3T3 cells which constitutively overexpress Mas. Mas-transformed cells showed enhanced membrane ruffling, which is a hallmark of Rac activation (71). Similarly, microinjection analyses in PAE cells showed that Mas induced the same changes in actin cytoskeletal organization as activated Rac, which were distinct from those caused by activated RhoA or CDC42. Therefore, in two separate cell types, Mas and Rac induced similar changes in the organization of the actin cytoskeleton.

Additional evidence implicating Rac or a Rac-related protein in Mas function was revealed by our comparison of Rac and Mas signaling activities. Like activated Rac1, Mas stimulated the strong activation of the MAPK family members JNK and p38, but not ERK, whereas RhoA(61L) is not an activator of any of these MAPK family members (16, 57, 60). Furthermore, both Rac1 and Mas stimulated transcription from a panel of reporter plasmids which contained Rac-responsive DNA elements, including the NF-κB-, Jun/ATF-2-, and SRF-responsive elements and the cyclin D1 promoter. Mas activation of SRF and cyclin D1 expression was blocked specifically by dominant negative Rac1. Interestingly, Mas activation of NF-κB was not blocked by dominant negative Rac1. Thus, although activated Rac1 has been shown to stimulate NF-κB transcription (61, 73), Mas activation of NF-κB may be mediated by a Rac-independent pathway.

We found that Mas focus-forming activity was greatly inhibited by coexpression of dominant negative Rac1. This result, taken together with our cytoskeleton and signaling analyses, strongly suggests that Mas signaling and transformation are mediated by its ability to activate Rac or a Rac-related protein. However, we observed that Mas transformation was also impaired by dominant negative RhoA, suggesting that RhoA function may also be necessary for full Mas transforming activity. This requirement may reflect the fact that activated Rac1 can activate RhoA function (71). Alternatively, Rho may be required for an autocrine loop that is essential for Mas transformation.

Although our actin cytoskeleton and signaling analyses implicate Rac1 as a necessary component of Mas function, we cannot exclude the possibility that Mas is an activator of an as yet to be identified Rac-related protein. A definitive demonstration that Rac proteins are activated by Mas will require an analysis showing that Rac1-GTP levels are elevated in Mas-transformed cells. However, our present efforts to perform this analysis has been hampered by the lack of anti-Rac antibodies that would be useful in such assays.

We found that dominant negative Ras also impaired Mas focus-forming activity, suggesting that Ras function may also be important for Mas transformation. The possibility that Ras is activated in Mas-transformed cells is suggested by our observation that Mas caused a reproducible, but weak, activation of ERK which was not caused by activated Rac1. Similarly, like activated Ras, we found that Mas, but not Rac1, caused activation of Elk-1 (Fig. 6). Elk-1 is phosphorylated and activated by ERKs (51). Additionally, activated Rac can also synergize with activated Raf to increase ERK2 activity (23). Thus, it is possible that Mas causes a weak activation of the Ras→Raf→MEK→ERK→Elk-1 pathway that contributes to Mas transforming activity. This idea may explain some of the differences that we have observed between Mas and Rac transforming activity: the far greater potency of Mas than of Rac1 in focus formation assays, the weaker synergy seen between Mas and activated Raf, and the more refractile and spindle-shaped appearance of Mas-transformed cells. G-protein-coupled receptors such as the M₁-muscarinic acetylcholine receptor have been reported to activate ERK1/2 by two mechanisms: activation of protein kinase C via Gα_q or increased phospholipase C activity and activation of Ras via release of βγ subunits (18, 22, 31, 45, 49; see reference 76 for a review). Further experimentation will be required to determine if Mas activates Ras and if it is mediated through activated Gα versus βγ subunits.

At present, the G protein(s) which couples Mas to Rac activation is not known. One logical candidate is Gα_q, a possibility supported by several observations. First, Mas has been shown to be an activator of phospholipase C (29, 54, 62), which is an effector of Gα_q (76). Second, bombesin receptor activation causes activation of phospholipase C (59), and activation of this receptor causes a Rac-dependent induction of lamellipodia (71). Third, the stimulation of two Gα_q-coupled receptors, the angiotensin II and M₁-muscarinic acetylcholine receptors, and activated Gα_q itself, leads to activation of JNK (15, 32, 88). Lastly, constitutively activated Gα_q causes transformation of NIH 3T3 cells (19, 39). However, we have found that microinjection of a constitutively activated Gα_q mutant did not cause induction of lamellipodia and instead stimulated stress fiber formation in porcine aortic endothelial cells (89). Therefore, while Gα_q may mediate some aspects of Mas function, it does not appear to be the link that connects Mas with Rac, at least in PAE cells. Other possible Gα subunits that may provide such a connection are Gα₁₂ and Gα₁₃. Both also exhibit Rho-like transforming activities in NIH 3T3 cells (11, 37, 84, 85) and have been shown to activate JNK (14, 63). However, microinjection analyses of activated Gα₁₂ and Gα₁₃ in Swiss 3T3 cells caused the induction of stress fibers rather than lamellipodia (7, 36). Thus, these two Gα subunits do not seem to provide the connection between Mas and Rac. Whether a novel Gα family protein is involved in the signaling pathway that promotes Mas activation of Rac remains to be determined. Alternatively, βγ subunits may mediate Mas activation of Rac. In Cos-7 cells, overexpression of βγ subunits potently induced JNK activity (17). Finally, it is likely that a Dbl family protein is required to mediate Mas activation of Rac. To date, several Dbl family proteins which have exchange activity for Rac1 (e.g., Tiam-1) have been described and represent possible candidates for mediating Mas activation of Rac (see reference 83 for a review). To address this possibility, we are presently evaluating whether dominant negative mutants of specific Dbl family proteins can block Mas signaling and transformation.

In rodents, Mas transcripts are expressed primarily in the brain, with the highest detected levels in the cerebral cortex, the hippocampus (dentate gyrus and CA3 and CA1 cell layers), the piriform cortex, and the olfactory bulb (8, 52, 55, 86). Expression of Mas is developmentally regulated and begins postnatally on day 1 in postmitotic neurons, continuing through adulthood. In the dentate gyrus and the CA1 layers in the hippocampus, Mas is expressed at a time in development when most neurons are postmitotic but have not begun extending their axons (52). Additionally, Mas mRNA expression is upregulated by seizure activity in the dentate gyrus and CA1 fields of the hippocampus (53). The selectivity and timing of Mas expression in plastic regions of the brain and its regulation by neuronal activity suggest that Mas may play a role in growth and plasticity in these regions. Therefore, Mas may function as a cell surface receptor to regulate synapse formation in the brain. The observation that Mas can activate Rac proteins, leading to lamellipodium formation in both fibroblasts and endothelial cells, may provide an insight into the function of Mas in the nervous system. Accumulating evidence suggests that Rac proteins are involved in growth cone formation in developing neurons (47, 50). These observations, taken together with the fact that Mas is expressed in developing neurons and in neurons which may be undergoing plastic changes, suggest a model where Mas may regulate growth cone formation by activation of Rac proteins. Experiments to determine if Mas can regulate growth cone formation in differentiating neurons will be necessary to address this possibility.