Rig is a novel Ras-related protein and potential neural tumor suppressor

Abstract

The Ras superfamily consists of a large group of monomeric GTPases demonstrating homology to Ras oncoproteins. Although structurally similar, Ras-superfamily proteins are functionally diverse. Whereas some members exhibit oncogenic properties, others may serve as tumor suppressors. We have identified a novel Ras-related protein that suppresses cell growth and have designated it Rig (Ras-related inhibitor of cell growth). Overexpression of Rig inhibited Ras-mediated cellular transformation and activation of downstream signaling in NIH 3T3 cells. rig mRNA is expressed at high levels in normal cardiac and neural tissue. However, Rig protein expression is frequently lost or down-regulated in neural tumor-derived cell lines and primary human neural tumors. Moreover, expression of exogenous Rig in human astrocytoma cells suppressed growth. Rig has a C-terminal CAAX motif that codes for posttranslational modification by both farnesyl and geranylgeranyl isoprenoid lipids. Consequently, Rig may play a role in the cellular response to farnesyl transferase inhibitors. Rig bears 63% overall sequence homology to a recently described Ras-family member Noey2, a tumor suppressor in breast and ovarian tissue. Therefore, Rig and Noey2 may represent a new subfamily of Ras-like tumor suppressors.

Small, Ras-related GTPases form a large superfamily of structurally related proteins in mammalian cells (1). These proteins are typically regulated by guanine nucleotide binding and undergo C-terminal isoprenylation. Ras-superfamily proteins mediate pleiotropic cellular effects ranging from growth control to cytoskeletal rearrangements, various aspects of intracellular transport, cell survival, and apoptosis (2, 3). Although the Ras-, Ral-, Rit-, and Rho-subfamily proteins function as oncoproteins (4, 5), Noey2/Ahri1 and Rap1A have been described as tumor suppressors (6, 7). Thus, even closely related Ras-superfamily members can exert quite different biological effects.

The relationship between structure and function of Ras-related proteins is now sufficiently well understood that certain biological and biochemical characteristics may be inferred simply from an examination of their primary amino acid sequence. For example, the region of these proteins essential for effector interaction (the effector domain) has been identified and the residues important for particular effector interactions characterized (8). Furthermore, many Ras-related proteins contain a C-terminal CAAX motif (cysteine-aliphatic-aliphatic-X) that serves as a target for posttranslational isoprenoid lipid modification (9). It is now known that the X amino acid residue is critical for determining the type of covalent isoprenylation, farnesyl (F) (15-carbon) or geranylgeranyl (GG) (20-carbon). Typically, serine, phenylalanine, methionine, and cysteine residues are substrates for F transferases whereas GG transferases recognize leucine and phenylalanine residues (9, 10).

The vast majority of Ras-related proteins undergo GG isoprenoid lipid modification whereas the relatively rare F isoprenoid modification occurs on only a subset of Ras-superfamily proteins (H/K/N Ras, Rap2A/B, Rheb, and RhoB/E) (11). At least for Ras, posttranslational farnesylation plays a key role in effector binding (12, 13) and subcellular localization (14). The importance of farnesylation in Ras function is underscored by the fact that disruption of the CAAX motif prevents cellular transformation by oncogenic Ras (15). F transferase has been successfully targeted for clinical drug development (11) by compounds called F transferase inhibitors (FTIs), designed to prevent membrane localization and therefore activity of Ras oncoproteins. Although FTIs revert the Ras-transformed phenotype in vivo (16–18), their antitumor effects seem to be largely Ras-independent (19, 20). Consequently, the effort to characterize the antitumor action of FTIs and identify farnesylated non-Ras cellular targets of FTIs continues (21, 22).

We explored the possibility that many Ras-related proteins remain unidentified and used a bioinformatic approach to search the expressed sequence tag database for novel monomeric GTP-binding proteins. Here we describe a new member of the Ras-related protein family designated Rig (Ras-related inhibitor of cell growth). Unlike Ras, Rig actually inhibits growth, survival, and transformation. Moreover, Rig expression is frequently lost in primary human astrocytomas. We suggest that Rig is a novel, neural-specific tumor suppressor, which may function in a similar manner to Noey2/Arhi1, a related Ras-superfamily member and tumor suppressor in breast and ovary tissue (7). Because Rig can be modified by F isoprenoid lipids, it may serve as a clinically relevant target of FTIs.

Methods

Identification of rig.

We used the amino acid sequence of Noey2/Arhi1 (U96750) to query the expressed sequence tag database in an advanced tblastn search (http://www.ncbi.nlm.nih.gov/blast/). We identified a 198-aa, hypothetical Ras-related protein. Further analysis identified the same the gene on a bacterial artificial chromosome clone (AC006538) containing DNA from chromosome 19p13.3. The Rig protein was then compared with other Ras family members with the pileup and growtree programs of the GCG suite of sequence analysis tools.

rig Cloning and Expression Plasmid Construction.

The rig gene was isolated from IMAGE clone (3363561) by PCR with 5′-gggggatccatgccggaacagagtaacgattac-3′ forward and 5′-gcggaattctcacatgagggtgcatttgccc-3′ reverse primers containing BamHI and EcoRI restriction sites, respectively. The resulting PCR fragment was sequenced and subcloned into pGEX-2T, pZIP-NeoHABE, or pcDNAF (23) to make N-terminal-tagged glutathione S-transferase (GST)-, hemagglutinin (HA)-, or Flag-Rig fusion proteins. Flag-tagged rig was cloned into a tetracycline-inducible retroviral vector, pLRT (24). This plasmid (pLRT-Rig/Flag) was transfected into retroviral packaging Phoenix cells (25). rig point mutations were introduced with a QuickChange Site-Directed Mutagenesis kit (Stratagene).

Cell Culture.

Cells were cultured in DMEM supplemented with 10% calf serum (NIH 3T3) or 10% FBS (HEK 293T and Phoenix) and penicillin (100 units/ml)/streptomycin (100 μg/ml) at 37°C in a 10% CO₂ incubator. Human U251 astrocytoma and A673 peripheral neuroepithilioma tumor (PNET) cells (American Type Culture Collection) were cultured in RPMI media 1640 supplemented with 10% FBS and penicillin (100 units/ml)/streptomycin (100 μg/ml). Plasmid DNA was transfected into NIH 3T3, HEK 293T, and Phoenix cells by CaPO₄ precipitation (26).

Cell Survival and Transformation Assays.

NIH 3T3 cells were cultured in 60-mm² dishes and transfected with indicated plasmids as described (26). To measure the effects on cell growth and survival, transfected NIH 3T3 cells were selected in media supplemented with 500 μg/ml G418 (Life Technologies, Grand Island, NY). Focus-formation assays were performed on NIH 3T3 cells transfected with pCGNH-HRas(G12V), -Rap1A(Q63E) (27), and pcDNAF-Rig(wt) (wt, wild type) as indicated. Cell culture medium was refreshed every 2–3 days and foci were scored after 10–14 days with an inverted microscope. Both focus-formation and colony-growth assays were performed three times in duplicate.

Elk-1 Dual-Luciferase Assays.

NIH 3T3 cells were cultured on 6-well plates and cotransfected by CaPO₄ precipitation with Gal-Elk-1, 5X Gal-Luc, pCGNH-HRas(G12V) (28), and pcDNAF-Rig(wt). The plasmid pRL-TK was also included in all samples as an internal control. pRL-TK encodes the Renilla reniformis (sea pansy) luciferase under the control of the herpes simplex virus thymidine kinase promoter, providing constitutive low-level activity. Cells were incubated for 48 h before shifting to 1% serum overnight. Quiescent cells were lysed for 30 min at 25°C and luciferase activity was determined in a Luminometer with the Dual-Luciferase Reporter Assay system (Promega). This assay enables the sequential measurement of firefly (Photinus pyralis) and sea pansy (R. reniformis) luciferase activity from the same cell lysate. The constitutive activity of Renilla luciferase may be used as a measure of transfection efficiency as well as cell viability. Data are derived from the relative light units (RLUs) detected during the sequential measurement of firefly (P. pyralis) and sea pansy (R. reniformis) luciferase activity in cell lysates. The absolute background RLU values (in the absence of reporter gene expression) were negligible. Each data point is the mean and SE of the mean of at least three independent experiments in triplicate.

In Vitro Prenylation Assay.

Recombinant GST-fusion proteins were generated from pGEX2T constructs in XL1-Blue bacteria (Stratagene) as described (8). Equal amounts (1–2 μg) of glutathione Sepharose-coupled GST-fusion proteins were incubated in rabbit reticulocyte lysates containing 2.5 μCi [³H]mevalonate ([³H]mev), [³H]F-pyrophosphate ([³H]FPP), or [³H]GGPP for 1 h at 30°C. Affinity-bound proteins were isolated by centrifugation (10,000 × g) and resolved by SDS/PAGE. The gel was fixed in 10% methanol/10% acetic acid, washed, and soaked in Autofluor before drying and exposure to x-ray film.

Northern Blot Analysis.

Analysis of rig expression in normal human tissue was performed by probing a MultiTissue Northern blot (CLONTECH) with full-length rig, labeled with [³²P]dCTP (Prime-IT II, Stratagene). Determination of rig mRNA expression in neural tumor cell lines was performed essentially as described (23).

Generation of Rig Abs and Immunoblotting.

A peptide consisting of Rig residues 175–194 (corresponding to the C-terminal hypervariable domain region LNIDGKRSGKQKRTDRVKGK) was used to generate polyclonal Abs in rabbits by standard methods (Research Genetics, Huntsville, AL). This sequence was determined to be unique by database query and the antiserum did not cross react with HRas, KRas, Rap1A, RRas, TC21, or Rac1 (data not shown). Human normal striatum, low-, and high-grade astrocytoma tissue were boiled in 2% SDS, dounce-homogenized, and clarified by centrifugation (10,000 × g). Equivalent amounts of clarified lysates were resolved by SDS/PAGE. Endogenous Rig protein was detected by immunoblotting with a 1:2000 dilution of the polyclonal Rig Ab. Coimmunoprecipitation of Rig and Raf-1 N-terminal regulatory domain (RafREG), from HEK-293T lysates, was performed as described (23).

Generation of Neural Tumor Cells Harboring an Inducible Rig Expression Vector.

The ecotropic viral envelope receptor (29) was cloned into pBabe (30) and transfected into PA317 amphotropic packaging cells (31). Retroviral-conditioned medium was used to infect human U251 astrocytoma and A673 PNET cell lines (both negative for Rig protein expression; C.A.E. and G.J.C., unpublished observations). Cells were then selected in puromycin (5 μg/ml). The resistant colonies were pooled and the resultant cell lines, which were now susceptible to infection with ecotropic virus, were designated U251E and A673E, respectively. Conditioned supernatants, obtained from ecotropic packaging Phoenix cells (25) transfected with pLRT retroviral vectors (24), were then used to infect ecotropic receptor-positive human U251E and A673E cells. After selection in blasticidin (5 μg/ml)-supplemented media, polyclonal U251E and A673E (pLRT-Rig/Flag) cells were examined for doxycycline-induced Rig/Flag expression. For growth assays, 4 × 10⁴ cells were seeded into 60-mm² dishes under reduced serum conditions 24 h before induction. After trypsinization, cells were quantitated with a Coulter Counter, pelleted by centrifugation (10,000 × g), and lysed in RIPA buffer (0.15 mM NaCl/0.05 mM Tris·HCl, pH 7.2/1% Triton X-100/1% sodium deoxycholate/0.1% SDS) (23). Each data point is the mean and SE of the mean for at three experiments in duplicate or triplicate.

Results

Sequence Comparison of Rig and Ras-Related Proteins.

rig was identified in the human database expressed sequence tag (EST) during an attempt to identify novel, farnesylated, small GTPases. The 594-nt gene encodes a hypothetical Ras-related protein on chromosome 19p13.3. The rig gene was isolated by PCR with the IMAGE EST clone (3363561) as a template, and the nucleotide sequence was verified against that in the database. Amino acid sequence alignment of Ras-related proteins with the predicted amino acid sequence of Rig revealed significant homology (Fig. 1A). The overall sequence similarity varied with Rig being most homologous to the Noey2 tumor suppressor (63%). The greatest identity, between Rig and Ras-family proteins, exists within conserved motifs responsible for guanine-nucleotide and effector protein binding. In addition, Rig contains a C-terminal CAAX motif with the X being a methionine, suggesting that Rig will be a substrate for F transferase. From the aligned Ras-family protein sequences a distance matrix was created to generate a dendogram (Fig. 1B). In this graphical representation of sequence relatedness, Ras, Rap, Ral, and RRas subgroups form discrete branches of the Ras-superfamily tree as do Rig and the Noey2 tumor suppressor. Thus, based on bioinformatic evaluation of the primary amino acid sequence we predict that Rig is posttranslationally farnesylated, binds Ras and non-Ras effectors, and exhibits tumor suppressor-like properties.

Figure 1.

Rig sequence analysis. Primary amino acid sequence comparison of Rig, Noey2/Arhi1, and HRas by progressive, pairwise alignment was performed with the PILEUP program of the Wisconsin Package. (A) Multiple sequence alignment. Key elements are capitalized: (1) guanine nucleotide-binding domains; (2) effector binding domain; and (3) CAAX motif. (B) Dendogram of Ras-family proteins. Amino acid sequences were imported into the GROWTREE program of the Wisconsin Package to obtain graphical representation of peptide relatedness. Bar length is proportional to amino acid substitutions/peptide length.

Rig Expression Inhibits Cell Growth and Survival.

In an attempt to elucidate the biological function of Rig, retroviral expression vectors (pZIP-NeoHABE) encoding HA-tagged, wt, and Rig(S21N) mutant proteins were introduced into NIH 3T3 cells and antibiotic (G418)-resistant colony formation was monitored (26). The S21N mutation in Rig, located within the guanine nucleotide-binding motif, is analogous to the dominant-negative HRas(S17N) (32). In untransfected and Rig(wt)-transfected NIH 3T3 cells, no drug-resistant colonies formed whereas numerous colonies formed in Rig(S21N)- and vector-transfected cells (Fig. 2A). Similar results were achieved in other tumor-derived cell types (data not shown). Lysates from pooled colonies of NIH 3T3 pZIP-NeoHABE and pZip-NeoHABE-Rig(S21N) stable cell lines were analyzed for HA-tagged protein expression (Fig. 2B). Data from these experiments indicate that Rig potently suppresses cell growth.

Figure 2.

(A) Rig expression inhibits clonogenic cell growth. The effect of Rig expression on NIH 3T3 cell growth was analyzed by transfecting pZIP-NeoHABE plasmids (500 ng) encoding Rig (wt or S21N) proteins into by CaPO₄ precipitation. To select cells stably expressing Rig, transfected NIH 3T3 cells were cultured in media supplemented with 500 μg/ml G418. In NIH 3T3 cells transfected with Rig(wt) expression constructs, no colonies survived G418 selection, whereas numerous colonies grew in vector- and Rig(S21N)-transfected cells. The Rig(S21N) mutant supported NIH 3T3 cell growth, after G418 selection, comparable to vector-transfected cells. Similar results were achieved with a pBabe retroviral vector encoding Rig(wt) and culturing cells in puromycin-supplemented medium (data not shown). (B) Anti-HA immunoblotting of lysates from NIH 3T3 cells transfected with pZIP-NeoHABE and pZIP-NeoHABE-Rig(S21N). (C) Focus-formation assay. NIH 3T3 cells were cotransfected with 500 ng of pcDNA3.1, Rig(wt), or pCGNH-Rap1A(Q63E) and -HRas(G12V) (50 ng) expression vectors by CaPO₄ precipitation. Rap1A(Q63E), which has been shown to antagonize Ras transformation, was used as a control. Data are the percent of Ras-induced focus formation. Results are the mean and SD of three experiments in triplicate. (D and E) Dual-luciferase reporter assay. NIH 3T3 cells on 6-well plates were cotransfected with the following reporter constructs (per-well quantities): 100 ng of pRL-TK, 125 ng of Gal-Elk-1, and 250 ng of 5X Gal-Luc as well as 50 ng of pCGNH-HRas(G12V) and 1,000 ng of pcDNAF-Rig(wt) plasmids (indicated by +), cultured and lysed as described. The activity of firefly (P. pyralis) and sea pansy (R. reniformis) luciferase from cellular lysates were measured sequentially (see Methods). (D) Effect of Rig overexpression (1,000 ng of pcDNAF-Rig) on Ras-mediated Elk-1 activity. Data are represented as fold increase in Elk-1-mediated luciferase activity [in relative light units (RLUs)] of each group relative to vector. (E) Herpes simplex virus thymidine kinase-driven R. reniformis luciferase activity in lysates from NIH 3T3 cells in Fig. 2D. Bar height is proportional to R. reniformis luciferase activity (in RLUs) ×100.

Rig Antagonizes Ras-Induced Signaling and Transformation.

We examined the effect of Rig on Ras-mediated cellular transformation by cotransfecting their respective expression vectors into NIH 3T3 fibroblasts and monitoring focus formation (26). The number of pCGNH-HRas(G12V)-induced foci (1–2 foci per ng of plasmid DNA/60-mm² dish) was reduced by ∼50% in the presence of Rig (Fig. 2C). The inhibitory effect of Rig on Ras-mediated transformation was similar to that seen by Rap1A(Q63E) (Fig. 2C; ref. 8). Ras exerts a biological effect by interacting with a broad array of effector proteins thus activating numerous signal transduction cascades (33, 34). Other Ras-like GTPases, namely Rap and Rheb, are capable of antagonizing Ras by nonproductive competition with Ras effectors (8, 35). To determine whether Rig could antagonize Ras signaling, we used an Elk-1-dependent luciferase reporter as an Erk/MAPK activity readout to measure Ras-mediated signaling. The ability of Ras to transactivate Elk-1 was diminished by >50% by as little as 10 ng (data not shown) and completely ablated by 1 μg of Rig expression vector (Fig. 2D). Renilla luciferase activity was not affected under these same conditions (Fig. 2E), indicating that inhibition of Ras-signaling is not the result of nonspecific cell death. These data indicate that Rig, like Rap1A and Rheb, could antagonize Ras-mediated signaling and transformation.

The Association of Rig with Potential Effector Molecules in Vivo.

The similarity between the core effector domains of Rig and Ras suggests they may interact with common effector molecules. By analogy to both Rap1A and Rheb, the antagonistic effect of Rig on Ras signaling and transformation may be the result of nonproductive association with Ras effector molecules, such as Raf-1 (8, 36). To determine their ability to interact in vivo, FLAG epitope-tagged Rig and HA epitope-tagged Raf-1 kinase N termini (containing the Ras binding domains) were overexpressed in HEK 293T cells. Cell lysates were subjected to immunoprecipitation–Western analysis, where the RafREG was immunoprecipitated with anti-HA mAbs and immune complexes were probed for Rig by anti-FLAG immunoblotting (Fig. 3). When overexpressed in 293T cells, Rig interacted with RafREG (Fig. 3 Upper), indicating Rig may use N-terminal Raf-1 domains similar to Ras. Fig. 3 Lower shows levels of ectopic protein expression in 293T cell lysates before immunoprecipitation. FLAG-tagged Rig could also be immunoprecipitated from lysates with an anti-Raf-1 (C-12) Ab, suggesting that Rig can interact with the endogenous Raf-1 (data not shown).

Figure 3.

Effector binding. 293T cells grown in 100-mm² dishes were transfected with HA-tagging vector, pcDNAH, encoding RafREG (amino acids 1–330) and FLAG epitope-tagged pcDNAF-Rig(wt) expression plasmids (5 μg) by CaPO₄ precipitation. After ∼48 h, cells were lysed and clarified by centrifugation (10,000 × g). With monoclonal anti-HA Abs, RafREG was immunoprecipitated (IP) from lysates and immune complexes were resolved by SDS/PAGE. Association of RafREG and Rig was determined by anti-Flag immunoblotting (1B).

In Vitro Prenylation of Rig by F and GG Isoprenoid Groups.

The presence of a C-terminal CAAX motif, on Ras-related proteins, signals for posttranslational prenylation. As the terminal amino acid (X) of Rig is methionine, we speculated that Rig, like Ras, might be farnesylated. We used recombinant GST-Rig, -HRas (F), and -Rap1A (GG) (37) fusion proteins to analyze prenylation in vitro after incubation in rabbit reticulocyte lysates containing [³H]mev. Mevalonate is an essential metabolic precursor for both 15-carbon FPP and 20-carbon GGPP isoprenoids. Metabolic incorporation of [³H]mev was detected in recombinant GST-Rig, -HRas, and Rap1A proteins after incubation in reticulocyte lysate (Fig. 4 Top). Thus, Rig was covalently modified by prenylation, as predicted by the C-terminal CAAX motif. To determine the precise biochemical nature of Rig prenylation, GST-Rig was incubated in the presence of [³H]FPP or [³H]GGPP. Whereas HRas and Rap1A were farnesylated and geranylgeranylated, respectively, Rig was posttranslationally modified by both types of isoprenoid lipids (Fig. 4 Middle images). These data indicate that the Rig CAAX motif, CTLM, is a substrate for F as well as GG type prenyltransferases.

Figure 4.

In vitro prenylation assay. Equal amounts of recombinant GST-Rap1A, -HRas, and -Rig proteins were incubated in rabbit reticulocyte lysates containing [³H]mev, [³H]FPP, or [³H]GGPP for 1 h at 30°C. Reactions were stopped by the addition of SDS sample buffer and resolved by SDS/PAGE. Labeled (³H) proteins were detected by autoradiography. As expected, HRas and Rap1A were modified by FPP and GGPP, respectively. Rig undergoes posttranslational prenylation, being modified by F and GG isoprenoid groups.

rig mRNA Expression Is Limited to Brain and Heart Tissue.

To determine the expression pattern of rig, we probed poly(A)⁺ RNA from normal adult human tissues with a rig gene probe. A single band representing the rig transcript was detected in heart and brain tissue but in no other normal tissue analyzed (Fig. 5A). We next probed a multitissue slot-blot containing poly(A)⁺ RNA from numerous human tissues, including different brain and cardiac structures as well as fetal tissue. Consistent with the Northern analysis, strong hybridization of rig was exclusively detected throughout the heart and brain (C.A.E. and G.J.C., unpublished observation). Hybridization, albeit to a lesser extent, was detected with fetal heart and brain mRNA.

Figure 5.

(A) Multitissue Northern analysis. Poly(A)⁺ RNA from multiple human tissues was probed with [³²P]dCTP-labeled full-length rig. Hybridization and washing were performed according to manufacturer's protocol. (B) Northern blot analysis of neural tumor cell lines. Total RNA was isolated from cultured human U251 TC32, TC106, SH5Y, and murine NE115 NB cell lines and resolved by gel electrophoresis. Denatured RNA was transferred to BrightStar nylon membranes (Ambion, Austin, TX) and the full-length rig probe was hybridized in UltraHyb (Ambion) according to the manufacturer's instructions. Virtually no rig mRNA was detected in U251 and TC106 cells whereas expression was relatively abundant in the other cell lines. (C) Loss of rig expression in early- and late-stage astrocytomas. Protein lysates were prepared from human astrocytomas (early stage: 1–13; late stage: 14–20) and oligodendrogliomas (21–23). Proteins (equivalent amounts) were resolved by SDS/PAGE, and Western analysis was performed with purified anti-Rig polyclonal (1:2,000) and anti-actin (1:2,000) Abs. N, normal striatal tissue.

Loss of Rig Expression in Neural Tumor-Derived Cell Lines and Primary Human Astrocytomas.

Bioinformatic evaluation of the Rig protein sequence suggests that, as it is closest to Noey2, it may possess tumor-suppressor properties. If so, one might expect Rig to be down-regulated during tumor development. Because rig mRNA was abundantly expressed in normal brain tissue, we examined various neural tumor-derived cell lines for altered rig mRNA expression. rig mRNA was readily detected in human SH5Y and murine NE115 neuroblastoma cells. Lower levels of expression were detected in human A673 PNET and TC32 neuroepithelial cells. rig mRNA expression, however, was virtually absent in U251 astrocytoma and TC106 PNET cells (Fig. 5B).

Rig Protein Expression Is Lost in Many Primary Human Neural Tumors.

To obtain a more direct analysis of Rig protein expression in human tumors, we developed a Rig-specific polyclonal Ab. We used this reagent to perform Western analysis of homogenates from primary astrocytomas (Fig. 5C) (samples 1–20) and oligodendrogliomas (samples 21–23). Rig protein was undetectable or severely down-regulated in 6/13 (46%) early- and 5/7 (71%) late-stage astrocytomas (GBM, glioblastoma multiforme). The reduced Rig protein levels detected in some samples possibly resulted from normal tissue resected with the tumor. Rig protein was abundantly expressed in oligodendrogliomas. As an internal control, samples were reprobed for actin expression. Collectively, these results demonstrate that Rig protein is frequently down-regulated in primary human neural tumors and neural tumor-derived cell lines. Also, Rig expression is lost as astrocytomas progress to GBM.

Induced Rig Expression Decreases Neural Tumor Cell Growth.

Because of the inhibitory effect of Rig on cell growth and survival, we resorted to a tetracycline-inducible retroviral vector system whereby we could regulate the extent of Rig expression. Human U251E astrocytoma and A673E PNET cells, which stably express the ecotropic viral receptor, were generated (see Methods). These cells were subsequently exposed to ecotropic viral particles produced in Phoenix cells transfected with pLRT-Rig/Flag. Equal numbers of U251E and A673E (pLRT-Rig/Flag) cells were seeded onto 60-mm² dishes, in reduced serum media, before induction. Cells were trypsinized, enumerated at indicated time intervals, and lysed for immunoblot analysis. U251E and A673E (pLRT-Rig/Flag) cells underwent approximately one doubling every 24 h in the absence of induction (Fig. 6 A and B, white squares), which was reduced by >50% in the presence of doxycycline (Fig. 6 A and B, black squares). Interestingly, there were no significant differences in cell viability between induced and uninduced groups, as determined by trypan blue exclusion. By analyzing cellular lysates for Rig expression, we observed elevated and sustained levels of Rig in lysates from doxycycline-treated U251E and A673E (pLRT-Rig/Flag) cells but not uninduced cells (Fig. 6 Inset). Results from these experiments show a direct correlation between increased Rig expression and reduced growth rate in U251E AND A673E neural tumor cells.

Figure 6.

Induced rig expression alters tumor cell growth. Growth curves of human U251E astrocytoma cells (Left) and A673E PNET cells (Right) infected with a doxycycline-inducible retroviral vector, pLRT (circles), or encoding the rig gene (squares). Cells were cultured in 1% serum medium supplemented with (black) or without (white) 2 μg/ml of doxycycline. (Insets) Anti-Flag immunoblots (IB:FL) of U251E astrocytoma and A673E neural tumor cell lysates (−, no doxycycline; +, 2 μg/ml of doxycycline).

Discussion

By using a bioinformatics approach, we have identified a novel Ras-related protein, Rig. Rig shares >50% sequence similarity with the Ras and Rap subfamilies but has the greatest homology to the Noey2 tumor suppressor. From the primary amino acid sequence, we predicted that Rig would behave as a tumor suppressor, bind to Raf-1, and be modified by F isoprenoid lipids. These predictions of Rig function, based on domain conservation, have been confirmed experimentally. The data presented here highlight the practical application of bioinformatics not only in the identification of homologous proteins, but also in inferring functional properties.

Expression of wt Rig potently inhibited the growth and survival of numerous cell types in addition to antagonizing Ras-mediated signaling and transformation. Although it is difficult to fully exclude the possibility that nonspecific toxicity contributed to inhibition of Ras foci formation, the use of a dual-luciferase reporter system confirms that toxicity did not play a role in the inhibition of Ras signaling pathways. Rig-meditated growth inhibition was GTP-dependent, as the RigS21N mutant had no effect. There are two likely explanations for the antagonistic effect of Rig on Ras signaling. First, the similarity of the effector domains of Ras and Rig suggests that Rig can interact with at least some of the same effector molecules as Ras. We have confirmed this for Raf-1. If the interaction of Rig with Ras effectors were nonproductive then Rig would serve as a competitive inhibitor of Ras. Second, it is possible that Rig activates unique, non-Ras effectors to inhibit cell growth and antagonize Ras function. The conservative differences between the core effector domains of Rig and Noey2 suggest they use overlapping effector pathways, a notion consistent with their similar biological function.

Although Rig protein is expressed at high levels in normal brain tissue, it is frequently lost in primary human astrocytomas. Astrocytomas, the most common form of primary adult brain tumor (38), progress from a nonaggressive, relatively normal low-grade form to an infiltrative, highly malignant glioblastoma multiforme (GBM) (39). In this multistep carcinogenic process, tumor cells undergo morphological as well as genetic alterations associated with increasing malignancy (40). Loss of heterozygosity has been observed on both the p and q arms of chromosome 19 in astrocytomas and oligodendrogliomas, respectively (41–43). The rig gene is located on the p arm of chromosome 19 and Rig protein expression is diminished in astrocytomas and GBM, but not oligodendrogliomas. Furthermore, restoration of Rig expression in U251 astrocytoma and A673 PNET cells severely impaired growth. By Southern analysis we observed no obvious deletions or rearrangements of the rig gene in genomic DNA from primary neural tumors and neural tumor-derived cell lines (C.A.E. and G.J.C., unpublished observations). However, although the underlying mechanism(s) governing Rig protein loss in neural tumors is not known, epigenetic inactivation is possible. Loss of Noey2 expression in breast and ovarian tumors occurs by promoter hypermethylation (7), which may be a common means of transcriptionally silencing tumor-suppressor genes (44). Overall, these results strongly imply a tumor-suppressor role for Rig in human astrocytic tumors.

Here we show that Rig is also one of few cellular proteins subject to posttranslational F isoprenoid modification. Therefore, Rig can be added to the growing list of potential cellular targets of FTI antitumor compounds (11, 45). FTIs exhibit antitumor activity independently of oncogenic Ras, thus determining their cellular target is of considerable interest. The best-characterized non-Ras FTI target is RhoB (22), which exists in both farnesylated and geranylgeranylated forms. In the presence of FTIs, RhoB becomes exclusively geranylgeranylated and exhibits growth-inhibitory activity (21). It has been suggested that this is the primary mechanism of action of FTI compounds in tumors expressing RhoB. We demonstrate the modification of Rig by F or GG isoprenoid lipids, in vitro. Consequently, in the presence of FTI we would expect an increase in the levels of Rig GG in a cell. We are currently examining the possibility that GG Rig has enhanced growth-suppressive properties. If so, then Rig may mediate some of the antitumor effects of FTIs, in a manner reminiscent of RhoB.

In summary, a novel member of the Ras superfamily of monomeric GTPases was identified with bioinformatic tools to search the public genome sequence database. rig is a farnesylated and geranylgeranylated inhibitor of cell growth and Ras-mediated transformation. Although expressed at high levels in normal brain tissue, Rig protein is frequently down-regulated in astrocytomas. Moreover, the rig locus is a site of frequent loss of heterozygosity in advanced astrocytoma (41, 43). As Rig bears ∼63% homology to previously described tumor suppressor Noey2, we propose that Rig is a novel, neural-specific tumor suppressor.

Abbreviations

F

farnesyl

GG

geranylgeranyl

FTI

F transferase inhibitor

PP

pyrophosphate

PNET

peripheral neuroepithilioma tumor

GST

glutathione S-transferase

HA

hemagglutinin

RafREG

Raf-1 N-terminal regulatory domain

wt

wild type