Oncogene (1998) 17, 1395 ± 1413  1998 Stockton Press All rights reserved 0950 ± 9232/98 $12.00 http://www.stockton-press.co.uk/onc Increasing complexity of Ras signaling

Sharon L Campbell1,4 Roya Khosravi-Far2, Kent L Rossman1, Geo€rey J Clark6 and Channing J Der3,4,5

1Department of Biochemistry and Biophysics; 2Department of Biology, MIT, Cambridge, Massachusetts 02139; 3Department of Pharmacology; 4Lineberger Comprehensive Cancer Center; 5Curriculum in Genetics and Molecular Biology, University of North Carolina, Chapel Hill, North Carolina 27599; 6Department of Cell and Cellular Cancer Biology, NCI/NIH, 9610 Medical Center Drive, Rm 307, Rockville, Maryland 20850-3300, USA

The initial discovery that ras genes endowed retroviruses Introduction with potent carcinogenic properties and the subsequent determination that mutated ras genes were present in a This review will summarize three major emerging themes wide variety of human cancers, prompted a strong that have refocused our perceptions on how Ras suspicion that the growth-promoting actions of mutated functions in signal transduction. First, it has become Ras proteins contribute to their aberrant regulation of clear that the role of Ras in recognition and activation of growth stimulatory signaling pathways. In 1993, a one its downstream targets, the Raf-1 serine/threonine remarkable convergence of experimental observations kinase, is much more complex than originally envisioned. from genetic analyses of Drosophila, S. cerevisiae and C. Earlier studies suggested that the primary role of Ras elegans as well as biochemical and biological studies in was to localize Raf-1 to the membrane where other mammalian cells came together to de®ne a clear role for factors were involved in the activation of Raf-1 kinase Ras in signal transduction. What emerged was an elegant activity. However, more recent data suggests that Raf-1 linear signaling pathway where Ras functions as a relay activation is a complex multi-step process, where Ras is switch that is positioned downstream of cell surface involved in the initial stages of activation, and possesses receptor tyrosine kinases and upstream of a cytoplasmic a direct role in both membrane localization and cascade of kinases that included the mitogen-activated activation of Raf-1. The recognition and pre-activation protein kinases (MAPKs). Activated MAPKs in turn of Raf-1 by Ras appears to be accomplished through regulated the activities of nuclear transcription factors. binding interactions between multiple sites in Ras and at Thus, a signaling cascade where every component least two distinct domains of Raf-1. Second, the Raf-1 between the cell surface and the nucleus was de®ned serine/threonine kinase is not the sole downstream and conserved in worms, ¯ies and man. This was a e€ector of Ras. Earlier proclamations that Ras served remarkable achievement in our e€orts to appreciate how simply as an activator of Raf-1 were clearly premature. the aberrant function of Ras proteins may contribute to Instead, what has emerged is that Ras utilizes a multitude the malignant growth properties of the cancer cell. of functionally diverse e€ector targets which are likely to However, the identi®cation of this pathway has proven to transduce signals through multiple pathways. Third, the be just the beginning, rather than the culmination, of our Ras subfamily consists of an expanding group of understanding of Ras in signal transduction. Instead, we GTPases that show high homology to Ras yet display now appreciate that this simple linear pathway represents distinct signaling properties. This, most likely re¯ects the but a minor component of a very complex signaling di€erential interaction between these Ras-related pro- circuitry. Ras signaling has emerged to involve a teins and additional cellular factors, regulators and complex array of signaling pathways, where cross-talk, e€ectors. feedback loops, branch points and multi-component Targeting components of the Ras signaling pathways signaling complexes are recurring themes. The simplest has been proposed as one approach for the development concept of a signaling cascade, where each component of anti-Ras drugs for cancer treatment. The realization simply relays the same message to the next, is clearly not that our comprehension of the complexities of Ras the case. In this review, we summarize our current signaling is quite limited, and simple-minded, may understanding of Ras signal transduction with an discourage any serious consideration that this is a fruitful emphasis on new complexities associated with the direction for drug discovery at this time. Thus, there are recognition and/or activation of cellular e€ectors, and some who are of the opinion that such e€orts are the diverse array of signaling pathways mediated by premature and must await a complete knowledge of this interaction between Ras and Ras-subfamily proteins with signaling circuitry. However, evidence that intervention multiple e€ectors. of Ras signaling at multiple points can signi®cantly impact the ability of Ras to cause cellular transformation, argues Keywords: Ras; review; signal transduction; e€ector; that a full unveiling of all the complexities of Ras signaling Ras sub-family will not be required before successful targeting of Ras signaling pathways for drug discovery can be achieved.

Ras is a point of convergence for diverse extracellular signal-stimulated pathways The three human ras genes (H-ras,N-ras and K-ras) Correspondence: SL Campbell encode four 188 ± 189 amino acid (p21 proteins) that Increasing complexity of Ras signaling SL Campbell et al 1396 function as GDP/GTP regulated switches (H-Ras, N- Ras activation of the Raf4MEK4MAPK kinase Ras, K-Ras4A and K-Ras4B) (Barbacid, 1987; Bourne cascade. Ras is a point of convergence for many et al., 1990; Boguski and McCormick, 1993; Quilliam signaling pathways (Khosravi-Far and Der, 1994). Ras et al., 1995). The two forms of K-Ras diverge solely in proteins are activated transiently in response to a their COOH-terminal 25 amino acids as a consequence diverse array of extracellular signals such as growth of alternate exon utilization. Ras proteins are factors, cytokines, hormones and neurotransmitters positioned at the inner surface of the plasma that stimulate cell surface receptors and include membrane where they serve as binary molecular receptor tyrosine kinases (RTKs), non-receptor tyr- switches to transduce extracellular ligand-mediated osine kinase-associated receptors, and G protein- stimuli into the cytoplasm to control signal transduc- coupled seven transmembrane receptors (SRs) (Satoh tion pathways that in¯uence cell growth, di€erentiation and Kaziro, 1992; Khosravi-Far and Der, 1994). The and apoptosis (Satoh and Kaziro, 1992; Khosravi-Far best characterized Ras-mediated signal transduction and Der, 1994). Ras biological activity is controlled by pathway involves the activation of peptide mitogen- a regulated GDP/GTP cycle (Figure 1). The intrinsic stimulated RTKs such as the epidermal growth factor GDP/GTP exchange and GTP hydrolytic activity of (EGF) receptor (EGFR) (Figure 1) (Egan and Ras is very low and hence, cellular control of GDP/ Weinberg, 1993; Khosravi-Far and Der, 1994). The GTP cycling modulated by two types of regulatory EGF-stimulated EGFR undergoes autophosphoryla- proteins. Guanine nucleotide exchange factors (GEFs; tion of speci®c tyrosine residues in its cytoplasmic RasGRF/mCDC25, SOS1/2) promote formation of the domain which creates phosphotyrosyl binding sites for active GTP-bound state and Ras GTPase activating the Src homology 2 (SH2) and/or phosphotyrosyl proteins (GAPs; p120 GAP, NF1-GAP/neuro®bromin) binding (PTB) domains of the Shc and/or Grb2 promote formation of the inactive GDP-bound state adaptor proteins (Williams, 1992; Mayer and Balti- (Boguski and McCormick, 1993; Quilliam et al., 1995). more, 1993; Schlessinger, 1993b). Shc becomes The single amino acid substitutions at 12, 13 or 61 that autophosphorylated on association with activated

unmask Ras transforming potential create mutant RTKs which creates recognition sites for the SH2 proteins that are insensitive to GAP stimulation (Bos, domain of Grb2. Since Grb2 is stably associated with 1989; Clark and Der, 1993). Consequently, these SOS, the Shc/Grb2- or Grb2-mediated translocation of oncogenic Ras mutant proteins are locked in the the Ras GEF SOS to the plasma membrane, where Ras active, GTP-bound state, leading to constitutive, resides, leads to the transient elevation of Ras-GTP deregulated activation of Ras function. levels (Feig, 1994; Schlessinger, 1993a). Shc, Grb2 and SOS provide the link between many types of activated cell surface receptors and Ras (Williams, 1992; Schlessinger, 1993b). Activated Ras relays its signal downstream through a cascade of cytoplasmic proteins (Figure 1). Sub- stantial biological, biochemical and genetic evidence has implicated the Raf-1 serine/threonine kinase as a critical e€ector of Ras function (Moodie and Wolfman, 1994). A key observation was that biologically active, but not inactive, Ras forms a high anity complex with Raf-1 (Van Aelst et al., 1993; Moodie et al., 1993; Zhang et al., 1993; Warne et al., 1993; Vojtek et al., 1993). The Ras:Raf association promotes a transloca- tion of the normally cytoplasmic Raf protein to the plasma membrane, where subsequent events lead to the activation of its kinase function. These events are complex and remain to be fully understood (Morrison and Cutler, 1997), as will be discussed in more detail below. Upon activation, Raf then phosphorylates and activates two MAPK kinases (MAPKKs; also called MEK1 and MEK2). MEKs directly associate with the COOH-terminal catalytic domain of Raf-1 and are phosphorylated by Raf (Crews and Erikson, 1993). Activated MEKs function as dual speci®city kinases, and phosphorylate tandem threonine and tyrosine residues (TEY motif) in two MAPKs (also referred to as extracellular signal regulated kinases; ERKs), designated p42MAPK/Erk2 and p44MAPK/Erk1 to activate them (Crews et al., 1992). Once activated MAPKs Figure 1 Ras regulation of a cascade of kinases. Ras serves as a GDP/GTP-related binary switch that resides at the inner surface translocate to the nucleus where they phosphorylate of the plasma membrane and acts as a relay for extracellular and activate a variety of substrates that include the ligand-stimulated signals to cytoplasmic signaling cascades. A Elk-1 nuclear transcription factor (Marais et al., 1993). linear pathway where Ras functions downstream of receptor Elk-1 forms a complex with serum response factor tyrosine kinases (RTKs, such as the EGFR receptor) and (SRF) at the serum response DNA element (SRE) upstream of a cascade of serine/threonine kinases (Raf4MEK4MAPK) provides a complete link between the cell present in many promoters such as the c-fos promoter surface and the nucleus (Treisman, 1996b). MAPKs also activate other kinases, Increasing complexity of Ras signaling SL Campbell et al 1397 such as the p90 RSK serine/threonine kinase, to mediating activation of Raf. However, the interaction regulate protein synthesis (Blenis, 1993). of Ras and Raf-1 in vitro is insucient to stimulate Much of the information that has allowed the Raf-1 kinase activity, (Traverse et al., 1993; Zhang et delineation of this Ras signaling pathway has come al., 1993) indicating that in addition to Ras, other from genetic studies of the fruit ¯y Drosophila factors are required for Raf-1 activation in vivo. Initial melanogaster, the nematode Caenorhabditis elegans, reports suggested that Ras bound to Raf-1 at a single the baker's yeast Saccharomyces cerevisiae and the site (Pumiglia et al., 1995) and served simply to ®ssion yeast Schizosaccharomyces pombe. Drosophila translocate Raf-1 to the plasma membrane (Stokoe et eye development is controlled by a signaling pathway al., 1994; Leevers et al., 1994). However, the ability of that begins with ligand stimulation of the Sevenless a membrane-free complex of B-Raf and 14-3-3 proteins RTK, leading to the activation of a Ras homolog, to be activated in vitro by Ras suggests that Ras may which in turn activates a MAPK cascade (Rubin, play a direct role in Raf activation (Kuroda et al., 1991). A near-identical scheme is seen for the pathway 1996). Furthermore, the identi®cation of Ras mutants that regulates proper development of the vulva in C. that bind to Raf-1 with high anity but are defective elegans, which involves a RTK (Let-23)-mediated in Raf-1 activation, suggested that a direct correlation pathway that activates the Ras homolog, Let-60 could not be made between the anity of Raf-1 to (Sternberg and Horvitz, 1991). S. pombe Ras1 also Ras-GTP and Raf-1 activation, as would be expected if activates a MAPK cascade (Nishida and Gotoh, 1993). the sole role of Ras in Raf activation is its recruitment In fact, much of the early clues that identi®ed the to the membrane (Akasaka et al., 1996). Hence these positive (CDC25) and negative (IRA) regulators of Ras observations, coupled with recent ®ndings that Ras

GDP/GTP cycling came from RAS studies in S. interacts with two distinct NH2-terminal regions of cerevisiae (reviewed in Tamanoi, 1988; Broach, 1991). Raf-1, suggests that Ras promotes more than just The remarkable evolutionary conservation of this Ras membrane translocation of Raf, and instead, may also signaling cascade re¯ects the central role of Ras in facilitate subsequent events that lead to Raf-1 diverse organisms. activation (Figure 2) (Brtva et al., 1995; Drugan et al., 1996; Hu et al., 1995). The prevailing data indicate Raf is a key e€ector of Ras Considerable evidence that the ®rst Ras binding site (RBS) in Raf-1 (residues supports the critical role of Raf-1 and the MAPK 55 ± 131) is required for Raf-1 translocation by Ras, cascade in Ras function. For example, dominant whereas interactions between processed Ras and the negative mutants of Raf-1, MEK, and MAPKs have second Ras binding site in the cysteine-rich domain been shown to impair Ras transforming activity (Kolch (CRD) of Raf-1 are required for full activation of Raf- et al., 1991; Schaap et al., 1993; Cowley et al., 1994; 1 activity by Ras (Roy et al., 1997; Clark et al., 1995; Westwick et al., 1994; Qiu et al., 1995a; Khosravi-Far Mineo et al., 1997). Other components that may et al., 1995). Similarly, loss of function mutations in contribute to Raf-1 activation include the Hsp90 and Raf, MEK or MAPK homologs disrupt these Ras- p50 molecular chaperones, phospholipids, serine/ mediated developmental pathways in ¯ies and worms threonine and tyrosine kinases, and the kinase (reviewed in Satoh and Kaziro, 1992). Conversely, suppressor of Ras (KSR) (reviewed in Morrison and constitutively activated mutants of Raf-1 cause a Cutler, 1997). transformed and tumorigenic phenotype which is A further level of regulation involves the interaction indistinguishable from that caused by oncogenic of Raf-1 with 14-3-3 proteins. This family of mutants of Ras, when analysed in rodent ®broblast ubiquitously expressed, highly conserved proteins can transformation assays (Bonner et al., 1985; Stanton, et exist as dimers and are known to complex with a al., 1989; Leevers et al., 1994; Stokoe et al., 1994). variety of proteins (BCR, cdc25, A20, PKC) suggesting Constitutively activated mutants of MEK also cause that they may act as an adaptor proteins (Braselmann morphologic and growth transformation of NIH3T3 and McCormick, 1995; Conklin et al., 1995; Vincenz cells (Alessi et al., 1994; Mansour et al., 1994). Gain of and Dixit, 1996). The role of 14-3-3 in Raf-1 regulation function mutations in homologs of Raf, MEK or remains controversial (Michaud et al., 1995; Suen et MAPK can overcome the loss of Ras function in the al., 1995; Li et al., 1995; Irie et al., 1994; Freed et al., fruit ¯y and nematode (Dickson et al., 1992; Sprenger 1994; Fantl et al., 1994; Chang and Rubin, 1997; Luo et al., 1993). Constitutively activated Raf-1 can et al., 1996; Braselmann and McCormick, 1995; overcome the loss of Ras function caused by dominant Shimizu et al., 1994). This may be due, at least in negative mutants of Ras (e.g., Ras(17N)) or by the part, to the complex nature of the Raf-1/14-3-3 Y13-259 anti-Ras neutralizing monoclonal antibody interaction which involves multiple binding sites, (Smith et al., 1986; Feig and Cooper, 1988). Finally, including the Raf-1 CRD and two sites containing a when it was demonstrated that targeting Raf-1 to the phospho-serine consensus sequence for 14-3-3 at plasma membrane alone was sucient to fully unmask positions 259 and 621 of Raf-1 (Clark et al., 1997b; its transforming activity (Leevers et al., 1994; Stokoe et Li et al., 1993). Moreover, the heterogeneity of the 14- al., 1994), it was suggested that the biochemical 3-3 family and the ability of these proteins to form function of Ras was simply to promote the membrane dimers and heterodimers complicates the interpretation association and activation of Raf-1. When taken of their role in Raf-1 regulation. We and others together, these observations supported the likelihood (Rommel et al., 1996, 1997; Clark et al., 1997b) have that the Raf4MEK4MAPK cascade was both recently proposed a dual role for 14-3-3 in Raf-1 necessary and sucient for Ras function. regulation, with 14-3-3 proteins either stimulating or inhibiting as appropriate by binding to di€erent sites Ras-mediated activation of Raf-1 As detailed above, a on Raf-1. Recent data using Raf-1 mutants which large body of evidence supports the role of Ras in selectively disrupt 14-3-3 binding sites on Raf-1 appear Increasing complexity of Ras signaling SL Campbell et al 1398

Figure 2 Ras-mediated activation of Raf is a complex multi-step process. Although Ras-GTP activation of Raf is mediated by causing the translocation of Raf from the cytosol to the inner surface of the plasma membrane, other events are required to activate Raf kinase function. Ras, via interaction with a second distinct binding site in the cysteine-rich domain of Raf (CRD), may also promote other events that allow the subsequent activation steps to occur. 14-3-3 interacts with multiple regions of Raf and may serve as a negative regulator. Other possible components that may be involved in Raf activation include KSR, phosphatidylserine, p50 and Hsp90, and both tyrosine and serine/threonine kinases. The requirement for multiple signals to activate Raf-1 underscores an emerging theme that e€ectors may promote complex formation at the level of Ras, rather than simply forming a linear signaling pathway

to support this latter hypothesis (Rommel et al., 1997; appears to be a multi-step activation process requiring Clark et al., 1997b). multi-complex formation to complete Raf activation. Yet another level of complexity was added to the As will be discussed, this theme of Ras-mediated multi- regulation of Raf-1 when it was demonstrated that the step e€ector activation and e€ector-mediated multi- cysteine-rich domain (CRD) of Raf-1 binds to the complex formation may be extended also to other Ras anionic membrane phospholipid, phosphatidylserine e€ectors. In other words, the complexities we are just (PS) (Ghosh et al., 1994). Moreover, it has now been now delineating in the recognition of Ras by Raf and demonstrated that the addition of PS to a cellular its role in Raf-1 activation, may serve as a prototype preparation of K-Ras and B-Raf enhances the for the interaction and activation of other e€ectors by activation of B-Raf (Kuroda et al., 1996) and that Ras and Ras-related proteins. the addition of a mixture of plasma membrane lipid Other observations also revealed the oversimplifica- comonents, including PS, allows Ras to activate c-Raf- tion of a Raf4MEK4MAPK linear cascade. For 1 in vitro (Stokoe and McCormick, 1997). We have example, the mammalian Raf protein family contains recently generated Raf-1 CRD variants which are three family members: Raf-1, A-Raf and B-Raf, with speci®cally impaired for binding of PS (Clark et al., B-Raf existing in multiple spliced forms (Barnier et al., 1997a), and found them to be defective in Raf-1 1995; Eychene et al., 1995). Recent data indicate that activation. However, the interaction with PS becomes unique regulatory events are likely to be involved in unnecessary for the activation of Raf-1 in CRD the activation of each of the three family members. For mutants which are defective in both PS and 14-3-3 example, it has been shown previously that maximal binding. Furthermore, we have found that the binding activation of Raf-1 is produced by synergistic signals sites for 14-3-3 and PS in the Raf-1 CRD appear to from oncogenic Ras and activated tyrosine kinases overlap, and that PS can compete with 14-3-3 for (Marais et al., 1995). In particular, A-Raf like Raf-1, binding to the CRD of Raf-1. appears to be weakly activated by oncogenic Ras and Thus, a model is presented in Figure 2 where the more strongly activated by oncogenic Src, and these activation of Raf-1 by Ras involves recruitment to the signals synergize to give maximal activation. In plasma membrane via binding to the ®rst Ras binding contrast, B-Raf is strongly activated by oncogenic site (RBS) of a Raf-1 molecule constitutively associated Ras alone and is not activated by oncogenic Src with 14-3-3 (Rommel et al., 1996). Once at the plasma (Marais et al., 1997). Similarly, protein kinase C (PKC) membrane, PS acts to promote the interaction of Ras can phosphorylate and activate Raf-1 directly and the second Ras binding site of Raf-1 (CRD) by (Morrison, 1994; Sozeri et al., 1992; Kolch et al., displacing 14-3-3 to permit full Raf-1 activation by 1993), whereas protein kinase A antagonizes EGF- other factors such as KSR, HSP90, and both serine/ activated MAPK signaling pathway (Wu et al., 1993). threonine and tyrosine kinases (as reviewed in The supporting data, although controversial, (Morri- Morrison and Cutler, 1997). Therefore, the connection son et al., 1993) suggests that MAPK inhibition by between Ras and Raf-1 alone is not simply linear and PKA results from phosphorylation and inhibition of Increasing complexity of Ras signaling SL Campbell et al 1399 Raf-1 activity (Chuang et al., 1994; Weissinger et al., tion and sporulation. Like Raf, Byr2 has a distinctive 1997). N-terminal kinase regulatory domain and a character- MAPKs can be activated by Ras-independent istic C-terminal kinase catalytic domain, where multi- mechanisms. For example, the activation of MEK ple protein interactions within the regulatory domain and ERK by an activated form of PKCd requires the appear to be required for control of Byr2 kinase presence of c-Raf and is independent of a dominant activity (Tu et al., 1997). Scd1 represents a second negative form of Ras (RasN17) (Ueda et al., 1996). In e€ector and functions as an exchange factor for the S. addition, evidence for P13K and cPKCs mediated pombe CDC42 homolog, Cdc42 (Chang et al., 1994). MEK-independent activation of MAPK exists (Gram- Scd1 regulates a pathway that includes CDC42 and mer and Blenis, 1997). Moreover, integrin-mediated Shk1 (PAK homolog) and controls cell morphology activation of MAPKs can occur by a process (Marcus et al., 1995). Hence, these data provide a link independent of Ras (Chen et al., 1996a). However, between Ras and Rho family proteins and de®ne two MEK transforming activity may still be dependent on distinct Ras1 e€ector mediated activities. Whether a Ras function (Cowley et al., 1994). Furthermore, speci®c mammalian Rho GEF(s), analogous to Scd1, MAPKs can phosphorylate MEK, although the can function as a Ras e€ector remains to be functional consequences are not clear. Raf and MEK determined. Finally, although the best characterized can cause activation of the p70 S6 kinase, but e€ector of S. cerevisiae RAS2 is not a Raf-related apparently independent of p42/p44 MAPK activation molecule, and is instead adenylate cyclase, a second (Lenormand et al., 1996). Finally, MEK1 has been Ras-dependent pathway was suggested by the fact that shown to mediate activation of Raf-1 by a novel loss of RAS, but not adenylate cyclase, is lethal (Toda positive feedback mechanism that is independent of et al., 1985). RAS2 also signals via a CDC42/STE20 Ras, Src and tyrosine phosphorylation of Raf-1 MAPK cascade to induce ®lamentous growth (MoÈ sch (Zimmermann et al., 1997). In particular, MEK has et al., 1996). been shown to phosphorylate both the amino- and Second, in addition to p42 and p44 MAPKs/ carboxyl terminus of Raf-1. Only phosphorylation of ERKs, activated Ras induces activation of the JNK/ the carboxyl terminal catalytic domain of Raf-1 SAPK and p38/HOG MAPK cascades, which in turn appears to correlate with increased Raf-1 kinase activate the Elk-1, c-Jun and ATF-2 nuclear activity. Taken together, each signaling protein in this transcription factors (De rijard et al., 1994). While cascade is likely to have additional targets and each these kinase cascades are analogous to the component may be activated by other upstream Raf4MEK4MAPK cascade (reviewed in Treisman, components. Furthermore, feedback loops and auto- 1996a; Kyriakis and Avruch, 1996) they de®ne crine mechanisms may also exist. For example, Raf parallel pathways that are stimulated independent of activation of a second MAPK pathway, involving the Raf activation (Minden et al., 1994; Olson et al.,

Jun NH2-terminal kinase (JNK; also known as SAPK) 1995). However, whereas ERK activation is asso- was found to be mediated by an autocrine loop ciated with growth stimulatory responses, JNK and involving the EGF receptor (Minden et al., 1994). p38 activation are typically associated with stress Hence, this signaling cascade is not a simple linear responses that result in apoptosis (Xia et al., 1995; pathway. Various stimuli and cell types are likely to Verheij et al., 1996; Chen et al., 1996b). For example, di€erentially utilize multiple, possibly temporally MEKK1 activates both JNK and p38 and causes distinct pathways converging on MAPK. Moreover, apoptosis (Johnson et al., 1996). Similarly, activation recent observations began to establish that Raf is not of JNKs is associated with apoptosis in PC12 the sole downstream target of Ras and that Ras pheochromocytoma, U937 leukemia and other cell function may require its association with a surprisingly types (Xia et al., 1995; Verheij et al., 1996; Chen et large number of other e€ectors, leading to transduction al., 1996b). However, we recently observed that of Ras-mediated signals through multiple pathways inhibition of JNK activation impaired Ras transfor- (Boguski and McCormick, 1993) mation suggesting a growth promoting role for this MAPK cascade (Clark et al., 1997d). Jun function is also required for Ras transformation (Granger- Ras activation of Raf-independent pathways contribute Schnarr et al., 1992; Clark et al., 1997d). Therefore, to Ras transformation JNK activation may promote di€erent cellular Despite the apparent linear nature of the consequences that depend on the coordinate activa- Ras4Raf4MEK4MAPK signaling cascade, there tion of other pathways. Thus, JNK activation by was also mounting evidence that it represented but a stress stimuli alone may cause apoptosis, whereas mere subset of a very complex array of signaling JNK activation under other conditions may synergis- interactions at several levels. For example, a consider- tically enhance the growth promoting action of able body of evidence supports the hypothesis that Raf activated MAPKs (Xia et al., 1995). Alternatively, is not the sole e€ector of Ras function and that Raf- JNK activation may possess cell type speci®c actions independent pathways are critical for mediating Ras (De rijard et al., 1994; Xia et al., 1995; Verheij et al., function. First, genetics studies in S. pombe revealed 1996; Chen et al., 1996b; Su et al., 1994), and serve a the involvement of at least two distinct downstream protective function, at least in ®broblasts, and e€ector-mediated signaling pathways that facilitate full thereby contribute to cell transformation. Ras1 function (Marcus et al., 1995). One e€ector of Third, transformation studies in several epithelial Ras1 is Byr2, a MEKK/Raf homolog (Van Aelst et al., cell lines including RIE-1 rat intestinal epithelial cells, 1993; Wang et al., 1991). Byr2 is a component of a showed that Ras activation of the protein kinase cascade that includes Byr1 (MEK) and Raf4MEK4MAPK pathway alone was not suffi- Spk1 (MAPK) that regulates agglutination, conjuga- cient to cause complete transformation (Oldham et al., Increasing complexity of Ras signaling SL Campbell et al 1400 1996). Whereas constitutively activated mutants of Ras two mutants have not been clearly established. and Raf-1 show equivalent transforming potencies However, since Ras(12V, 40C) caused a transformed when assayed in rodent ®broblast transformation morphology similar to Rho proteins, retained the analyses, activated Raf failed to cause morphologic ability to activate JNK (Khosravi-Far et al., 1996), and growth transformation of RIE-1 cells. Instead, Ras and induced membrane ru‚ing in REF52 cells which activation of Raf-independent pathways, which led to could be blocked by dominant negative Rac1 (Joneson the induction of an EGFR-dependent autocrine growth et al., 1996), an e€ector pathway leading to Rac1 may pathway, was found to be critical for oncogenic Ras be important for the transforming activity of this transformation of these cells (Gangarosa et al., 1997). mutant protein. Other evidence for the inability of Raf to promote full Other studies utilizing these Ras e€ectors mutants Ras function was provided by a study that showed that have been conducted recently to investigate how Ras Ras, but not Raf, could trigger the morphologic binding to Raf-1 mediates Ras transforming activity, as changes and actin reorganization associated with well as the roles of other candidate Ras e€ectors in cellular hypertrophy (Thorburn et al., 1993). transformation (Marte et al., 1997; Rodriguez-Viciana Fourth, it has been shown that constitutively et al., 1997; Winkler et al., 1997; Stang et al., 1997). activated mutants of two members of the Ras branch While these Ras e€ector mutants aid in identifying and of Ras-related proteins (TC21/R-Ras2 and R-Ras; characterizing Ras-mediated signal pathways, they 55% identity with Ras) can cause tumorigenic remain incompletely characterized with regard to their transformation of rodent ®broblasts (Figure 3) ability to bind and activate various Ras e€ector relative (Graham et al., 1994; Chan et al., 1994; Saez et al., to WT Ras or other activated forms of Ras. Moreover, 1994; Cox et al., 1994). Although both TC21 and R- given di€erences in binding, cell type and cell-based Ras share complete sequence identity with the core Ras assays employed in these studies, caution should be e€ector domain (residues 32 ± 40), as shown in Figure exercised when interpreting the resulting data. 4, and can interact with an isolated Ras-binding

domain present in the NH2-terminus of Raf-1, these proteins failed to interact with and activate Raf kinases in vivo or to directly activate ERKs (Graham et al., 1996; Hu€ et al., 1997). In particular, we have found that activated TC21 exhibits the same di€erentiation- inducing (PC12 cells) or di€erentiation-inhibiting (C2 myoblasts) phenotype that has been observed with activated Ras (Graham and Der, unpublished observa- tions). Therefore, Raf-independent pathways are clearly sucient to mediate the transforming and di€erentiating functions of TC21 and, presumably, Ras. While the Raf-independent pathways that promote transformation by these two Ras-related proteins are presently not known, it is suspected that Ras may activate some of these Raf-independent pathways. This hypothesis is based in part on the fact that TC21 and R-Ras can interact with many of the Ras-GTP binding partners that have been identi®ed. Finally, important and direct evidence for the involvement of multiple signaling pathways down- stream of Ras comes from the identi®cation of Ras e€ector domain mutants that are defective in interac- tion and activation of Raf (White et al., 1995). These Raf-binding defective mutants, designated Ras(12V, 37G) and Ras(12V, 40C), no longer activate Raf-1 or MAPKs, but retain the ability to bind to other putative Ras targets (Khosravi-Far et al., 1996). Surprisingly, although these e€ector mutants did not cause signi®cant morphologic transformation when stably expressed in NIH3T3 cells, they promoted a growth- transformed phenotype (growth in low serum, colony formation in soft agar and tumor formation in nude mice) similar to that caused by oncogenic Ras. In contrast to WT Ras, however, these mutants caused a transformed phenotype distinct from that caused by Figure 3 Ras subfamily dendrogram. Primary sequences corresponding to Ras subfamily proteins were aligned using Raf-1, which was similar to that caused by constitu- ClustalW (Thompson et al., 1994), a dynamic sequence alignment tively activated mutants of Rho family proteins. program. From the resulting multiple sequence alignment, a Furthermore, co-expression of these Ras e€ector distance matrix was prepared and used to construct the Ras domain mutants with activated Raf-1 resulted in family dendrogram. The branch lengths in the dendrogram are proportional to the estimated divergence along each branch. With potent synergistic transforming activity. The non-Raf the exception of the mouse homolog, M-Ras, only human e€ectors that mediate the transforming actions of these sequences are presented Increasing complexity of Ras signaling SL Campbell et al 1401

Figure 4 Ras e€ector-interaction sequences. Multiple, distinct sequences of Ras proteins are involved in interactions with Raf and other candidate Ras e€ector targets. The core Ras sequence domain (residues 32 ± 40), containing part of the conformationally sensitive switch I region, may represent a common interaction site for all Ras-GTP binding proteins, and shows complete identity with the equivalent sequences of TC21/R-Ras2, R-Ras and Rap-1A,1B proteins. Sequences that ¯ank this core sequence have been shown to be involved in additional binding contacts that promote speci®city and/or activation of downstream e€ectors. Sequence di€erences are seen in the closely related, but biologically-distinct, Rap and R-Ras proteins. Other sequences that may in¯uence e€ector interactions have been identi®ed in the switch II domain, residues spanning 92 ± 106 and the lipid-modi®ed carboxy terminus

domain, a region of Ras (residues 30 ± 37) whose Ras mediates its actions through interaction with conformation di€ers between GTP- and GDP-bound multiple e€ectors forms (Tong et al., 1989; Milburn et al., 1990; Krengel The existence of Raf-independent Ras signaling path- et al., 1990). In addition to the three Raf's, several ways is further insinuated by the expanding roster of proteins have been identi®ed with these characteristics candidate Ras e€ectors that has emerged over the last (Figure 5), but a central goal in the ®eld, is to decide several years (Figure 5) (Van Aelst et al., 1994; which of these are actually downstream targets of Ras. Marshall, 1996). Like Raf-1, these structurally and Critical aims of current studies are to verify whether functionally diverse proteins show preferential binding these candidate Ras e€ectors actually serve as to the active GTP-bound form of Ras and this physiologically relevant e€ectors of Ras function and interaction requires an intact core Ras e€ector to de®ne their precise contribution to Ras signaling domain. These candidate e€ector molecules have been and transformation. A demonstration of binding, in identi®ed by a variety of experimental approaches either yeast two-hybrid analyses or with puri®ed including biochemical methods, yeast two hybrid recombinant proteins in vitro, support the candidacy library screening and yeast functional screening. It is of a particular protein as a Ras e€ector. However, highly likely that additional e€ectors will be identi®ed. while a demonstration of interaction in cells is a key In this section, we summarize our current under- requirement, assessment of binding interactions with standing of the involvement of these candidate Ras full length proteins is also essential. The use of isolated e€ectors in mediating Ras function. We also speculate fragments of Raf, that contain RBS1, to assess on the structural requirements for Ras interaction with interaction with Ras, Ras e€ector mutants or Ras- such a diverse spectrum of functionally divergent related proteins has led to misleading conclusions. For targets. example, it has now been shown that some Ras e€ector domain mutations which do not impair binding to the A growing roster of candidate e€ectors of Ras Putative isolated Raf-RBS, abrogate interaction with the full e€ectors of Ras have been characterized as binding length Raf-1 protein (Winkler et al., 1997). Moreover, preferentially to the GTP-bound form of Ras, with the Ras-related protein, TC21, can interact with the binding determinates requiring what has been geneti- isolated Raf-RBS in yeast two-hybrid binding assays, cally de®ned as the core Ras e€ector domain (residues yet failed to bind to and activate full length Raf-1 in 32 ± 40). This region comprises part of the switch I vivo (Graham et al., 1996). Increasing complexity of Ras signaling SL Campbell et al 1402

Figure 5 Ras interacts with multiple e€ectors. In addition to the three Raf serine/threonine kinases (Raf-1, A-Raf and B-Raf), a number of other proteins share properties of a Ras e€ector, and bind preferentially to active, Ras-GTP. The interaction is dependent on an intact core e€ector domain. Critical aims of current studies are to verify whether these candidate Ras e€ectors actually serve as physiologically relevant e€ectors of Ras function and to de®ne their precise contribution to Ras signaling and transformation

Some additional criteria for validating the authenti- de®cient mutants of oncogenic Ras still retained city of a protein as an e€ector of Ras are based on transforming potential (Khosravi-Far et al., 1996), it evidence for Raf-1 e€ector function. These include the has become apparent that multiple e€ectors are detection of stable association of Raf-1 with activated required for full Ras function. Therefore, the loss of Ras in vivo, an increased plasma membrane association binding of any one e€ector alone may not cause a of Raf-1 upon Ras activation, and stimulation of Raf-1 signi®cant alteration to Ras transforming activity. kinase activity upon activation of Ras. The use of Thus, these e€ector domain mutagenesis results alone e€ector domain mutants that have selectively impaired do not rule out an e€ector role for GAPs. Instead, a the ability to activate Raf-1, have aided in assessing the signi®cant body of evidence has established intriguing, contribution of Raf-1 to Ras transforming activity. but incomplete, support for p120 GAP as a Ras Does loss of interaction correlate with loss of e€ector (Tocque et al., 1997). transforming activity or loss of a particular Ras Evidence for p120 GAP e€ector function was function or signaling activity? Finally, high anity derived from studies that evaluated the activity of

binding would be expected of a physiologcially NH2-terminal fragments of p120 GAP that lack the signi®cant e€ector. However, it should be cautioned catalytic GTPase stimulatory domain and Ras-binding

that these critiera are not absolute and that bona ®de sequences. The NH2-terminus of p120 GAP contains

e€ectors may not always share these characteristics an SH3 domain ¯anked by two SH2 domains, a with Raf. pleckstrin homology (PH) domain, a calcium binding The best studied Ras GAPs are p120 GAP and domain, and a site of interaction with phospholipids. neuro®bromin (NF1) (as reviewed in McCormick, In one study, McCormick and colleagues found that a 1995; McCormick et al., 1991). Although p120 GAP Ras : p120 GAP complex was required for inhibition of and NF1/neuro®bromin clearly function as negative muscarinic receptor-activated potassium channel open-

regulators of Ras function (Boguski and McCormick, ing, whereas an NH2-terminal fragment of p120 GAP 1993), p120 GAP represented the ®rst candidate Ras could cause this independent of Ras (Yatani et al., e€ector (Adari et al., 1988; Cale s et al., 1988). 1990; Martin et al., 1992). In a separate study, it was

Apparent evidence against an e€ector function for observed that an antibody directed against the SH3 Ras GAPs came from studies of Ras e€ector domain domain of p120 GAP blocked Ras-induced germinal mutants that impaired p120 or NF1 interactions, yet vesicle breakdown, but not MAPK activation, when retained strong focus-forming activity (Marshall and assayed in Xenopus oocytes (Pomerance et al., 1996;

Hettich, 1993). These observations, together with Hartwell, 1992). Similarly, co-expression of a NH2- evidence establishing Raf as a true Ras e€ector, terminal fragment was shown to block oncogenic H- signaled the demise of the notion that GAPs served Ras, but not Raf-1, transforming activity (Clark et al., as e€ectors of Ras. However, since Raf-binding 1993). Moreover, this fragment blocked Ras activation Increasing complexity of Ras signaling SL Campbell et al 1403 of JNK, but not MAPKs (Clark et al., 1997d). Pawson In summary, the jury is still out with regard to an and colleagues found that overexpression of a p120 e€ector function for Ras GAPs. In the case of both

GAP NH2-terminal fragment altered cell morphology p120 Ras GAP and NF1, a considerable body of and organization of the actin cytoskeleton (McGlade et indirect evidence suggests that these molecules may al., 1993). Furthermore, observations that the p120 play roles in controlling non-Raf/MAPK pathways GAP-associated protein, p190, which is primarily a activated by Ras which may be essential for such GAP for RhoA, but is also active on CDC42 and Rac neoplastic properties as invasiveness (Kim et al., 1997). (Settleman et al., 1992), provides evidence that p120 p120 GAP- or NF1-de®cient mice have been described GAP may serve as an e€ector that facilitates Ras and both lead to death during embryogenesis regulation of Rho protein function. These ®ndings, (Henkenmeyer et al., 1995; Bollag et al., 1996). The together with others (reviewed in Tocque et al., 1997), analysis of whether Ras transforming potential is suggest that p120 GAP may serve dual roles as both a impaired in ®broblasts derived from the embryos of negative regulator of Ras and as a sca€olding protein these animals will provide further assessment of that promotes formation of a Ras-GTP dependent whether Ras GAPs are important e€ectors of Ras signaling complex with Ras. A current working model transformation and may identify the signaling path- proposes that Ras-GTP binding to the COOH-terminal ways that they regulate. catalytic domain would expose the protein-ligand The Ral guanine nucleotide dissociation stimulator binding motifs present in the NH2-terminus, and the (RalGDS) and two closely related proteins (RGL and subsequent association of components with the NH2- RGL2/Rlf) represent intriguing candidate e€ectors of terminus would activate a signaling function. The Ras that may link Ras with other Ras-related proteins. identi®cation of other p120 GAP-binding proteins in Members of this family have been identi®ed repeatedly addition to p190 Rho GAP, (e.g., p62dok, G3BP) by yeast two-hybrid library screening searches for provide support for such a scenario (Settleman et al., candidate e€ectors of Ras and Ras-related proteins 1992; Yamanashi and Baltimore, 1997; Ellis et al., (Rap, R-Ras and TC21/R-Ras2) (Kikuchi et al., 1994; 1990; Parker et al., 1996). Hofer et al., 1994; Spaargaren and Bischo€, 1994; Evidence for NF1 as a Ras e€ector is less Wolthuis et al., 1996; Peterson et al., 1996; Lo pez- compelling and is based largely on a detection of Barahona et al., 1996). RalGDS was identi®ed NF1 function beyond its role as a Ras GTPase originally as a CDC25-related protein (yeast RAS stimulator (McCormick, 1995). Like p120 GAP, a GEF) but was found instead to function as a GEF for signi®cant portion of the 220 kDa NF1 protein is the two Ras-related proteins RalA and RalB (85% unrelated to its GTPase catalytic function, and identical to each other; Figures 3 and 4) (Albright et consequently, may be involved in an e€ector al., 1993). The COOH-terminal noncatalytic domains function. Clues to such a function are limited, but of RalGDS, RGL and RGL2/Rlf interact with the they suggest an e€ector function involved in growth e€ector domain region of Ras in a GTP-dependent suppression rather than promotion. For example, in manner in vitro and can compete with Raf-1 for some studies, no elevation in Ras-GTP was seen in a binding to the Ras e€ector domain region in vitro. Ras number of NF1-de®cient tumors (Andersen et al., association with RalGDS has also been observed when 1993; The et al., 1993; Johnson et al., 1993), yet overexpressed in COS cells (Kikuchi and Williams, introduction of full length NF1 into these cells 1996). Furthermore, the GAP activity of both p120 resulted in severe reductions in growth (Johnson et GAP and NF1-GAP on Ras were competitively al., 1994). These observations suggest that a loss of inhibited by RalGDS (Kikuchi et al., 1994), suggesting NF1 function, other than as a Ras GTPase that RalGDS, like Raf-1, interacts with Ras in a GTP- stimulator, contributes to the development of these dependent manner through its e€ector domain. tumors. In further support of this hypothesis, two Although RalGDS and related proteins also bind to separate studies showed that NF1 overexpression Ras-related proteins, transient expression assays in inhibited the transformed growth properties of cells COS cells showed that Ras, but not Rap1A or R-Ras, that expressed constitutively activated Ras mutants. enhanced the Ral GEF activity of RalGDS (Urano et This would be unexpected if NF1 served simply as a al., 1996). This observation underscores an important GAP, since the GTPase activity of mutant Ras is not caution with regard to the interpretation of data stimulated by NF1. First, Lowy and colleagues showing the ability of a particular protein to bind in showed that overexpression of neuro®bromin drasti- vitro. Again, whether full length proteins or fragments cally reduced (®vefold) viral H-Ras, but not Raf, are employed in binding analyses and whether binding focus-forming activity (Johnson et al., 1994). Second, occurs in vivo needs to be assessed. Moreover, a link Li and White found that the introduction of full between association and stimulation of e€ector length NF1 into HCT116 colon carcinoma cells, which function needs to be established to validate a true harbor mutated K-Ras, suppressed soft agar growth in interaction. vitro and tumor formation in nude mice (Li and Evidence that RalGDS may serve as a positive White, 1996). NF1 overexpression was found to regulator of Ras transformation comes from several reduce MAPK activation, and introduction of independent observations. First, co-expression of activated Raf could overcome the NF1 suppression isolated Ras-binding domains of RGL and RGL2/Rlf (Li and White, 1996). Thus, these investigators inhibited Ras, but not Raf, transforming activity in suggested that only modest elevations (2 ± 3-fold) of NIH3T3 cells (Okazaki et al., 1996; Peterson et al., NF1 expression can block Ras activation of Raf. 1996). Second, although constitutively activated Ral Third, recent experiments in Drosophila systems have alone does not cause transformed foci, one study linked neuro®bromin to PKA signaling pathways (The reported that its co-expression enhanced Ras transfor- et al., 1997; Izawa et al., 1996). mation, whereas dominant negative Ral impaired Ras Increasing complexity of Ras signaling SL Campbell et al 1404 focus-forming activity (Urano et al., 1996). However, Although the e€ect of Ras on PI3K activity in vitro these e€ects were not dramatic and a second study did is modest (*twofold) using bacterially expressed not ®nd signi®cant regulation of Ras transforming recombinant Ras, it is possible that like Raf-1, PI3K activity by mutant RalA or RalB proteins (White et al., activation requires posttranslational modi®cation of 1996). We also found no signi®cant regulation of Ras Ras and/or other factors to achieve full stimulation of focus-forming potential by mutants of RalB (Jordan PI3K activity (Kodaki et al., 1994). In fact, is has been and Der, unpublished observation). Third, co-expres- recently shown that Ras-GTP can activate the lipid sion of RalGDS cooperated synergistically with kinase activity of PI3K directly in vitro and that this activated Raf-1 to induce transformation of NIH3T3 e€ect is synergistic with tyrosine phosphopeptide cells (White et al., 1996). While RalGDS alone showed binding to p85 (Rodriguez-Viciana et al., 1996). The no focus-forming activity, we also found that co- site of interaction of Ras with PI3K resides between expression of RalGDS caused a threefold enhancement residues 133 ± 314 on the catalytic p110 subunit, with of R-Ras, but not Ras or TC21, focus-forming activity lysine residue 227 being essential for the interaction (Brtva and Der, unpublished observation). It was not (Rodriguez-Viciana et al., 1996). This is a region established whether the growth-promoting activity seen immediately carboxy terminal to the site of interaction with RalGDS was due to its activation of Ral proteins. between p110 and the p85 regulatory subunit, It is conceivable that RalGDS activation of Ral suggesting a model where PI3K is regulated through proteins may in turn lead to either activation of at least two domains in its amino-terminal regulatory CDC42/Rac or phospholipase D (Jiang et al., 1995). domain, one from tyrosine phosphoproteins through its

The identi®cation of a putative RalA e€ector, RalBP1 SH2 binding site and Ras-GTP. The closely related that is a novel GAP for CDC42 and Rac, suggests that members for the Ras family Rap1A/B, R-Ras and RalBP1 may regulate these Ras-related GTPases TC21 all interact with PI3K, although Rap2 and Ral (Cantor et al., 1995). However, it is also possible that did not. However, with the exception of R-Ras it is not RalGDS may have functions distinct from its Ral GEF yet clear whether binding to these Ras-related GTPases activity. to PI3K leads to its activation in vivo (Marte et al., Phosphoinositol 3-kinase (PI3K) is a lipid kinase 1997). that phosphorylates phosphoinositides at the 3' A number of signaling molecules have been position of the inositol ring (Carpenter and Cantley, implicated downstream of PI3K, including Rac, 1996a) and has also been implicated as a Ras e€ector p70S6 kinase, PKB/AKT, and novel and atypical (Figure 5). PI3K is composed of a p110 catalytic and a isoforms of protein kinase C (Carpenter and Cantley, p85 regulator subunit and both are members of a 1996b). One of the downstream targets of activated family of related proteins. Multiple isoforms of PI3K PI3K, protein kinase B (PKB; also Akt), has received a can associate with Ras and Ras can activate both great deal of attention recently (Franke et al., 1995; wortmanin-sensitive and -insensitive isoforms (Rodri- Klippel et al., 1996; Marte et al., 1997), as a series of guez-Viciana et al., 1997). Five isoforms of the studies have identi®ed a role for PKB in cell survival regulatory subunit and ®ve forms of the catalytic (as reviewed in Franke et al., 1997). A recent study subunit have been identi®ed (as reviewed in Carpenter using Ras e€ector domain mutants implicated the and Cantley, 1996b). The catalytic p110 subunits di€er PI3K/PKB in oncogenic Ras suppression of Myc- in their substrate speci®city and regulation. A novel induced apoptosis in Rat-1 ®broblasts, whereas Ras PI3K isoform has recently been identi®ed (p110d), activation of the Raf4MEK4MAPK pathway was which unlike p110a, does not phosphorylate p85 but found to promote the Myc apoptotic response instead harbors an intrinsic autophosphorylation (Kau€mann-Zeh et al., 1997). Since oncogenic Ras capacity. The p110d cataytic domain contains unique also potentiated the apoptotic response, the Raf- potential protein-protein interaction modules such as a mediated pathway may be dominant over the PI3K proline-rich region and a basic-region leucine-zipper pathway. In addition, R-Ras has been shown to bind like domain (Vanhaesebroeck et al., 1997). Addition- to and activate PI3K (Marte et al., 1997), and this ally, another p110 isoform designated p100O was also interaction leads to PI3K-PKB/Akt pathway activa- identi®ed during a yeast two-hybrid library screen for tion, but not the Raf-MAP kinase pathway (Marte et Ras-interacting proteins (Vojtek et al., 1993). al., 1997). These results indicate the existence of a Ras Recombinant p110 (a- and b-) showed high anity signaling cascade parallel to the Raf/MEK/MAPK interaction with the GTP-bound form of recombinant pathway where Ras binds and activates PI3K, causing Ras through the Ras e€ector domain (Rodriguez- activation of PKB. A direct mechanism of PKB Viciana et al., 1994, 1996). In intact cells, there is activation by PI3K that involves the binding of

evidence that Ras can stimulate PI3K activity and is PtdIns-3,4,5,-P3 to the PKB PH domains has been required for its optimal activation in response to described recently (Franke et al., 1997; Klippel et al.,

growth factors (Kodaki et al., 1994; Rodriguez- 1997). Since the PI3K products, PtdIns(3,4)P2 and

Viciana et al., 1994). Furthermore, dominant negative PtdIns(3,4,5)P3, can regulate the activities of a wide Ras inhibited platelet-derived growth factor (PDGF) variety of proteins, including PKC isoforms and activation of PI3K. In addition, introduction of proteins that regulate the actin organization (e.g. activated Ras, but not Raf, into COS cells resulted in pro®lin), the consequences of PI3K activation are a signi®cant induction of inositol phosphates demon- likely to be beyond AKT and will be complex. strating the convergence of Raf and PI3K pathways at On the other hand, several observations implicated the level of Ras. Taken together, these observations PI3K as an upstream activator of Ras function. First, suggest strongly the involvement of PI3K as a phosphorylation of PDGFR at sites that are involved downstream e€ector of Ras function (Rodriguez- in p85 binding is required for the activation of Ras by Viciana et al., 1994). PDGF (Fantl et al., 1992). Second, an activated form Increasing complexity of Ras signaling SL Campbell et al 1405 of p110 caused a small elevation in the level of Yeast studies looking for genes that interfere with activated Ras-GTP in vivo (Hu et al., 1995). One Ras function identi®ed the Ras interaction/interference possible resolution to these apparently contradictory gene 1 (Rin1) that suppressed activated RAS2-induced roles of PI3K is that, depending on the speci®c cyclic AMP activation (Han and Colicelli, 1995). This extracellular stimulus or cell type, PI3K isoforms may interference is at the level of RAS2, since Rin1 cannot act as either an upstream activator or a downstream suppress the activity of components downstream of mediator of Ras function. RAS2 in the adenylyl cyclase pathway. Additionally, in Another candidate e€ector of Ras is the mitogen- vitro binding and yeast two hybrid analysis demon- activated protein kinase kinase kinase 1 (MEKK1), a strated that Rin1 directly interacted with RAS2 as well serine-threonine kinase that is an upstream activator as the mammalian H-Ras. This interaction was GTP- of SEK (a MAPKK), which in turn activates JNK/ dependent and required an intact Ras e€ector domain. SAPK (Figure 5) (Lange-Carter et al., 1993; Yan et Furthermore, Rin1 competed with Raf-1 in binding to al., 1994). MEKK1 can be activated in response to a Ras in vitro (Han and Colicelli, 1995). A 434 amino variety of extracellular stimuli, ultraviolet radiation, as acid region of Rin has been shown to be required for well as activated Ras. Evidence for its role as a Ras ecient Ras binding (Han et al., 1997). Finally, the e€ector comes from observations that MEKK1 bound demonstration that Ras and Rin1 can form a stable directly to GST-Ras(12V) in a GTP-dependent complex in vivo provides strong evidence that Rin1 is a manner via its COOH-terminal kinase domain in physiologically relevant e€ector of Ras (Han et al., vitro (Russell et al., 1995). This binding was blocked 1997). Although overexpression of Rin1 alone in by a Ras e€ector peptide. Whether MEKK1 is a true NIH3T3 cells showed no growth-promoting activity, Ras e€ector remains to be critically tested. Since we have observed that co-expression of Rin1 with overexpression of MEKK1 causes apoptosis (Johnson activated R-Ras showed a synergistic enhancement of et al., 1996; Cardone et al., 1997), it seems unlikely focus-forming activity (Brtva, Colicelli and Der, that it would be an important positive e€ector for Ras unpublished). Thus, Rin1 may be involved in positive transforming activity. MEKK1 has also been shown growth regulation. to bind to GTP-complexed Cdc42 and Rac1 in vitro Recent evidence suggests a possible role of Rin1 in and kinase-dead MEKK1 can block Cdc42/Rac linking Abl tyrosine kinase function with Ras. First, activation of JNK (Fanger et al., 1997). Thus, the Abl tyrosine kinase was found to bind Rin1 in vitro MEKK1 may serve as e€ectors of Cdc42 and Rac and served as a substrate for its kinase activity (Han et as well. Finally, MEKK1 showed a nuclear location al., 1997). However, no association between Abl and and an association with vesicular-like structures in the Rin1 was detected in vivo. Second, co-expression of cytoplasm (Fanger et al., 1997). However, the Rin1 greatly potentiated the transforming activity of relevance of this subcellular location and the the chimeric BCR-Abl fusion protein (Afar et al., functional interaction with plasma membrane asso- 1997). BCR-Abl is implicated in the development of ciated Ras proteins is not clear. Philadelphia Chromosome-positive human leukemias. AF6/Rsb1 is yet another candidate Ras e€ector that This interaction required tyrosine phosphorylation of was identi®ed by yeast two-hybrid library screening for Rin1 to mediate its interaction with the Abl SH2 and

Ras-binding proteins and by anity chromatography SH3 domains. Stable complex formation between Rin1 puri®cation of Ras-GTP-binding proteins (Figure 5) and BCR-Abl was seen. Whether Rin1 can complex (Van Aelst et al., 1994; Kuriyama et al., 1996). AF-6 simultaneously with BCR-Abl and Ras is not known. shows a high degree of sequence similarity to Moreover, like the cysteine-rich domain of Raf-1 Drosophila Canoe, which is assumed to function in (Clark et al., 1997b), a domain of Rin1 that binds signaling pathways downstream from the Notch Ras also binds 14-3-3 proteins (Han et al., 1997). Thus, receptor, acting as a regulator of cellular differentia- it appears that Rin1 can interact with multiple tion (Hunter, 1997). The p180 AF-6 protein was also signaling molecules and may possess functions distinct identi®ed independently as one of up to ten fusion from those of Ras. partners of the MLL protein associated with chromo- The zeta isoform of protein kinase C (PKCB) has some translocation events in human leukemias (Prasad also been implicated as a Ras e€ector. First, PKCB was et al., 1993). The MLL/AF-6 chimeric protein is the required for Ras-induced maturation in Xenopus gene product of a reciprocal translocation oocytes (Dominguez et al., 1992) and serum-stimu- t(6;11)(q27;q23) associated with a subset of human lated mitogenic signaling in mammalian cells (Berra et acute lymphoblastic leukemias (Prasad et al., 1993; al., 1993). Second, the regulatory NH2-terminal Tanabe et al., 1996; Taki et al., 1996). Transcripts for fragment of PKCB showed preferential binding to the reciprocal AF6/MLL fusion have not been detected Ras-GTP in vitro and a peptide corresponding to Ras (Tanabe et al., 1996; Taki et al., 1996). Interestingly, all residues 17 ± 44 blocked this interaction. (Diaz-Meco et MLL/AF-6 fusion proteins contain the identical AF-6 al., 1994). Furthermore, PDGF stimulation promoted residues, 36 ± 1612. In vitro analyses showed that the PKCB association with Ras in vivo. Finally, dominant

NH2-terminal domains of AF6 and Canoe interacted negative H-Ras(17N) blocked PDGF-stimulated acti- speci®cally with GTP-bound form of Ras and this vation of PKCB. Taken together, these observations interaction interferes the binding of Ras with Raf suggest that PKCB may serve as a positive regulator of (Kuriyama et al., 1996). AF-6 appears to be expressed Ras growth stimulation. in a variety of tissues. Presently, a function for AF-6 In summary, a growing family of candidate Ras has not been determined. However, limited sequence e€ectors has emerged. However, to date, the precise homology was observed between AF-6 and proteins roles, if any, of these Ras-GTP binding proteins in that may be involved in signal transduction at special mediating Ras downstream signal transduction and cell-cell junctions. transformation remains to be established. It is likely Increasing complexity of Ras signaling SL Campbell et al 1406 that some of these proteins may be involved in GAPs for Ras, as well as Ras-related proteins have mediating normal Ras function while others may be been implicated as possible e€ectors. The X-ray critical mediators of Ras transforming activity. A structure of the p120 Ras GAP catalytic domain (Ras subset of these proteins may actually be negative GAP-CAT) has recently been solved (Sche€zek et al., regulators of Ras function by preventing the interac- 1996) and shown to have an all-helical fold that is tion of Ras with bona ®de Ras e€ectors. Moreover, it is distinct from GAP domains of two RhoGAPs (p85 and possible that distinct mammalian Ras (H-Ras, K-Ras, p50 RhoGAP) (Musacchio et al., 1996; Barrett et al.,

N-Ras) proteins and/or Ras mutants (for example, 1997) and Gai1-RGS (Tesmer et al., 1997). Hence, even positions 12,13,61) may di€er in their ability to GAP domains from Ras-related proteins and hetero- recognize and activate downstream targets, resulting trimer G-proteins with similar functional properties in di€erential signaling (Yan et al., 1997). For example, show distinct tertiary folds. The overall topology of gene targeting studies indicate an essential role for K- Ras GAP-CAT, in turn, is quite distinct from that of ras, but not H- or N-Ras, for normal development in Raf-RBS (Emerson et al., 1995; Nassar et al., 1995), in mice (Koera et al., 1997; Johnson et al., 1997). This that Raf-RBS possesses a mixed a/b fold similar to that data, taken together with earlier observations that K- of ubiquitin. Interestingly, despite the low level of Ras-4B possesses distinct carboxy terminus modifica- sequence homology between Raf-RBS and the Ras tions (James et al., 1995) and the speci®c association of interaction domain of RalGDS (RalGDS-RID), Raf- KRas4B with smg-GDP dissociation stimulator (GEF) RBS has been shown to have a similar tertiary fold to (Mizuno et al., 1991), suggests that K-Ras may have RalGDS-RID (Huang et al., 1997; Geyer et al., 1997). speci®c functions in signal transduction not shared by In fact, NMR chemical shift mapping studies of other family members. This might explain why most RalGDS-RID in a complex with Ras, indicate an Ras mutants in human tumors contain K-Ras interaction similar to the intermolecular beta-sheet mutations at position 12 (Barbacid, 1987; Bos, 1989; observed for the complex between Ras and Raf-RBS Clark and Der, 1993). Finally, it is likely that a portion (Geyer et al., 1997). Furthermore, mutagenesis studies of these candidate e€ectors may not serve as true Ras indicate that several residues in RalGDS-RID that e€ectors, but rather e€ectors for one or more of the mediate binding to Ras are located at sites that closely related proteins such as TC21, R-Ras or Rap. correpond to those important for binding by Raf- Analysis of the role of these proteins in the function of RBS (Huang et al., 1997; Geyer et al., 1997). Hence, it Ras and Ras-related proteins is the focus of ongoing is tempting to speculate, given the structural and investigations in many laboratories. mutagenesis data, that the RalGDS-Ras complex will be similar to the Ras-Raf complex, and may provide a Structural requirements for Ras : e€ector interac- common structural basis for interaction with a subset tions The current roster of candidate Ras e€ectors is of Ras e€ectors. Moreover, using N- and C-terminal comprised of a very diverse collection of structurally deletions of Rlf, the minimal e€ective Ras binding and functionally distinct proteins. The conformation of domain of the RalGDS-related protein, RLF, was Ras-GTP di€ers from Ras-GDP in two distinct determined to encompass residues 657 ± 778 (O'Gara et regions, designated switch I (Ras residues 30 ± 37) and al., 1997). The secondary structure of this domain was switch II (Ras residues 59 ± 76) (Tong et al., 1989; also predicted to be similar to the ubiquitin fold Milburn et al., 1990; Krengel et al., 1990). Since switch previously determined for Raf-RBS. Alignment of Rlf- I overlaps with the Ras e€ector domain sequences, a RBS with Raf-RBS and mutagenesis of K687 in Rlf- common feature of candidate Ras e€ectors is the loss RBS suggest conservation of amino acids in Raf-1 of binding to Ras-GTP due to mutations in the core essential for Ras binding to Rlf (O'Gara et al., 1997). Ras e€ector domain (32 ± 40). Thus, it is not surprising Thus, it appears, that both a common structural that this region is important for e€ector interaction manifold and conservation of critical complementary and may represent a shared binding region important Ras e€ector interactions may represent shared Ras- for all Ras e€ectors. A Ras-GTP binding motif has binding properties among a subset of Ras e€ectors. It recently-been proposed based on limited sequence and will be interesting to see whether other candidate Ras predicted structural similarity between RalGDS and e€ectors share a similar fold. AF-6, and designated the RA domain (Ponting and A characteristic feature of the GTP-bound form of Benjamin, 1996). The RA domain is predicted in a Ras is that the switch regions are dynamic and undergo number of candidate Ras e€ectors (RalGDS, AF-6/ interconversion between two or more conformers. This Canoe and Rin-1), but not in others (Ras GAPs, PI3K, conformational polymorphism has been proposed to MEKK1). Although the alignment and hypothesis is mediate recognition and interaction with regulators intriguing, Kalhammer et al. (1997) failed to detect and downstream targets of Ras, such as Raf-1 and Ras direct binding between the putative RA domain in Myr GAPs (Ito et al., 1997; Geyer et al., 1996). Although a 5 and Ras, suggesting that the identifed RA domain common feature of Ras targets, are binding determi- does not reliably predict Ras-binding domains. nants present in the GTP-bound form of the Ras The speculation that some candidate e€ectors share a e€ector domain, it is possible that each target common tertiary structure, despite the distinct primary preferentially recognizes and binds a distinct confor- sequences seen among some Ras-GTP binding proteins, mer presented by the polysteric binding interface of is intriguing. However, the ability of speci®c amino acid Ras. Moreover, structural and mechanistic studies of mutations to cause selective impairment of only a subset the GAP-stimulated GTP hydrolysis of Ras indicate of Ras-GTP binding partners suggests that the binding that RasGAPs supply a catalytic residue, termed the determinants will be distinct. The recent determination arginine ®nger, into the active site of Ras to neutralize of the structures of three Ras-GTP binding domains developing charges in the transition state. Moreover, it support both possibilities. has been proposed, that GAP proteins also facilitate Increasing complexity of Ras signaling SL Campbell et al 1407 GTP hydrolysis by binding and reducing the mobility whereas only WT Ras-GTP, binds to Raf-1 with high within the GTPase catalytic center (Sche€zek et al., anity leading to Raf-1 activation, both Q61L Ras- 1997; Ahmadian et al., 1997; Wittinghofer et al., 1997). GDP and Q61L Ras-GTP bind Raf-1 with reasonable Therefore, a number of factors appear to be involved anity and can lead to its activation (Moodie et al., in target/regulator interactions with Ras including: the 1995). A possible explanation for these observations, is structural manifold, presence of both common and that the mutation alters another a region of the protein distinct complementary binding interactions, and whose conformation is not GTP-dependent such that it conformational mobility. increases binding anity to Raf-1 in a GTP- One additional complexity to this issue is the independent manner. It is intriguing that the corre- likelihood that the various Ras-binding proteins are sponding mutation in Rac restores binding and likely to interact with distinct and multiple sequences activation of NADPH oxidase interactions when in Ras (Figure 4). Thus, the original de®nition of the placed in the context of Rac e€ector domain Ras e€ector domain (Sigal et al., 1986) is clearly an mutations (Dorseuil et al., 1996) and lends further over simpli®cation. The crystal structure of Rap bound support for this hypothesis. So, the unusual nature of to the slowly-hydrolyzable GTP analog, GppNHp, this mutation may be a conserved feature of at least of demonstrated that Raf-RBS (residues 51 ± 131) contacts subset of Ras-related GTPases. Whether switch II is Rap primarily though residues 30 ± 41, in and around important for interaction with all Ras e€ectors is switch I region (Nassar et al., 1995). However, not all presently not known. Mutation of yet other residues the genetic data can be rationalized from these critical (e.g., residues 92 or 106) suggests a third region of contacts between Ras/Rap and Raf-1. Mutation of contact between Ras and Ras GAPs and possibly other residues ¯anking the core e€ector region (spanning Ras e€ectors (Morcos et al., 1996; Parrini et al., 1996; residues 26 ± 45) can also in¯uence Ras transforming Yoder-Hill et al., 1995). The three dimensional activity and Ras interaction with candidate e€ectors structure of Ras GDP bound to GAP-CAT has (Shirouzu et al., 1994; Fujita-Yoshigaki et al., 1995). In recently been solved (Sche€zek et al., 1997) and particular, Ras variants containing mutations at 26 or consistent with mutational studies, residues of the 45 which are close to the interface but do not make phosphate binding loop (residues 10 to 16), switch contact with RBS, retain high anity binding to RBS, regions I (30 to 37) and II (59 to 76) and possibly helix but cannot activate Raf-1. These data suggest that a3 (87 to 98) on Ras participate in binding interactions. other elements of Ras and/or Raf-1 are involved in Furthermore, COOH-terminal farnesylation of Ras interactions important for activation, consistent with appears to be important for high anity interaction recent data that Ras regions ¯anking residues 31 ± 41 with Raf, suggesting that the Ras COOH-terminus may (particularly 26 and 45) are involved in additional also be involved in e€ector interaction (Hu et al., 1995; contacts with Raf-1 via the Raf-CRD domain (Hu et Luo et al., 1997). The presence of the farnesyl group al., 1995). Interestingly, the binding contacts between also appears to be important for Ras2 activation of unprocessed and processed forms of Ras and the adenylate cyclase (Kuroda et al., 1993). cysteine-rich domain do not show a strong preference Although it is clear that Ras sequences important for either Ras-GTP or Ras-GDP, consistent with for e€ector interactions are complex and involve binding interactions outside the switch domains (Hu multiple distinct sequences, the core e€ector region et al., 1995; Drugan et al., 1996). In Raf-1, the initial (32 ± 40) appears to be an essential element for all contact appears to be mediated by Raf-RBS and Ras- e€ector interactions. Several members of the Ras GTP, which is necessary for Raf-1 translocation, branch of the Ras superfamily share complete identity whereas the cysteine-rich domain of Raf-1 appears to with Ras in this core e€ector domain, including the be required for full activation of Raf-1 activity by Ras. Rap proteins, R-Ras and TC21 (Figure 4). Whether It will be interesting to see if other Ras targets possess these Ras-related proteins share e€ector utilization similar modes of recognition and activation, where with Ras, or whether some Ras binding proteins are membrane localization is mediated by the Ras-GTP really e€ectors for Ras-related proteins remains binding to the switch I domain, and target activation incompletely understood. Hence, studies of Ras- by other regions of Ras. related proteins may shed some light on the Ras Similarly, mutagenesis of switch II residues also sequences and e€ectors important for Ras transforming support their importance in interaction with at least activity. some candidate e€ectors (Moodie et al., 1995; Hwang Although several members of the Ras branch of the et al., 1996; Drugan et al., 1996). For example, Y64G/ Ras superfamily share complete identity with Ras S65G or Y64G/Y71G double mutants of H-Ras were residues 32 ± 40 and can also bind to many Ras-GTP impaired in complex formation with NF1 or PI3K binding proteins, they exhibit distinct biochemical and (Moodie et al., 1995). Furthermore, mutation of biological properties. For example, although activated residues 60 or 64 also perturbed Ras interaction with forms of TC21 cause the same transforming and Raf-1 (Drugan et al., 1996). However, NMR structural di€erentiating actions as Ras, TC21 fails to bind to analysis of these Ras variants suggest that the and activate Raf kinases (Graham et al., 1996). We mutations produce structural changes remote from have found that, like Ras, TC21 can cause growth the site of mutation and thus these residues may not be transformation of NIH3T3 and other cells (Graham et directly involved in contacts with Raf-1 (Sharon al., 1994), promote the di€erentiation of PC12 Campbell, unpublished observations). The Q61L pheochromocytoma cells, and block serum starvation- oncogenic mutation increases anity for both GAPs induced di€erentiation of C2 skeletal myoblasts (Vogel et al., 1988), consistent with the recently (Graham et al., in preparation). Thus, it will be determined structure of Ras complexed with the interesting to determine if TC21 utilizes any of the catalytic domain of p120 Ras-GAP. Interestingly, candidate e€ectors for Ras to mediate these activities. Increasing complexity of Ras signaling SL Campbell et al 1408 Our yeast two-hybrid binding analyses have deter- proteins, although the previously noted sequence mined that TC21 can bind some (e.g., RalGDS, AF6), homology with Rap1A is not observed in our but not all (Rin1 or Raf-1) known Ras binding sequence alignment (Yamagata et al., 1994). For proteins (Graham et al., in preparation). These example, constitutively activated mutants of Rheb properties of TC21 emphasize the critical role of failed to cause growth or morphologic transformation sequences outside the core e€ector region in specifying of NIH3T3 cells. Rheb has also been shown to interact not only e€ector binding but also e€ector activation, with Raf-1 (Clark et al., 1997c). Hence, the lack of and further support the importance of Raf-independent transforming potential exhibited by Rheb may result pathways in mediating transformation. from antagonizing Ras signaling and transformation Even though activated mutants of R-Ras can cause possibly by formation of a nonproductive complex transformation of NIH3T3 cells, they clearly di€er with Raf and other Ras e€ectors. Recent ®ndings that from Ras and TC21 in biological activity. For example, lend support for this hypothesis indicate that Rheb R-Ras transformed NIH3T3 cells lack the strong binds but does not activate Raf-1, e€ectively inhibiting morphologic transformation seen with Ras trans- the activation of Raf-1 by Ras. Intriguingly, although formed cells. Furthermore, whereas activated Ras or Rap proteins do not activate Raf-1, they will stimulate TC21 can overcome the loss of endogenous Ras the kinase activity of B-Raf (Vossler et al., 1997). We function caused by the Ras(17N) dominant negative, have recently found that activated forms of Rheb may activated R-Ras apparently lacks the Ras functions exhibit similar properties (Geo€ Clark, unpublished required to maintain normal cell proliferation (Hu€ et observations). al., 1997). R-Ras may regulate cellular processes Recently, additional small GTPases have been (apoptosis, integrin-mediated cellular adhesion) dis- identi®ed that also share strong, but incomplete, tinct from Ras (Zhang et al., 1996; Hu€ et al., 1997; identity with the core Ras e€ector domain and Marte et al., 1997). Like TC21, R-Ras does not ¯anking sequences. Using a PCR-based approach demonstrate high anity binding to Raf or promote with degenerate primers corresponding to conserved kinase activation. However, like Ras, R-Ras can regions in Ras and employing as template, mouse activate a PI3K/PKB pathway (Marte et al., 1997). retinal cDNA, three novel small GTPases were cloned Nevertheless, since activated PI3K or PKB does not (Lee et al., 1996). One of the two GTPases, named Rin share the same transforming potential as R-Ras, R-Ras for Ras-like protein expressed in neurons, binds must utilize other e€ectors as well. calmodulin in a Ca2+-dependent manner, suggesting it Although Rap1A/Krev-1 can interact with most Ras may be involved in mediating calcium-dependent binding proteins, such as to Raf-1, GAPs and RalGDS signaling within neurons (Lee et al., 1996). Rit, on (Zhang et al., 1993; Frech et al., 1990; Hata et al., the other hand, stands for Ras-like protein expressed in 1990; Kikuchi et al., 1994; Hofer et al., 1994; Wolthuis many tissues. Like Rin, Rit does not contain a et al., 1996; Spaargaren and Bischo€, 1994; Peterson et sequence for prenylation or palmitoylation but al., 1996; Lo pez-Barahona et al., 1996), activated forms nontheless, both appear to be plasma membrane of Rap1A fail to cause transformation. Instead, Rap1A localized. Both proteins contain a similar core effector has been shown to antagonize Ras transforming domain to Ras but contain histidine instead of tyrosine activity. This antagonism has been attributed to the at position 32 and an alanine instead of serine at ability of Rap1A to form nonproductive complexes position 39 (Figure 4). Finally, an unpublished Ras- with key Ras e€ectors important for Ras transforming related protein, designated M-Ras, has also been activity. This may occur because Rap1A may bind, but identi®ed. Whether Rit, Rin or M-Ras share common fail to activate some Ras e€ectors such as Raf-1 e€ectors with Ras, and whether constitutively activated (Kitayama et al., 1989, 1990). Alternatively, since mutants will cause cellular transformation is presently Rap1A is located in the endoplasmic reticulum and not known. The signaling pathways mediated by these Golgi, it may sequester Ras e€ectors in a subcellular novel GTPases are currently under investigation. location that prevents their complete activation to or from functional signaling complexes (Beranger et al., Increasing complexity of Ras signal transduction: a boon 1991; Sato et al., 1994; Cook et al., 1993). The ability or a bust for drug discovery and the development of anti- of Rap1A to block Ras activation of ERKs suggests Ras drugs for cancer treatment? that Raf-1 is one of the Ras e€ectors targeted by this Ras antagonist. However, since it has been described A promising approach for targeting Ras for cancer that activated Rap can promote the proliferation of treatment involves the use of farnesyltransferase Swiss 3T3 cells, Rap1A may use Ras e€ectors, or Rap- inhibitors (FTIs) (Cox and Der, 1997). FTIs block speci®c e€ectors, to mediate functions beyond antag- Ras function by preventing its posttranslational onizing Ras (Yoshida et al., 1992; Franke et al., 1997). modi®cation by the farnesyl isoprenoid. Although Another Ras subfamily protein, designated Rheb, current evidence support the potential use of FTIs for Ras homolog enriched in brain, has been described for cancer treatment, some concerns exist since FTIs recently (Yamagata et al., 1994). The observation that block all Ras function. Thus, since Ras protein Rheb retains conservation with Ras in six out of nine function is believed to be central to so many cellular residues in the core e€ector domain suggests that Rheb processes, targeting only a subset of Ras functions by may share common e€ector interactions with Ras downstream intervention may provide signi®cant (Figure 4). Similar to Ras, Rheb also encodes a advantages. Therefore, another approach for targeting CAAX box that is farnesylated and sensitive to Ras protein function involves the development of farnesyltransferase inhibition (Clark et al., 1997c). inhibitors that prevent activated Ras from relaying its Our analyses of Rheb function suggest that Rheb downstream signal by interfering at di€erent points in shares greater functional relationship with Rap its signaling cascades (Gibbs and Oli€, 1994; Der et al., Increasing complexity of Ras signaling SL Campbell et al 1409 1996). For example, components of the of Ras function. If Raf serves as a prototypic Ras Raf4MEK4MAPK cascade has been targeted by e€ector, Ras binding and e€ector activation are likely ongoing e€orts at a number of pharmaceutical to be mediated through multiple interaction sites, via a companies. However, because of the revelations multistep activation process requiring complex forma- indicating the complexity of Ras signaling pathways tion with other factors. Will other e€ectors prove to be required for transformation are complex, such e€orts equally important? Of those that have been identi®ed, may be premature. Instead, it may be better to wait which Ras-binding partners will be legitimate Ras until full appreciation of the involvement of Ras in the targets? Clearly, Raf and PI3K appear to be important signaling circuitry that regulates cell growth and ones. If other e€ectors are validated, de®ning their di€erentiation can be achieved. contribution to Ras function will be important. Some, While the unknown complexities of Ras signaling but not all, are likely to be important mediators of Ras pose serious concerns with regard to identifying the transformation. Are more e€ectors yet to be found? most appropriate and key downstream components for This seems very likely, including tissue-speci®c intervention, there is also evidence that intervention at e€ectors. any number of points may be promising. Ras Several questions relating to the regulation of Ras transformation of rodent ®broblasts can be antago- function via e€ector utilization remain. If Raf and nized by dominant negative mutants of Raf, MEK or other e€ectors lack GAP function, how is the MAPK (Kolch et al., 1991; Schaap et al., 1993; Cowley interaction terminated? Most are ubiquitously ex- et al., 1994; Westwick et al., 1994; Qiu et al., 1995a; pressed. Consequently, how is speci®c e€ector utiliza- Khosravi-Far et al., 1995), by dominant negative Rho tion regulated? Does the cell regulate e€ector family proteins (Qiu et al., 1995a,b; Khosravi-Far et utilization by controlling levels of expression? Will al., 1995; Prendergast et al., 1995), by blocking the Ras activation by di€erent Ras GEFs cause di€erent action of speci®c transcription factors (Myc, Fos, Jun, subcellular locations that then promote distinct e€ector Ets and NFkB) (Sklar et al., 1991; Granger-Schnarr et utilization? Will the formation of distinct signaling al., 1992; Langer et al., 1992; Clark et al., 1997d; Finco complexes, with di€erent combinations of e€ectors, et al., 1997; Mayo et al., 1997). These observations dictate the function of each e€ector? Will cell type provoke an apparent paradox. If mutants of Ras that di€erences in e€ector utilization be seen? Will each fail to activate the Raf4MEK4MAPK pathway can e€ector mediate the distinct biological properties of still cause transformation, then why does blocking this Ras (growth versus di€erentiation versus apoptosis). kinase cascade still result in potent inhibition of Ras For example, recent studies have implicated speci®c transformation? One possible resolution to this Rho family proteins as Raf-independent mediators of paradox is that full Ras transformation requires the Ras transformation (Rac, Rho and CDC42). Which action of multiple pathways. However, the impairment e€ector(s) connect Ras to Rho family proteins will be of any one of several key pathways alone will important to determine. Will common e€ector utiliza- signi®cantly impair Ras transforming function, and tion provide cross talk between di€erent Ras-related consequently, may have a signi®cant impact on proteins? Will distinct e€ector utilization be seen for reversing the malignant and invasive growth proper- the four Ras proteins? Will one e€ector pathway be a ties of cancer cells. Thus, while many components of key target for drug discovery or will di€erent e€ector Ras signaling remain to be identi®ed, and their precise pathways be important in di€erent tumors? role in Ras transformation established, we may already have sucient knowledge to target Ras signaling to initiate these drug discovery e€orts. Acknowledgements Future directions We thank Adrienne Cox for stimulating discussions and communicating unpublished results and Jennifer Parrish Despite a rapid and impressive accumulation of for excellent assistance in the preparation of the text, information regarding Ras e€ector utilization, we are references and ®gures. Our studies are supported by NIH at a point in discovery where there are more questions grants to CJD (CA42978, CA52072, CA55008 and than answers. What is clear is that Ras does not simply CA67771), GJC grants (CA72644-10 and DAMD17-97-1- activate Raf. Presently, Raf remains the major e€ector 7050) and SC (CA64569 and CA70308).

References

Adari H, Lowy DR, Willumsen BM, Der CJ and McCormick Alessi DR, Saito Y, Campbell DG, Cohen P, Sithanandam F. (1988). Science, 240, 518 ± 521. G, Rapp U, Ashworth A, Marshall CJ and Cowley S. Afar DE, Han L, McLaughlin J, Wong S, Dhaka A, Parmar (1994). EMBO J., 13, 1610 ± 1619. K, Rosenberg N, Witte ON and Colicelli J. (1997). Andersen LB, Fountain JW, Gutmann DH, Tarle SA, Immunity, 6, 773 ± 782. Glover TW, Dracopoli NC, Housman DE and Collins Ahmadian MR, Stege P, Sche€zek K and Wittinghofer A. FS. (1993). Nature Genet., 3, 118 ± 126. (1997). Nat. Struct. Biol., 4, 686 ± 689. Barbacid M. (1987). Annu.Rev.Biochem.,56, 779 ± 827. Akasaka K, Tamada M, Wang F, Kariya K, Shima F, Barnier JV, Papin C, Eychene A, Lecoq O and Calothy G. Kikuchi A, Yamamoto M, Shirouzu M, Yokoyama S and (1995). J. Biol. Chem., 270, 23381 ± 23389. Kataoka T. (1996). J. Biol. Chem., 271, 5353 ± 5360. Barrett T, Xiao B, Dodson EJ, Dodson G, Ludbrook SB, Albright CF, Giddings BW, Liu J, Vito M and Weinberg RA. Nurmahomed K, Gambin SJ, Musacchio A, Smerdon SJ (1993). EMBO J., 12, 339 ± 347. and Eccleston JF. (1997). Nature, 385, 458 ± 461. Increasing complexity of Ras signaling SL Campbell et al 1410 Beranger F, Goud B, Tavitian A and de Gunzburg J. (1991). Crews CM, Alessandrini A and Erikson RL. (1992). Cell Proc. Natl. Acad. Sci. USA, 88, 1606 ± 1610. Growth and Di€., 3, 135 ± 142. Berra E, Diaz-Meco MT, Dominguez I, Municio MM, Sanz Crews CM and Erikson RL. (1993). Cell, 74, 215 ± 217. L, Lozano J, Chapkin RS and Moscat J. (1993). Cell, 74, Der CJ, Cox AD, Sebti SM and Hamilton AD. (1996). Anti- 555 ± 563. Cancer Drugs, 7, 165 ± 172. Blenis J. (1993). Proc. Natl. Acad. Sci. USA, 90, 5889 ± 5892. De rijard B, Hibi M, Wu I-H, Barrett T, Su B, Deng T, Karin Boguski MS and McCormick F. (1993). Nature, 366, 643 ± M and Davis RJ. (1994). Cell, 76, 1025 ± 1037. 654. Diaz-Meco MT, Lozano J, Municio MM, Berra E, Frutos S, Bollag G, Clapp DW, Shih S, Adler F, Zhang YY, Sanz L and Moscat J. (1994). J. Biol. Chem., 269, 31706 ± Thompson P, Lange BJ, Freedman MH, McCormick F, 31710. Jacks T and Shannon K. (1996). Nature Genet., 12, 144 ± Dickson B, Sprenger F, Morrison D and Hafen E. (1992). 148. Nature, 360, 600 ± 603. Bonner TI, Kerby SB, Sutrave P, Gunnell MA, Mark G and Dominguez I, Diaz-Meco MT, Municio MM, Berra E, Rapp UR. (1985). Mol. Cell. Biol., 5, 1400 ± 1407. GarcõÁ a de Herreros A, Cornet ME, Sanz L and Moscat J. Bos JL. (1989). Cancer Res., 49, 4682 ± 4689. (1992). Mol. Cell. Biol., 12, 3776 ± 3783. Bourne HR, Sanders DA and McCormick F. (1990). Nature, Dorseuil O, Reibel L, Bokoch GM, Camonis J and Gacon G. 349, 117 ± 126. (1996). J. Biol. Chem., 271, 83 ± 88. Braselmann S and McCormick F. (1995). EMBO J., 14, Drugan JK, Khosravi-Far R, White MA, Der CJ, Sung Y-J, 4839 ± 4848. Huang Y-W and Campbell SL. (1996). J. Biol. Chem., 271, Broach JR. (1991). TIG, 7, 28 ± 33. 233 ± 237. Brtva TR, Drugan JK, Ghosh S, Terrell RS, Campbell-Burk Egan SE and Weinberg RA. (1993). Nature, 365, 781 ± 783. S, Bell RM and Der CJ. (1995). J. Biol. Chem., 270, 9809 ± Ellis C, Moran M, McCormick F and Pawson T. (1990). 9812. Nature, 343, 377 ± 380. Cale s C, Hancock JF, Marshall CJ and Hall A. (1988). Emerson SD, Madison VS, Palermo RE, Waugh DS, Nature, 332, 548 ± 551. Sche‚er JE, Tsao KL, Kiefer SE, Liu SP and Fry DC. Cantor SB, Urano T and Feig LA. (1995). Mol. Cell. Biol., (1995). Biochemistry, 34, 6911 ± 6918. 15, 4578 ± 4584. Eychene A, Dusanter-Fourt I, Barnier JV, Papin C, Charon Cardone MH, Salvesen GS, Widmann C, Johnson G and M, Gisselbrecht S and Calothy G. (1995). Oncogene, 10, Frisch SM. (1997). Cell, 90, 315 ± 323. 1159 ± 1165. Carpenter CL and Cantley LC. (1996a). Curr. Opin. Cell Fanger GR, Johnson NL and Johnson GL. (1997). EMBO Biol., 8, 153 ± 158. J., 16, 4961 ± 4972. Carpenter CL and Cantley LC. (1996b). Biochim. Biophys. Fantl WJ, Escobedo JA, Martin GA, Turck CW, del Rosario Acta., 1288, M11 ± M16. M, McCormick F and Williams LT. (1992). Cell, 69, 413 ± Chan AML, Miki T, Meyers KA and Aaronson SA. (1994). 423. Proc. Natl. Acad. Sci. USA, 91, 7558 ± 7562. FantlWJ,MuslinAJ,KikuchiA,MartinJA,MacNicolAM, Chang EC, Barr M, Wang Y, Jung V, Xu H-P and Wigler Gross RW and Williams LT. (1994). Nature, 371, 612 ± MH. (1994). Cell, 79, 131 ± 141. 614. Chang HC and Rubin GM. (1997). Genes Dev., 11, 1132 ± Feig LA. (1994). Curr. Opin. Cell Biol., 6, 204 ± 211. 1139. Feig LA and Cooper GM. (1988). Mol. Cell. Biol., 8, 3235 ± Chen Q, Lin TH, Der CJ and Juliano RL. (1996a). J. Biol. 3243. Chem., 271, 18122 ± 18127. Finco TS, Westwick JK, Norris JL, Beg AA, Der CJ and Chen YR, Meyer CF and Tan T-H. (1996b). J. Biol. Chem., Baldwin Jr AS. (1997). J. Biol. Chem., 272, 24113 ± 24116. 271, 631 ± 634. Franke B, Akkerman JW and Bos JL. (1997). EMBO J., 16, Chuang E, Barnard D, Hettich L, Zhang X-F, Avruch J and 252 ± 259. Marshall MS. (1994). Mol. Cell. Biol., 14, 5318 ± 5325. Franke TF, Ynag S-I, Chan TO, Datta K, Kazlauskas A, Clark GJ, Quilliam LA, Hisaka MM and Der CJ. (1993). Morrison DK, Kaplan DR and Tsichlis PN. (1995). Cell, Proc. Natl. Acad. Sci. USA, 90, 4887 ± 4891. 81, 727 ± 736. Clark GJ, Drugan JK, Terrell RS, Bradham C, Der CJ, Bell Franke TF, Kaplan DR and Cantley LC. (1997). Cell, 88, RM and Campbell-Burk S. (1995). Proc. Natl. Acad. Sci. 435 ± 437. USA, 93, 1577 ± 1581. Frech M, John J, Pizon V, Chardi P, Tavitian A, Clark R, Clark GJ, Drugan JK, Ghosh S, Rossman K, Bell RM, Der McCormick F and Wittinghofer A. (1990). Science, 249, CJ and Campbell S. (1997a). J. Biol. Chem., submitted. 169 ± 171. Clark GJ, Drugan JK, Rossman KL, Carpenter JW, Rogers- Freed E, Symons M, Macdonald SG, McCormick F and Graham K, Fu H, Der CJ and Campbell SL. (1997b). J. Ruggieri R. (1994). Science, 265, 1713 ± 1716. Biol. Chem., 272, 20990 ± 20993. Fujita-Yoshigaki J, Shirouzu M, Ito Y, Hattori S, Furuyama Clark GJ, Kinch MS, Rogers-Graham K, Sebti SM, S, Nishimura S and Yokoyama S. (1995). J. Biol. Chem., Hamilton AD and Der CJ. (1997c). J. Biol. Chem., 272, 270, 4661 ± 4667. 10608 ± 10615. Gangarosa LM, Sizemore N, Graves-Deal R, Oldham SM, Clark GJ, Westwick JK and Der CJ. (1997d). J. Biol. Chem., Der CJ and Co€ey RJ. (1997). J. Biol. Chem., 272, 18926 ± 272, 1677 ± 1681. 18931. Clark GJ, Quilliam LA, Hisaka MM and Der CJ. (1993). Geyer M, Schweins T, Herrmann C, Prisner T, Wittinghofer Proc. Natl. Acad. Sci. USA, 90, 4887 ± 4891. A and Kalbitzer HR. (1996). Biochemistry, 35, 10308 ± Conklin DS, Galaktionov K and Beach D. (1995). Proc. 10320. Natl. Acad. Sci. USA, 92, 7892 ± 7896. Geyer M, Herrmann C, Wohlgemuth S, Wittinghofer A and Cook SJ, Rubinfeld B, Albert I and McCormick F. (1993). Kalbitzer HR. (1997). Nat. Struct. Biol., 4, 694 ± 699. EMBO J., 12, 3475 ± 3485. Ghosh S, Xie WQ, Quest AFG, Mabrouk GM, Strum JC, Cowley S, Paterson H, Kemp P and Marshall CJ. (1994). and Bell RM. (1994). J. Biol. Chem., 269, 10000 ± 10007. Cell, 77, 841 ± 852. Gibbs JB and Oli€ A. (1994). Cell, 79, 193 ± 198. Cox AD, Brtva TR, Lowe DG and Der CJ. (1994). Oncogene, GrahamSM,CoxAD,DrivasG,RushMR,D'EustachioP 9, 3281 ± 3288. and Der CJ. (1994). Mol. Cell. Biol., 14, 4108 ± 4115. Cox AD and Der CJ. (1997). Biochimica. Biophysica. Acta., 1333, F51 ± F71. Increasing complexity of Ras signaling SL Campbell et al 1411 Graham SM, Vojtek AB, Hu€ SY, Cox AD, Clark GJ, Kikuchi A, Demo SD, Ye Z-H, Chen Y-W and Williams LT. Cooper JA and Der CJ. (1996). Mol. Cell. Biol., 16, 6132 ± (1994). Mol. Cell. Biol., 14, 7483 ± 7491. 6140. Kikuchi A and Williams LT. (1996). J. Biol. Chem., 271, Grammer TC and Blenis J. (1997). Oncogene, 14, 1635 ± 588 ± 594. 1642. Kim HA, Ling B and Ratner N. (1997). Mol. Cell. Biol., 17, Granger-Schnarr M, Benusiglio E, Schnarr M and Sassone- 862 ± 872. Corsi P. (1992). Proc. Natl. Acad. Sci. USA, 89, 4236 ± Kitayama H, Sugimoto Y, Matsuzaki T, Ikawa Y and Noda 4239. M. (1989). Cell, 56, 77 ± 84. HanL,WongD,DhakaA,AfarD,WhiteM,XieW, Kitayama H, Matsuzaki T, Ikawa Y and Noda M. (1990). Herschman H, Witte O and Colicelli J. (1997). Proc. Natl. Proc. Natl. Acad. Sci. USA, 87, 4284 ± 4288. Acad. Sci. USA, 94, 4954 ± 4959. Klippel A, Reinhard C, Kavanaugh M, Apell G, Escobedo Han L and Colicelli J. (1995). Mol. Cell. Biol., 15, 1318 ± M-A and Williams LT. (1996). Mol. Cell. Biol., 16, 4117 ± 1323. 4127. Hartwell L. (1992). Cell, 71, 543 ± 546. Klippel A, Kavanaugh WM, Pot D and Williams LT. (1997). Hata Y, Kikuchi A, Sasaki T, Schaber MD, Gibbs JB and Mol. Cell. Biol., 17, 338 ± 344. Takai Y. (1990). J. Biol. Chem., 265, 7104 ± 7107. Kodaki T, Woscholski R, Hallberg B, Rodriguez-Viciana P, Henkenmeyer M, Rossi DJ, Holmyard DP, Puri MC, Downward J and Parker PJ. (1994). Curr. Biol., 4, 798 ± Mbamalu G, Harpal K, Shih TS, Jacks T and Pawson T. 806. (1995). Nature, 377, 695 ± 701. Koera K, Nakamura K, Nakao K, Miyoshi J, Toyoshima K, Hofer F, Fields S, Schneider C and Martin GS. (1994). Proc. Hatta T, Otani H, Aiba A and Katsuki M. (1997). Natl. Acad. Sci. USA, 91, 11089 ± 11093. Oncogene, 15, 1151 ± 1160. Hu C, Kariya K, Tamada M, Akasaka K, Shirouzu M, KolchW,HeideckerG,LloydPandRappUR.(1991). Yokoyama S and Kataoka T. (1995). J. Biol. Chem., 270, Nature, 349, 426 ± 428. 30274 ± 30277. Kolch W, Heidecker G, Kochs G, Hummel R, Vahidi H, Hu Q, Klippel A, Muslin AJ, Fantl WJ and Williams LT. Mischak H, Finkenzeller G, Marme D and Rapp UR. (1995). Science, 268, 100 ± 102. (1993). Nature, 364, 249 ± 252. Huang L, Weng X, Hofer F, Martin GS and Kim SH. (1997). Krengel U, Schlichting L, Scherer A, Schumann R, Frech M, Nat. Struct. Biol., 4, 609 ± 615. John J, Kabsch W, Pai EF and Wittinghofer A. (1990). Hu€ SY, Quilliam LA, Cox AD and Der CJ. (1997). Cell, 62, 539 ± 548. Oncogene, 14, 133 ± 143. Kuriyama M, Harada N, Kuroda S, Yamamoto T, Hunter T. (1997). Cell, 88, 333 ± 346. Nakafuku M, Iwamatsu A, Yamamoto D, Prasad R, Hwang M-CC, Sung Y-J and Hwang Y-W. (1996). J. Biol. Croce C, Canaani E and Kaibuchi K. (1996). J. Biol. Chem., 271, 8196 ± 8202. Chem., 271, 607 ± 610. Irie K, Gotoh Y, Yashar BM, Errede B, Nishida E and Kuroda S, Ohtsuka T, Yamamori B, Fukui K, Shimizu K Matsumoto K. (1994). Science, 265, 1716 ± 1719. and Takai Y. (1996). J. Biol. Chem., 271, 14680 ± 14683. Ito Y, Yamasaki K, Iwahara J, Terada T, Kamiya A, Kuroda Y, Suzuki N and Kataoka T. (1993). Science, 259, Shirouzu M, Muto Y, Kawai G, Yokoyama S, Laue ED, 683 ± 686. Walchli M, Shibata T, Nishimura S and Miyazawa T. Kyriakis J and Avruch J. (1996). BioEssays, 18, 567 ± 577. (1997). Biochemistry, 36, 9109 ± 9119. Lange-Carter CA, Pleiman CM, Gardner AM, Blumer KJ Izawa I, Tamaki N and Saya H. (1996). FEBS Lett., 382, 53 ± and Johnson GL. (1993). Science, 260, 315 ± 319. 59. Langer SJ, Bortner DM, Roussel MF, Sherr CJ and James GL, Goldstein JL and Brown MS. (1995). J. Biol. Ostrowski MC. (1992). Mol. Cell. Biol., 12, 5355 ± 5362. Chem., 270, 6221 ± 6226. Lee C-HJ, Della NG, Chew CE and Zack DJ. (1996). J. Jiang H, Luo J-Q, Urano T, Frankel P, Lu Z, Foster DA and Neurosci., 16, 6784 ± 6794. Feig LA. (1995). Nature, 378, 409 ± 412. Leevers SJ, Paterson HF and Marshall CJ. (1994). Nature, Johnson L, Greenbaum D, Cichowski K, Mercer K, Murphy 369, 411 ± 414. E, Schmitt E, Bronson RT, Umano€ H, Edelmann W, Lenormand P, McMahon M and Pouysse gur J. (1996). J. Kucherlapati R and Jacks T. (1997). Genes Dev., 11, Biol. Chem., 271, 15762 ± 15768. 2468 ± 2481. Li R, Havel C, Watson JA and Murray AW. (1993). Nature, Johnson MR, Look AT, DeClue JE, Valentine MB and 366, 82 ± 84. Lowy DR. (1993). Proc. Natl. Acad. Sci. USA, 90, 5539 ± Li S, Janosch P, Tanji M, Rosenfeld GC, Waymire JC, 5543. Mischak H, Kolch W and Sedivy JM. (1995). EMBO J., Johnson MR, DeClue JE, Felzmann S, Vass WC, Xu G, 14, 685 ± 696. White R and Lowy DR. (1994). Mol. Cell. Biol., 14, 641 ± Li Y and White R. (1996). Cancer Res., 56, 2872 ± 2876. 645. Lo pez-Barahona M, Bustelo XR and Barbacid M. (1996). Johnson NL, Gardner AM, Diener KM, Lange-Carter C, Oncogene, 12, 463 ± 470. Gleavy J, Jarpe MB, Minden A, Karin M, Zon LI and Luo Z, Tzivion G, Belshaw PJ, Vavvas D, Marshall M and Johnson GL. (1996). J. Biol. Chem., 271, 3229 ± 3237. Avruch J. (1996). Nature, 383, 181 ± 185. Joneson T, White MA, Wigler MH and Bar-Sagi D. (1996). Luo Z, Diaz B, Marshall MS and Avruch J. (1997). Mol. Science, 271, 810 ± 812. Cell. Biol., 17, 46 ± 53. Kalhammer G, Bahler M, Schmitz F, Jockel J and Block C. Mansour SJ, Matten WT, Hermann AS, Candia JM, Rong S, (1997). FEBS Lett., 414, 599 ± 602. Fukasawa K, Vande Woude GF and Ahn NG. (1994). Kau€mann-Zeh A, Rodriguez-Viciana P, Ulrich E, Gilbert Science, 265, 966 ± 970. C, Co€er P, Downward J and Evan G. (1997). Nature, 385, Marais R, Wynne J and Treisman R. (1993). Cell, 73, 381 ± 544 ± 548. 393. Khosravi-Far R, Solski PA, Clark GJ, Kinch MS, Burridge Marais R, Light Y, Paterson HF and Marshall CJ. (1995). K and Der CJ. (1995). Mol. Cell. Biol., 15, 6443 ± 6453. EMBO J., 14, 3136 ± 3145. Khosravi-Far R, White MA, Westwick JK, Solski PA, Marais R, Light Y, Paterson HF, Mason CS and Marshall Chrzanowska-Wodnicka M, Van Aelst L, Wigler MH CJ. (1997). J. Biol. Chem., 272, 4378 ± 4383. and Der CJ. (1996). Mol. Cell. Biol., 16, 3923 ± 3933. Marcus S, Polverino A, Chang E, Robbins D and Cobb MH. Khosravi-Far R and Der CJ. (1994). Cancer Metastasis Rev., (1995). Proc. Natl. Acad. Sci. USA, 92, 6180 ± 6184. 13, 67 ± 89. Marshall CJ. (1996). Curr. Opin. Cell Biol., 8, 197 ± 204. Increasing complexity of Ras signaling SL Campbell et al 1412 Marshall MS and Hettich LA. (1993). Oncogene, 8, 425 ± 431. Prasad R, Gu Y, Alder H, Nakamura T, Canaani O, Saito H, Marte BM, Rodriguez-Viciana P, WennstroÈ mS,WarnePH Huebner K, Gale RP, Nowell PC, Kuriyama K, Miyazaki and Downward J. (1997). Curr. Biol., 7, 63 ± 71. Y, Croce CM and Canaani E. (1993). Cancer Res., 53, Martin GA, Yatani A, Clark R, Conroy L, Polakis P, Brown 5624 ± 5628. AM and McCormick F. (1992). Science, 255, 192 ± 194. Prendergast GC, Khosravi-Far R, Solski PA, Kurzawa H, Mayer BJ and Baltimore D. (1993). Trends Cell Biol., 3, 8± Lebowitz PF and Der CJ. (1995). Oncogene, 10, 2289 ± 13. 2296. Mayo MW, Wang C-Y, Cogswell PC, Rogers-Graham KS, Pumiglia K, Chow Y-H, Fabian J, Morrison D, Decker S Lowe SW, Der CJ and Baldwin Jr AS. (1997). Science, in and Jove R. (1995). Mol. Cell. Biol., 270, 398 ± 406. press, QiuR-G,ChenJ,KirnD,McCormickFandSymonsM. McCormickF,MartinGA,ClarkR,BollagGandPolakisP. (1995a). Nature, 374, 457 ± 459. (1991). Cold Spring Harb. Sym. Quant. Biol., 56, 237 ± 241. Qiu R-G, Chen J, McCormick F and Symons M. (1995b). McCormick F. (1995). Curr. Opin. Genet. Dev., 5, 51 ± 55. Proc. Natl. Acad. Sci. USA, 92, 11781 ± 11785. McGlade J, Brunkhorst B, Anderson D, Mbamalu G, Quilliam LA, Khosravi-Far R, Hu€ SY and Der CJ. (1995). Settleman J, Dedhar S, Rozakis-Adcock M, Chen LB BioEssays, 17, 395 ± 404. and Pawson T. (1993). EMBO J., 12, 3073 ± 3081. Rodriguez-Viciana P, Warne PH, Dhand R, Vanhaesebroeck Michaud NR, Fabian JR, Mathes KD and Morrison DK. B, Gout I, Fry MJ, Water®eld MD and Downward J. (1995). Mol. Cell. Biol., 15, 3390 ± 3397. (1994). Nature, 370, 527 ± 532. Milburn MV, Tong L, DeVos AM, Brunger A, Yamaizumi Rodriguez-Viciana P, Warne PH, Vanhaesebroeck B, Z, Nishimura S and Kim S-H. (1990). Science, 247, 939 ± Water®eld MD and Downward J. (1996). EMBO J., 15, 945. 2442 ± 2451. Minden A, Lin A, McMahon M, Lange-Carter C, Derijard Rodriguez-Viciana P, Warne PH, Khwaja A, Marte BM, B, Davis RJ, Johnson GL and Karin M. (1994). Science, Pappin D, Das P, Water®eld MD, Ridley A and 266, 1719 ± 1723. Downward J. (1997). Cell, 89, 457 ± 467. Mineo C, Anderson RGW and White MA. (1997). J. Biol. Rommel C, Radziwill G, Lovrie J, Noeldeke J, Heinicke T, Chem., 272, 10345 ± 10348. Jones D, Aitken A and Moelling K. (1996). Oncogene, 12, Mizuno T, Kaibuchi K, Yamamoto T, Kawamura M, 609 ± 619. Sakoda T, Fujioka H, Matsuura Y and Takai Y. (1991). Rommel C, Radziwill G, Moelling K and Hafen E. (1997). Proc. Natl. Acad. Sci. USA, 88, 6442 ± 6446. Mech. Dev., 64, 95 ± 104. Moodie SA, Willumsen BM, Weber MJ and Wolfman A. Roy S, Lane A, Yan J, McPherson R and Hancock JF. (1993). Science, 260, 1658 ± 1661. (1997). J. Biol. Chem., 272, 20139 ± 20145. Moodie SA, Paris M, Villafranca E, Kirshmeier P, Willumsen Rubin GM. (1991). TIG, 7, 372 ± 377. BM and Wolfman A. (1995). Oncogene, 11, 447 ± 454. Russell M, Lange-Carter CA and Johnson GL. (1995). J. Moodie SA and Wolfman A. (1994). Trends Genet., 10, 14 ± Biol. Chem., 270, 11757 ± 11760. 18. Saez R, Chan AM-L, Miki T and Aaronson SA. (1994). Morcos P, Thapar N, Tusneem N, Stacey D and Tamanoi F. Oncogene, 9, 2977 ± 2982. (1996). Mol. Cell. Biol., 16, 2496 ± 2503. Sato KY, Polakis PG, Haubruck H, Fasching CL, Morrison DK, Heidecker G, Rapp UR and Copeland TR. McCormick F and Stanbridge EJ. (1994). Cancer Res., (1993). J. Biol. Chem., 268, 17309 ± 17316. 54, 552 ± 559. Morrison DK. (1994). Science, 266, 56 ± 57. Satoh T and Kaziro Y. (1992). Cancer Biol., 3, 169 ± 177. Morrison DK and Cutler Jr RE. (1997). Curr. Opin. Cell Schaap D, van der Wal J, Howe LR, Marshall CJ and van Biol., in press, Blitterswijk WJ. (1993). J. Biol. Chem., 268, 20232 ± 20236. MoÈ sch H-U, Roberts RL and Fink GR. (1996). Proc. Natl. Sche€zek K, Lautwein A, Kabsch W, Ahmadian MR and Acad. Sci. USA, 93, 5352 ± 5356. Wittinghofer A. (1996). Nature, 384, 591 ± 596. Musacchio A, Cantley LC and Harrison SC. (1996). Proc. Sche€zek K, Ahmadian MR, Kabsch W, WiesmuÈ ller L, Natl. Acad. Sci. USA, 93, 14373 ± 14378. Lautwein A, Schmitz F and Wittinghofer A. (1997). Nassar N, Horn G, Herrmann C, Scherer A, McCormick F Science, 277, 333 ± 338. and Wittinghofer A. (1995). Nature, 375, 554 ± 560. Schlessinger J. (1993a). Trends Biochem. Sci., 18, 273 ± 275. Nishida E and Gotoh Y. (1993). TIBS, 18, Schlessinger J. (1993b). TIBS, 18, 273 ± 275. O'Gara MJ, Zhang XF, Baker L and Marshall MS. (1997). Settleman J, Narasimhan V, Foster LC and Weinberg RA. Biochem. Biophys. Res. Commu., 238, 425 ± 429. (1992). Cell, 69, 539 ± 549. Okazaki M, Kishida S, Murai H, Hinoi T and Kikuchi A. Shimizu K, Kuroda S, Yamamori B, Matsuda S, Kaibuchi (1996). Cancer Res., 56, 2387 ± 2392. K, Yamauchi T, Isobe T, Irie K, Matsumoto K and Takai Oldham SM, Clark GJ, Gangarosa LM, Co€ey Jr RJ and Y. (1994). J. Biol. Chem., 269, 22917 ± 22920. Der CJ. (1996). Proc.Natl.Acad.Sci.USA,93, 6924 ± Shirouzu M, Koide H, Fujita-Yoshigaki J, Oshio H, Toyama 6928. Y, Yamasaki K, Fuhrman SA, Villafranca E, Kaziro Y Olson MF, Ashworth A and Hall A. (1995). Science, 269, and Yokoyama S. (1994). Oncogene, 9, 2153 ± 2157. 1270 ± 1272. Sigal IS, Gibbs JB, D'Alonzo JS and Scolnick EM. (1986). Parker F, Maurier F, Delumeau I, Duchesne M, Faucher D, Proc. Natl. Acad. Sci. USA, 83, 4725 ± 4729. Debussche L, Dugue A, Schweigho€er F and Tocque B. Sklar MD, Thompson E, Welsh MJ, Liebert M, Harney J, (1996). Mol. Cell. Biol., 16, 2561 ± 2569. Grossman HB, Smith M and Prochownik EV. (1991). Parrini MC, Bernardi A and Parmeggiani A. (1996). EMBO Mol. Cell. Biol., 11, 3699 ± 3710. J., 15, 1107 ± 1111. Smith MR, DeGudicibus SJ and Stacey DW. (1986). Nature, Peterson SN, Trabalzini L, Brtva TR, Fischer T, Altschuler 320, 540 ± 543. DL, Martelli P, Lapetina EG, Der CJ and White II GC. Sozeri O, Vollmer K, Liyanage M, Frith D, Kour G, Mark (1996). J. Biol. Chem., 271, 29903 ± 29908. GE and Stabel S. (1992). Oncogene, 7, 2259 ± 2262. Pomerance M, Thang MN, Tocque B and Pierre M. (1996). Spaargaren M and Bischo€ JR. (1994). Proc. Natl. Acad. Sci. Mol. Cell. Biol., 16, 3179 ± 3186. USA, 91, 12609 ± 12613. Ponting CP and Benjamin DR. (1996). Trends Biochem. Sci., Sprenger F, Trosclair MM and Morrison DK. (1993). Mol. 21, 422 ± 425. Cell. Biol., 13, 1163 ± 1172. Increasing complexity of Ras signaling SL Campbell et al 1413 Stang S, Bottor€ D and Stone JC. (1997). Mol. Cell. Biol., 17, Vincenz C and Dixit V. (1996). J. Biol. Chem., 271, 20029 ± 3047 ± 3055. 20034. Stanton Jr VP, Nichols DW, Laudano AP and Cooper GM. Vogel US, Dixon RAF, Schaber MD, Diehl RE, Marshall (1989). Mol. Cell. Biol., 9, 639 ± 647. MS, Scolnick EM, Sigal IS and Gibbs JB. (1988). Nature, Sternberg PW and Horvitz HR. (1991). TIG, 7, 366 ± 370. 335, 90 ± 93. Stokoe D, Macdonald SG, Cadwallader K, Symons M and Vojtek AB, Hollenberg SM and Cooper JA. (1993). Cell, 74, Hancock JF. (1994). Science, 264, 1463 ± 1467. 205 ± 214. Stokoe D and McCormick F. (1997). EMBO J., 16, 2384 ± Vossler MR, Yao H, York RD, Pan M-G, Rim CS and Stork 2396. PJS. (1997). Cell, 89, 73 ± 82. SuB,JacintoE,HibiM,KallunkiT,KarinMandBen- Wang Y, Xu H-P, Riggs M, Rodgers L and Wigler, M. Neriah Y. (1994). Cell, 77, 727 ± 736. (1991). Mol. Cell. Biol., 11, 3554 ± 3563. Suen KL, Bustelo XR and Barbacid M. (1995). Oncogene, 11, Warne PH, Viciana PR and Downward J. (1993). Nature, 825 ± 831. 364, 352 ± 355. TakiT,HayashiY,TaniwakiM,SetoM,UedaR,Hanada Weissinger EM, Eissner G, Grammer C, Fackler S, Haefner R, Suzukawa K, Yokota J and Morishita K. (1996). B, Yoon LS, Lu KS, Bazarov A, Sedivy JM, Mischak H Oncogene, 13, 2121 ± 2130. and Kolch W. (1997). Mol. Cell. Biol., 17, 3229 ± 3241. Tamanoi F. (1988). Biochim. Biophys. Acta., 948, 1 ± 15. WestwickJK,CoxAD,DerCJ,CobbMH,HibiM,KarinM Tanabe S, Zeleznik-Le NJ, Kobayashi H, Vignon C, and Brenner DA. (1994). Proc. Natl. Acad. Sci. USA, 91, Espinosa RI, LeBeau MM, Thirman MJ and Rowley 6030 ± 6034. JD. (1996). Genes Chromosom. and Cancer, 15, 206 ± 216. White MA, Nicolette C, Minden A, Polverino A, Van Aelst Tesmer JJ, Berman DM, Gilman AG and Sprang SR. (1997). L, Karin M and Wigler MH. (1995). Cell, 80, 533 ± 541. Cell, 39, 251 ± 261. White MA, Vale T, Camonis JH, Schaefer E and Wigler MH. The I, Murthy AE, Hannigan GE, Jacoby LB, Menon AG, (1996). J. Biol. Chem., 271, 16439 ± 16442. Gusella JF and Bernards A. (1993). Nature Genet., 3, 62 ± Williams LT. (1992). Curr. Opin. Cell Biol., 2, 601 ± 603. 66. Winkler DG, Johnson JC, Cooper JA and Vojtek AB. (1997). The I, Hannigan GE, Cowley GS, Reginald S, Zhong Y, J. Biol. Chem., 272, 24402 ± 24409. Gusella JF, Hariharan IK and Bernards A. (1997). Wittinghofer A, Sche€zek K and Ahmadian MR. (1997). Science, 276, 791 ± 794. FEBS Lett., 410, 63 ± 67. Thompson JD, Higgins DG and Gibson TJ. (1994). Nucl. Wolthuis RM, Bauer B, van't Veer LJ, de Vries-Smits AM, Acids Res., 22, 4673 ± 4680. Cool RH, Spaargaren M, Wittinghofer A, Burgering BM Thorburn A, Throburn J, Chen S-Y, Powers S, Shubeita HE, and Bos JL. (1996). Oncogene, 13, 353 ± 362. Feramisco JR and Chien KR. (1993). J. Biol. Chem., 268, Wu J, Dent P, Jelinek T, Wolfman A, Weber MJ and Sturgill 2244 ± 2249. TW. (1993). Science, 262, 1065 ± 1069. Tocque B, Delumeau I, Parker F, Maurier F, Multon M-C Xia Z, Dickens M, Raingeaud J, Davis RJ and Greenberg and Schweigho€er F. (1997). Cell Signaling, in press, ME. (1995). Science, 270, 1326 ± 1331. Toda T, Uno I, Ishikawa T, Powers S, Kataoka T, Broek D, Yamagata K, Sanders LK, Kaufmann WE, Yee W, Barnes Cameron S, Broach J, Matsumoto K and Wigler M. CA, Nathans D and Worley PF. (1994). J. Biol. Chem., (1985). Cell, 40, 27 ± 36. 269, 16333 ± 16339. Tong L, de Vos AM, Milburn MV, Jancarik J, Noguchi S, Yamanashi Y and Baltimore D. (1997). Cell, 88, 205 ± 211. Nishimura S, Miura K, Ohtsuka E and Kim S-H. (1989). Yan M, Dai T, Deak JC, Kyriakis JM, Zon LI, Woodgett JR Nature, 337, 90 ± 93. and Templeton DJ. (1994). Nature, 372, 798. Traverse S, Cohen P, Paterson H, Marshall C, Rapp U and Yan Z, Deng X, Chen M, Ahram M, Sloane BF and Grand RA. (1993). Oncogene, 8, 3175 ± 3181. Friedman E. (1997). J. Biol. Chem., 272, 27902 ± 27907. Treisman R. (1996a). Curr. Opin. Cell Biol., 8, 205 ± 215. Yatani A, Okabe K, Halenbeck R, Polakis P, McCormick F Treisman R. (1996b). Semin. Cancer Biol., 1, 47 ± 58. and Brown AM. (1990). Cell, 60, 769 ± 776. Tu H, Barr M, Dong DL and Wigler M. (1997). Mol. Cell. Yoder-Hill J, Golubic M and Stacey DW. (1995). J. Biol. Biol., 17, 5876 ± 5887. Chem., 270, 27615 ± 27621. Ueda Y, Hirai S, Osada S, Suzuki A, Mizuno K and Ohno S. Yoshida Y, Kawata M, Miura Y, Musha T, Sasaki T, (1996). J. Biol. Chem., 271, 23512 ± 23519. Kikuchi A and Takai Y. (1992). Mol. Cell. Biol., 12, Urano T, Emkey R and Feig LA. (1996). EMBO J., 15, 810 ± 3407 ± 3414. 816. Zhang X, Settleman J, Kyriakis JM, Takeuchi-Suzuki E, Van Aelst L, Barr M, Marcus S, Polverino A and Wigler M. Elledge SJ, Marshall MS, Bruder JT, Rapp UR and (1993). Proc. Natl. Acad. Sci. USA, 90, 6213 ± 6217. Avruch J. (1993). Nature, 364, 308 ± 313. Van Aelst L, White MA and Wigler MH. (1994). Cold Spring ZhangZ,VuoriK,WangHG,ReedJCandRuoslahtiE. Harbor Symp. Quant. Biol., 59, 181 ± 186. (1996). Cell, 85, 61 ± 69. Vanhaesebroeck B, Welham MJ, Kotani K, Stein R, Warne Zimmermann S, Rommel C, Ziogas A, Lovric J, Moelling K PH, Zvelebil MJ, Higashi K, Volinia S, Downward J and and Radziwill G. (1997). Oncogene, 15, 1503 ± 1511. Water®eld MD. (1997). Proc. Natl. Acad. Sci. USA, 94, 4330 ± 4335. VerheijM,BoseR,LinXH,YaoB,JarvisWD,GrantS, Birrer MJ, Szabo E, Zon LI, Kyriakis JM, Haimovitz- Friedman A, Fuks Z and Kolesnick RN. (1996). Nature, 380, 75 ± 79.