Ras Uses the Novel Tumor Suppressor RASSF1 as an Effector to Mediate Apoptosis*

Although activated Ras proteins are usually associated with driving growth and transformation, they may also induce senescence, apoptosis, and terminal differentiation. The subversion of these anti-neoplastic effects during Ras-dependent tumor development may be as important as the acquisition of the pro-neoplastic effects. None of the currently identified potential Ras effector proteins can satisfactorily explain the apoptotic action of Ras. Consequently, we have sought to identify novel Ras effectors that may be responsible for apoptosis induction. By examining the EST data base, we identified a potential Ras association domain in the tumor suppressor RASSF1. We now show that RASSF1 binds Ras in a GTP-dependent manner, both in vivo and directlyin vitro. Moreover, activated Ras enhances and dominant negative Ras inhibits the cell death induced by transient transfection of RASSF1 into 293-T cells. This cell death appears to be apoptotic in nature, as RASSF1-transfected 293-T cells exhibit membrane blebbing and can be rescued by the addition of a caspase inhibitor. Thus, the RASSF1 tumor suppressor may serve as a novel Ras effector that mediates the apoptotic effects of oncogenic Ras.

Ras proteins serve as a node in the transduction of information from a variety of cell surface receptors to an array of intracellular signaling pathways (1)(2)(3). Mutations that lock Ras protein in the active state are frequently found in ras genes in human cancers (4,5). Moreover, deletion of the activated ras gene from tumor cell lines impairs their tumorigenicity (6,7). Therefore, activated Ras proteins appear to play a key role in human cancer.
Activated Ras proteins mediate a broad range of biological effects associated with enhanced growth and transformation. These include the induction of DNA synthesis (8), reduced growth factor dependence (9), loss of contact inhibition (10), inhibition of terminal differentiation (11), resistance to apoptosis (12), enhanced motility (13), metastasis/invasion (14,15), and tumorigenic transformation (16,17). Yet Ras proteins may also induce growth inhibitory effects. These include senescence (18), necrosis (19), apoptosis (20 -22), and terminal differentiation (23). This versatility is facilitated by the ability of Ras to bind and activate a diverse array of effector proteins (1,24). Some of these effectors, for example Raf-1 and phosphatidylinositol 3-kinase, are known to be oncoproteins in their own right and have a well characterized enzymatic activity (25)(26)(27)(28)(29). Other members of the Ras effector family are less well characterized (2,30,31). Moreover, it seems certain that there are Ras effectors remaining to be identified. Consequently, the mechanisms behind Ras mediated growth inhibition and death remain poorly understood.
The observation that oncoproteins can promote cell death as well as transformation has led to the hypothesis that the signaling pathways that drive death and proliferation are tightly coupled to protect against oncogenic transformation (32,33). Understanding how Ras subverts this balance of life and death in a successful tumor is critical to understanding the role of Ras in human cancer.
There are two likely mechanisms behind Ras-mediated growth inhibition. First, certain Ras effectors may play a dual role, being capable of either inducing or inhibiting growth. This seems to be the case with Raf-1. Here, moderate activation can be shown to promote growth, but excessive, prolonged activation may cause growth arrest and senescence (34,35). Second, certain Ras effectors may be specialized to induce only growth inhibition. These effectors would have to be rendered ineffective during the development of the tumor. In effect, these proteins would serve as Ras activated tumor suppressors.
A novel tumor suppressor, RASSF1, has recently been identified, which has a putative Ras association (RA) 1 domain and produces two main splice variants, A and C (36). Here, we show that the expression of RASSF1 is frequently down-regulated in ovarian tumor cell lines. Moreover, the C splice variant of RASSF1 binds Ras in a GTP-dependent manner and mediates a potent, Ras-dependent apoptosis. Our results provide a mechanistic explanation for the pro-apoptotic functions of Ras and how these functions may be subverted in the development of human tumors.

EXPERIMENTAL PROCEDURES
Cloning and Vectors-The EST data base was searched with Advanced tBLASTn using the amino acid sequence of the Nore1 RA domain. An RA-containing protein was identified that was later described as RASSF1C (36). The gene was cloned by PCR from the IMAGE consortium clone (632948; GenBank TM accession number AA205984) as a BamHI/EcoRI PCR fragment using primers (gacggatccatgggcgaggcggaggcgcc) and (gacgaattcacccaagggggcaggcg). The isolated RA domain was cloned as a BamHI/EcoRI PCR fragment using an internal 5Ј oligomer (gacggatcctctcaagctgagattgag). pZip-Neo SV(X)1 HA is a modified form of pZip-Neo SV(X)1 (37). This version has the internal EcoRI site deleted and the cloning site modified from a single BamHI site to a BamHI/HindIII/EcoRI sequence downstream of an HA tag. (The frame is GGA TTC). PCDNAFLAG is a version of pCDNA3 that has been modified in a similar manner with an upstream FLAG epitope. The RA domain was cloned in a modified version of pMal (New England Biolabs) in which we have changed the orientation of the EcoRI/BamHI sites in the cloning site to BamHI/EcoRI. Ras/RASSF1 Binding Assays-Maltose-binding protein (MBP) fu-* The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. sion proteins were prepared essentially as described previously (38) and purified on MBP-conjugated Sepharose beads. The fusion proteins were quantified by SDS-polyacrylamide gel electrophoresis followed by Coomassie staining and a comparison with known standards. Recombinant Ras protein was prepared as described previously (38). Binding assays were performed at 4°C for 2 h in phosphate-buffered saline/25 mM MgCl 2 and washed four times before Western analysis using the 146 Ha-Ras monoclonal (Quality Biotech).
In vivo association was determined by co-transfecting pCGNHA (39) Ha-Ras12v or wild type with pCDNAFLAG RASSF1 in 293-T cells. After 48 h, the cells were lysed in EDTA-free RIPA buffer (38), immunoprecipitated with anti-HA (BabCo) and subjected to Western analysis using the M2 anti-FLAG monoclonal (Kodak). Western blots were developed using an ECL kit (Amersham Pharmacia Biotech).
Cell Culture-NIH 3T3 cells were propagated in Dulbecco's modified Eagle's medium and 10% calf serum (Life Technologies, Inc.). 293-T cells were grown in Dulbecco's modified Eagle's medium and 10% Fetal Calf serum. Cells were transfected with the calcium phosphate precipitation technique described previously (40). z-VAD-fmk assays were performed by adding the drug at 30 M to the cells immediately after transfection. The drug concentration was maintained during subsequent changes of medium.
Northern Analysis-The RASSF1C cDNA clone was random primed with [ 32 P]dCTP and hybridized to a multiple tissue Northern blot (CLONTECH) using standard procedures. Total RNA was prepared from ovarian tumor cell lines using gaunidinium isothiocyanate lysis followed by cesium chloride gradient purification and run on an 0.8% formaldehyde gel. The gel was transferred to nylon membrane and after hybridization was washed to high stringency before autoradiography.

RASSF1 Has the Characteristics of a Novel Ras Effector-
Many Ras effectors share a conserved structural domain that is responsible for mediating the interaction with Ras. This is known as the Ras association or RA domain. (41). In an attempt to discover novel Ras effectors responsible for Ras-mediated growth inhibition, we conducted a data base search for potential RA domain-containing tumor suppressors. We identified a novel protein that possesses an RA domain. Analysis of genomic sequence data bases showed that the gene is localized to the 3p21.3 region of the genome. This region is frequently deleted or rearranged in human lung and ovarian carcinomas (42). The genomic structure of the gene and further analysis of transcripts present in the EST data base suggested that the gene could produce three transcripts.
During the preparation of this manuscript, the gene was identified as a tumor suppressor and designated RASSF1. Production of three separate transcripts A, B, and C, was confirmed (36). We cloned splice variant C of RASSF1 (which we will designate RASSF1C) as well as the isolated RA domain from the IMAGE consortium clone (632948) using PCR. The clones were sequenced to confirm fidelity before use.
To determine whether RASSF1C is a Ras effector, we examined its Ras binding properties. Ras is a small G protein and shuttles between an inactive, GDP-bound state and an active, GTP-bound state. Only the active, GTP-bound form of Ras adopts the appropriate conformation for effector binding (43). Therefore, if RASSF1C is a Ras effector, then its RA domain should bind GTP but not GDP Ras. We cloned the RA domain into a MBP fusion vector, pMal (New England Biolabs). The MBP-RASSF1(RA) fusion protein was induced in XL1 Blue bacteria (Stratagene) and purified on maltose-conjugated Sepharose beads. We then used the fusion protein as an affinity reagent to precipitate recombinant GDP/GTP Ha-Ras from solution in vitro (Fig. 1a). The precipitate was then subjected to Western analysis using an Ha-Ras antibody. The Ras binding domain of Raf-1 was used as a positive control. The RA domain of RASSF1 clearly binds Ras in a GTP preferential manner.
To confirm that RASSF1C could interact with Ras in vivo, RASSF1C was cloned into a pCDNA3 variant (Invitrogen), which adds a FLAG epitope tag to the amino-terminal end of the protein. The RASSF1C construct was co-transfected into 293-T cells with HA epitope-tagged activated or wild type Ha-Ras. Activated forms of Ras are typically more than 70% GTP bound in the cell, and wild type Ras is typically only 5% bound to GTP. After 48 h, the cells were lysed and immunoprecipitated with HA-conjugated Sepharose beads (BabCo). The immunoprecipitate was then subjected to Western analysis with an anti-FLAG antibody (Sigma). We found that RASSF1C coprecipitated preferentially with activated, GTP-bound Ras (Fig. 1b). The direct, GTP-dependent association of RASSF1C with Ras confirms this protein as a new candidate Ras effector.

RASSF1 Is Expressed in Most Normal Tissue but Infrequently in Ovarian Tumor Cell Lines-Northern analysis of
RASSF1 expression showed that the mRNA is present in most normal tissue (Fig. 2a). As RASSF1 has been mapped to a region of the genome that is frequently deleted or rearranged in lung and ovarian tumors (42), we examined a series of ovarian carcinoma cell lines for RASSF1 expression by Northern blot. We found that six of nine transformed ovarian cell lines had lost the expression of RASSF1 (Fig. 2b).
RASSF1C Mediates Ras-dependent Apoptosis-To examine the biological role of RASSF1C, we attempted to make stable cell lines over-expressing RASSF1C. RASSF1C was cloned into an HA-tagged version of pZIP-Neo SV(X)1 (37) and transfected into NIH 3T3 cells. The cells were then selected in G418. No colonies survived the selection in the RASSF1C-transfected dishes (Fig. 3a); moreover, co-transfection with activated Ras failed to rescue the cells (data not shown). Therefore, deregulated expression of RASSF1C inhibits cell growth and survival. As we could not study the action of RASSF1C in stable cell lines, we examined the effects of RASSF1C in transient transfection experiments using 293-T cells. The cells were transfected with RASSF1C and examined for growth effects at 72 h. The right hand panel of Fig. 3a shows that 293-T cell growth is inhibited by RASSF1C.
If RASSF1C is a Ras-activated tumor suppressor, then the presence of activated Ras should stimulate its biological activity. The effects of Ras on RASSF1C-mediated growth inhibition were determined by co-transfecting 293-T cells with RASSF1C and Ha-Ras mutants. Activated Ha-Ras (G12V) dramatically stimulated the growth inhibitory effects of RASSF1C (Fig. 3b). This stimulation was dependent upon an intact effector domain, as an effector mutant (44) (Ha-RasG12V/E37G) was unable to activate RASSF1C. The presence of a dominant negative form of Ha-Ras (Q61L/C186S) served to completely block the growth inhibitory effects of RASSF1C. Thus, the growth inhibition appears to be Ras-dependent.
To examine the mechanism of the growth inhibition, we compared the RASSF1C transfectants with cells transfected with pCDNA Fas. Fas is a well characterized inducer of apoptosis (45). Cells transfected with RASSF1C and Fas exhibited similar morphological changes such as membrane blebbing, a hallmark of apoptosis (Fig. 3c) (46). Apoptosis requires the activation of caspase proteases (47). To confirm that apoptotic cell death was occurring, the transfection experiments were repeated in the presence of the caspase inhibitor z-VAD-fmk (Calbiochem). In these experiments, the ability of RASSF1C to impair cell growth was severely reduced (Fig. 3c). DISCUSSION Recently, our understanding of the role of oncogenes has begun to change to incorporate the concept that they not only provoke transformation but may also induce growth inhibition and death. It appears that the signaling pathways for growth and for death are linked, such that deregulation of growth pathways can cause the activation of growth inhibitory pathways (32,33). This struggle between the forces of transformation and death persists even in a successful tumor, where high levels of apoptosis can still be detected (48,49). Understanding the mechanisms by which these controls are mediated, and ultimately subverted, may give rise to new avenues of cancer therapy.
Although Ras oncoproteins were initially characterized as suppressors of apoptosis (12,50) it is now clear that they also have the ability to promote apoptosis (20,21). We now provide part of the explanation for this apparent paradox by identifying the first component of a specific Ras/apoptosis pathway, the tumor suppressor RASSF1.
The tumor-suppressing properties of RASSF1 have been described recently (36). The gene produces three splice variants, A, B, and C. RASSF1A expression, and to a lesser extent RASSF1C expression, is impaired in a number of lung tumor cell lines (36). We now show that RASSF1C is frequently downregulated in ovarian tumor cell lines. It will be interesting to determine whether the greater incidence of RASSF1C loss in ovarian versus lung tumor cell lines reflects tissue-dependent differences in biological activity of the isoforms.
All three splice variants of RASSF1 contain the putative RA domain. Here, we show that this domain does indeed code for an RA domain, as it binds directly to Ha-Ras in a GTP-dependent manner in vitro. Moreover, RASSF1C preferentially binds the activated form of Ras in vivo.
RASSF1C has an endogenous ability to promote apoptosis in 293-T cells. However, this activity is dramatically stimulated by activated Ras. The stimulation is dependent upon the pres- ence of an intact Ras effector domain. Furthermore, a dominant negative form of Ras inhibits the apoptotic inducing effects of RASSF1C. As RASSF1C binds Ras in a GTP-dependent, effector domain-dependent manner, it meets the minimal criteria for a potential Ras effector. Moreover, unlike many candidate Ras effectors, we have shown a biological activity for RASSF1 that may be modulated by Ras. Thus, we present strong evidence that the RASSF1 tumor suppressor may be a Ras effector.
The growth inhibition assays were performed mainly in 293-T cells. This cell system is not typically used for the study of Ras biological effects (although it has been used extensively to study Ras-mediated signaling); this is because the cells are already transformed, and typical bioassays for Ras involve studying the induction of transformation. Here, we studied growth inhibition, a novel effect of Ras. Therefore, we felt the use of 293-T cells to be entirely appropriate. However, we intend to extend our studies into ovarian epithelial cell lines to confirm the physiological relevance of our results. The fact that RASSF1C also inhibits the growth of NIH 3T3 cells and MCF7 human breast tumor cells (data not shown) suggests that the biological effects of RASSF1 are consistent between cell systems.
Previous reports have shown that Ras-mediated apoptosis is p53-independent, can be promoted by inhibitors of NF-B (20), and can be suppressed with activators of NF-B (22). As we found that RASSF1C-mediated apoptosis is also p53-independent (in 293-T cells p53 should be inactivated by SV40 large-T antigen), we intend to examine the connections between the RASSF1 pathway and NF-B in subsequent studies.
The primary amino acid sequence of the RASSF1C gives few immediate clues to the mechanism behind its action. At first sight, the carboxyl-terminal sequences suggest a potential CAAX motif for protein prenylation (51). However, the amino acid immediately after the cysteine is a proline; this does not fit the rule that two aliphatic residues are required at this position, and the protein is not prenylated in our hands (data not shown). RASSF1 bears some homology to an as yet uncharacterized Ras effector, Nore1 (52). Thus, RASSF1 may be just the first of a family of related proteins that serve as Ras-regulated tumor suppressors.
In summary, we show that one of the isoforms of the novel tumor suppressor RASSF1 can function as a novel Ras effector, which may be used by Ras to mediate apoptotic cell death. Activated Ras enhances RASSF1C-induced cell death, and dominant negative Ras blocks it. Cells that acquire a Ras mutation and express RASSF1 will be predisposed to die by apoptosis. Therefore, the loss or inactivation of RASSF1 may be a critical component of the evolution of Ras-dependent tumors.