key: cord-0777254-che3iwu3 authors: Hart, Kristen C; Donoghue, Daniel J title: Derivatives of activated H-ras lacking C-terminal lipid modifications retain transforming ability if targeted to the correct subcellular location date: 1997-12-18 journal: Oncogene DOI: 10.1038/sj.onc.1200908 sha: ebb33b727c09a42ae8af8c72db56b4658073d3ab doc_id: 777254 cord_uid: che3iwu3 To examine the ability of ras to activate signal transduction pathways in the absence of lipid modifications, fusion proteins were constructed that target ras(WT) or activated ras(61L) to cellular membranes as integral membrane proteins, using the first transmembrane domain of the E1 protein of avian infectious bronchitis virus (IBV), which contains a cis-Golgi targeting signal. Golgi-targeted derivatives of activated ras were completely inactive in transformation assays. However, when examined in focus formation assays, transformation of NIH3T3 cells were seen with derivatives of ras(61L) containing a mutated E1 targeting sequence that results in plasma membrane localization. Removal of the lipid modification sites in and upstream of the CAAX motif did not abrogate the transforming activity of plasma membrane-localized ras(61L) derivatives, indicating that these lipid modifications are not essential for ras activity, as long as the protein is correctly localized to the plasma membrane. Interestingly, the activity of integral membrane versions of ras(61L) was strictly dependent on a minimum distance between the transmembrane domain anchor region and the coding sequence of ras. Derivatives with only a 3-amino acid linker were inactive, while linkers of either 11- or 22-amino acids were sufficient to restore transforming activity. These results demonstrate that: (1) activated ras targeted to Golgi membranes is unable to cause transformation; (2) lipid modifications at the C-terminus are not required for the transforming activity of plasma membrane-anchored ras(61L) derivatives, and serve primarily a targeting function; (3) a transmembrane domain can effectively substitute for C-terminal modifications that would normally target ras to the inner surface of the plasma membrane, indicating that ras(61L) does not need to reversibly dissociate from the membrane as might be allowed by the normal lipidation; and (4) in order to function properly, there exists a critical distance that the ras protein must reside from the plasma membrane. Introduction ras proteins are critical modulators of signaling pathways involved in cell growth and dierentiation. p21ras is a member of a large family of small proteins that act as molecular`switches', cycling between an active GTP-bound state and an inactive GDP-bound state . ras is activated in response to a wide variety of stimuli, including platelet-derived growth factor (PDGF). Binding of PDGF to its receptors induces their dimerization and transphosphorylation, creating binding sites for eector molecules (reviewed in Hart et al., 1995) , some of which regulate or activate ras. Activation of ras by growth factors recruits Raf-1 to the plasma membrane (Stokoe et al., 1994; Leevers et al., 1994) , leading to Raf-1 activation and subsequent initiation of the MAPK cascade. ras also has been shown to activate Rac and Rho proteins, which control signaling pathways that lead to cytoskeletal modi®cations Khosravi-Far et al., 1995; Qiu et al., 1995a,b) . Activation of this pathway has recently been shown to be crucial for full transformation by ras (Prendergast et al., 1995; Khosravi-Far et al., 1995; Qiu et al., 1995a,b) . Thus, combined activation of the Rac/Rho and MAPK pathways may constitute the entire intracellular response necessary for transformation by activated ras. There are multiple p21 ras proteins, including N-ras, K-ras, and H-ras (reviewed in Santos and Nebreda, 1989 ; Lowy and Willumsen, 1993) , which are localized to the inner surface of the plasma membrane (Willingham et al., 1980; Sefton et al., 1982) through speci®c sequences in the C-terminus (Willumsen et al., 1984) . The mechanism by which ras proteins are targeted to the plasma membrane has been the subject of intense research. Since ras is frequently mutated in tumors, interference with ras localization could become an important means of treating some forms of cancer. Post-translational modi®cations of ras proteins occur at the C-terminus, which includes a domain known as the`CAAX' box, where A is an aliphatic amino acid, and X represents any amino acid. Processing of this region consists of farnesylation of the conserved Cys residue four amino acids from the C-terminus (Cys186 in H-ras and N-ras), cleavage of the last three amino acids of the protein, and carboxymethylation of the new C-terminal Cys Jackson et al., 1990; Casey et al., 1989; Gutierrez et al., 1989; Clarke et al., 1988; Lowy and Willumsen, 1986; Shih and Weeks, 1984; Willumsen et al., 1984; Sefton et al., 1982) . N-ras and Hras have cysteines N-terminal of the CAAX motif, which become palmitoylated after the CAAX processing (Buss and Sefton, 1986) . In the case of K-ras-4B, the C-terminus contains a polybasic domain in addition to the CAAX motif which contributes to plasma membrane localization (Hancock et al., 1990 (Hancock et al., , 1991 Jackson et al., 1994) . The palmitoylation reactions are reversible, allowing for potentially dynamic localization of ras proteins (Magee et al., 1987) . Although truncation of the C-terminus does not aect the inherent structure or biochemical properties of ras , mutations in these motifs render ras proteins cytosolic and inactive, implying that lipid modi®cations are important for ras function. There is recent evidence suggesting that the lipid modi®cations of ras are also necessary for activation of Raf-1 and B-Raf (Okada et al., 1996; Kikuchi and Williams, 1994) . However, since these lipid modifications are also responsible for targeting ras to the plasma membrane, it is unclear from these recent reports whether the lipid modi®cations per se are a necessity, or just localization of ras to the plasma membrane. The palmitoylation reactions that occur in the C-terminus of ras proteins have been shown to be reversible (Magee et al., 1987) . There is also some speculation that ras may require reversible association with membranes, and the lipid modi®cations in the Cterminus provide the potential for this dynamic localization. However, the experimental methods used to date are unable to conclusively distinguish the precise role of lipid modi®cations in the membrane anchoring and activation of ras. In order to examine these questions, we have used a novel method of targeting ras to the plasma membrane by creating fusion proteins using the transmembrane targeting signal from the E1 protein of avian IBV placed at the N-terminus (Swift and Machamer, 1991) . In its wild-type form, this transmembrane domain is capable of targeting heterologous proteins to cis-Golgi membranes. When the point mutation Gln37?Ile is present, proteins are instead targeted to the plasma membrane. These fusions allow us to address the question of whether ras must be able to reversibly associate with the plasma membrane for biological function as assayed by transformation and GTPase activity. To examine whether C-terminal modi®cations of ras are required for functions other than targeting, we created similar fusion proteins that contain mutations which abolish processing of the CAAX box. Results presented here demonstrate that ras 61L can transform ®broblasts if targeted to the plasma membrane by an N-terminal transmembrane domain, yet is not active when localized to Golgi membranes. We also show that integral membrane versions of ras 61L require a minimum distance from the membrane in order to activate signal transduction pathways leading to transformation. This work describes the ®rst derivatives of ras 61L that lack any lipid modi®cations, and demonstrate that Cterminal lipid modi®cations are not required for transformation by ras 61L . To examine the possibility that activated H-ras can transform if targeted to membranes by an N-terminal transmembrane anchor, a variety of fusion proteins were designed. As illustrated in the left side of Figure 1 , ras WT and ras 61L proteins are localized at the inner surface of the plasma membrane by means of lipid modi®cations at the C-terminus. The right side of Figure 1 depicts the orientation and design of the transmembrane-anchored ras derivatives. For derivatives containing the E1 cis-Golgi targeting signal, the transmembrane domain anchors the fusion proteins at Golgi membranes as Type I integral membrane proteins, with the ras-derived portion of the protein still in the cytosol. These derivatives allowed us to examine whether ras could function if localized to any membranous environment, thereby assessing the importance of the interaction between ras and the plasma membrane. Fusion proteins were also designed with a mutated E1(QI) targeting signal and are expected to be transported to the plasma membrane, with the Nterminus outside the cell, and the ras portion of the proteins on the inner surface of the plasma membrane. Figure 2 represents all of the fusion proteins designed for this study. The feasibility of targeting ras to the plasma membrane by the N-terminus was previously demonstrated in work by Buss et al. (1989) and Lacal et al. (1988) , which examined the eect of Nterminal myristylation on ras 61L activity. Buss et al. (1989) described a naturally-occurring myristylated derivative of ras, p29Gag-ras, containing a 59-amino linker region between the myristylated Gly and the ras coding sequence. Lacal and colleagues generated Nterminally anchored ras derivatives by attaching the 15amino acid myristylation signal from pp60c-src. Our derivatives dier in that ras is anchored permanently to membranes as an integral membrane protein. The myristylation targeting signal used in previous studies has the potential to allow reversible association of ras with the plasma membrane. This is an important distinction, which allows us to examine whether ras needs to be able to come o of the membrane in order to function. By exploiting the E1 localization signal, we were also able to target ras speci®cally to dierent membranes in the cell. We were also able to examine in more detail the minimum distance from the membrane that ras requires for activity, by designing constructs with linker regions of either 3, 11 or 22 amino acids. Additional fusion proteins were designed to examine the general requirement for lipid modi®cations for ras activity. ras derivatives with 3-or 11-amino acid linkers Figure 1 Schematic of ras fusion proteins. All fusion proteins have either the normal C-terminus (amino acids 181 ± 189) of ras WT or ras 61L (CMSCKCVLS), or contain SAAX (CMSCKSVLS) or 3S (SMSSKSVLS) mutations, which alter or abolish the normal post-translational processing and lipid modi®cations of the proteins. The N-terminus of each E1 derivative is oriented either into the lumen of the Golgi, or is extracellular if the point mutation Q?I is present. The whole of the ras protein is always cytosolic. The transmembrane domain (designated by the shaded box) is either the cis-Golgi targeting signal contained in the ®rst transmembrane domain of the IBV E1 protein, or a mutated version (Q37?I) that targets the protein to the plasma membrane. The linker region between the transmembrane domain and the ras protein denoted by the black box is either 3-, 11-, or 22-amino acids in length ras anchored as an integral membrane protein KC Hart and DJ Donoghue were constructed in the context of the unmutated (CAAX) C-terminus, as well as the SAAX and 3S Cterminal mutations.`CAAX' refers to the wild-type Cterminal sequences of ras.`SAAX' indicates a mutation of Cys186 in the CAAX motif to Ser, which abolishes the farnesylation, cleavage, and carboxymethylation reactions.`3S' refers to mutation of Cys181, 184, and 186 to Ser, which in addition destroys the palmitoylation sites . The 22-amino acid linker derivatives were only made in the context of the SAAX mutant C-terminus. Since mutation of all three C-terminal Cys residues results in a ras protein completely unable to be modi®ed by lipid moieties, these mutants allowed us to assay whether these lipid modi®cations are necessary for activity, or just utilized in a membrane targeting capacity. ras derivatives are unable to cause transformation if targeted to Golgi membranes When tested in focus forming assays, Golgi-targeted derivatives of ras WT with 3-, 11-, or 22-amino acid Figure 2 Localization and transforming activity of ras derivatives. Bars depict the ras WT and ras 61L constructs tested in this paper. The following abbreviations are used to describe the dierent derivatives.`61L' refers to mutation of codon 61 of ras from Gln to Leu, which oncogenically activates the protein (Sekiya et al., 1984) .`E1' refers to the ®rst transmembrane domain (amino acids 1 ± 45) of the E1 protein of avian IBV, which constitutes a cis-Golgi targeting signal that can target heterologous proteins to the Golgi (Swift and Machamer, 1991) .`E1(QI)' indicates a point mutation in the transmembrane domain (Gln37?Ile) that abrogates its Golgi-targeting function, targeting proteins instead to the plasma membrane (Swift and Machamer, 1991) . Shaded boxes indicate the E1 or E1(QI) transmembrane domain; black boxes indicate the linker region; striped boxes indicate the C-termini of the clones which is either wild-type (CAAX) or mutant (SAAX or 3S). Localization was determined by indirect immuno¯uorescence, as described in the legend for Figure 4 . Transforming activity of the clones was analyzed in a focus forming assay, where NIH3T3 cells were transfected with the plasmids encoding the fusion proteins, and examined for foci after approximately 12 ± 14 days. Each experiment was repeated at least three times for transforming derivatives, and at least two times for non-transforming constructs. Results are expressed as an average of all experiments performed with each given construct. 7=0 ± 5% of ras 61L ; ++=20 ± 25% of ras 61L ; +++=50 ± 60% of ras 61L ; ++++=80 ± 100% of ras 61L transforming activity linker regions were all inactive. This was expected, since ras WT did not induce focus formation in our assays. The Golgi-targeted derivatives of ras 61L were also unable to cause transformation. This was irrespective of whether the linker region was 3-, 11-, or 22-amino acids in length, or whether the C-terminal post-translational modi®cation signals were intact or not. Figure 4E and I demonstrate localization of E1-11ras 61L -SAAX and E1-22-ras 61L -SAAX, respectively, to a perinuclear region typical of Golgi staining. Panel K shows the localization of E1-11-ras 61L -CAAX, and Figure 4L shows that this fusion protein colocalizes with mAb 10E6 (see arrows), which speci®cally reacts with a cis-Golgi epitope, and has been used as a marker for the early Golgi (Hart et al., 1994; Wood et al., 1991) . This con®rms that the Golgi-targeted fusion proteins are localized correctly, and therefore we conclude that ras 61L cannot initiate transforming signal transduction pathways from Golgi membranes, indicating that attachment to any lipid-®lled environment is not sucient for ras to be active. ras 61L derivatives with 11-or 22-amino acid linkers can transform if anchored to the plasma membrane, irrespective of C-terminal lipid modi®cations The plasma membrane-targeted derivatives of ras 61L with the 3-amino acid linkers (E1(QI)-3-ras 61L -CAAX, E1(QI)-3-ras 61L -SAAX, and E1(QI)-3-ras 61L -3S) were all inactive in focus formation assays. However, derivatives with 11-or 22-amino acid linker regions were very active in transformation of NIH3T3 cells. The results of these assays are summarized in Figure 2 , which indicates the focus forming activity of each ras derivative as an average of at least three independent experiments. E1(QI)-11-ras 61L -CAAX, E1(QI)-11-ras 61L -SAAX, E1(QI)-11-ras 61L -3S and E1(QI)-22-ras 61L - Focus forming assays of ras derivatives targeted to the plasma membrane as integral membrane proteins. Plasmids encoding ras fusion proteins were transfected into NIH3T3 cells, and foci were counted after 12 ± 14 days. Cells were ®xed with methanol, stained with Giemsa, and photographed. ras anchored as an integral membrane protein KC Hart and DJ Donoghue SAAX exhibited signi®cant transforming activity in comparison with ras 61L (See Figure 3E ± I). It is of interest to note that mutations in the C-terminus of ras 61L that abolish the lipidation of the proteins do not aect the ability of activated ras to transform NIH3T3 cells. Indeed, the derivative with an intact C-terminus was consistently three-to four-fold less ecient at transformation than derivatives with mutated Ctermini (See Figure 2 ; also compare Figure 3E , F and G). This could re¯ect conformational strains conferred on the protein due to anchoring both termini to the membrane. Plasma membrane-targeted derivatives containing 3-, 11-, or 22-amino acid linkers were also examined by immuno¯uorescence. As seen in Figure 4 , derivatives that were positive in transformation assays localize to the plasma membrane (F, G, H, and J), and exhibit staining patterns comparable to that of ras 61L (B). Also, derivatives with 3-amino acid linkers were still targeted to the plasma membrane (D), although unable to transform cells. Thus, derivatives of ras 61L targeted to the plasma membrane via an N-terminal transmembrane domain are able to cause transformation, provided that there is a¯exible linker region of sucient length between the transmembrane anchor and the N-terminus of ras. These results also clearly demonstrate that ras 61L does not require lipid modi®cations in order to transform cells, and suggests that the C-terminal processing of ras proteins primarily functions in a membraneanchoring capacity. In order to verify that the fusion proteins being expressed were of the correct size, proteins were immunoprecipitated from transiently transfected cells ( Figure 5) . ras proteins typically run as a doublet on SDS ± PAGE gels, one band representing the unprocessed form of ras, the other indicating the mature, processed form of the protein (see arrows) (Shih et al., , 1982 . The SAAX and 3S mutations prevent the C-terminal cleavage step and subsequent lipid mod-i®cations from occurring, leaving only a single band (lanes 3 and 4, arrows) . The subtleties of these processing steps are dicult to detect in the remaining fusion proteins shown in Figure 5 , due to their increased molecular weight. However, it is quite apparent that all fusion proteins do express the additional N-terminal transmembrane domain sequences (lanes 5 ± 12), and these targeting domains are not cleaved o during maturation of the proteins. The fusion proteins migrate slower than Figure 4 Indirect immuno¯uorescence con®rms localization of ras derivatives. NIH3T3 cells were transiently transfected with various ras constructs, ®xed, permeabilized, and incubated with rat mAb Ab-2 (clone Y13-238) and¯uorescein-conjugated goat arat secondary antibody. For (L) mAb 10E6 was used to detect the cis-Golgi, and visualized with rhodamine-conjugated goat amouse secondary antibody. (A) mock; (B) ras 61L -CAAX; (C) E1-3-ras WT -3S; (D) E1(QI)-3-ras 61L -3S; (E) E1-11-ras 61L -SAAX; (F) E1(QI)-11-ras 61L -CAAX; (G) E1(QI)-11-ras 61L -SAAX; (H) E1(QI)-11-ras 61L -3S; (I) E1-22-ras 61L -SAAX; (J) E1(QI)-22-ras 61L -SAAX; (K) E1-11-ras 61L -CAAX -a-ras antibody (¯uorescein); (L) E1-11-ras 61L -CAAX -mAb 10E6 a-cis-Golgi antibody (rhodamine). Arrows indicate CAAX -mAb in K and L expected, which may be due to the presence of an Nlinked glycosylation site in the extreme N-terminus of the E1-derived sequence (Machamer et al., 1990) . The transforming ability of some of the plasma membrane-targeted derivatives of ras 61L indicates that their function was not adversely aected by addition of an N-terminal transmembrane domain. To analyse whether fusion proteins that were non-transforming were still able to function in some manner, GTPase activity was examined. ras 61L remains locked in the GTP-bound state longer due to a decreased rate of GTP hydrolysis (Temeles et al., 1985) , despite a 50-fold increased binding anity for rasGAP (Krengel et al., 1990) . If the N-terminal transmembrane domain does not interfere with normal protein function, one would expect to see a higher GTPase activity exhibited by ras WT fusion proteins, as compared to ras 61L derivatives. Using thin-layer chromatography, this dierence in GTPase activities is apparent, as shown by a higher proportion of labeled GDP versus GTP in ras WT samples as compared to ras 61L samples ( Figure 6, lanes 1 and 2) . Interestingly, ras 61L with the 3S mutation in the Cterminus, which renders the protein cytosolic and inactive in transformation, does not aect the GTPase activity (lane 3). Derivatives of ras WT with the 3-amino acid linker hydrolyze GTP, irrespective of whether they are targeted to the plasma membrane (lane 5) or Golgi membranes (lane 4). ras 61L derivatives with transmembrane anchors and 3-amino acid linkers exhibit impaired GTPase activity as expected (data not shown). Finally, the 11-and 22-amino acid derivatives of ras WT (lanes 6 and 8) or ras 61L (lanes 7 and 9) exhibit the expected GTPase activity pro®les, regardless of the transmembrane domain attached at the N-terminus or the modi®cations present at the C-terminus. These results demonstrate that ras proteins can be targeted to the plasma membrane via a transmembrane anchor without aecting the expected GTPase activity of the proteins. Also, we ®nd that C-terminal lipid modifications do not play a role in maintaining the intrinsic GTPase activity of wild-type or activated ras. The experiments described above demonstrate that ras 61L can activate transforming signal transduction pathways when targeted by the N-terminus to the plasma membrane as an integral membrane protein. This activity was completely independent of the normal post-translational processing that occurs at the C-terminus of ras proteins. Since ras 61L derivatives (E1(QI)-11-ras 61L -SAAX, E1(QI)-11-ras 61L -3S and E1(QI)-22-ras 61L -SAAX) lacking the C-terminal sequences required for normal processing of ras are still able to cause transformation, it is clear that lipid modi®cations per se are not required for activity of oncogenic ras. This is the ®rst demonstration of nonlipidated transforming derivatives of ras. Our data suggests that post-translational modification of H-ras serves primarily to target the protein to the plasma membrane. Whether there are other, more subtle functions of these C-terminal lipid modi®cations Note that all ras WT derivatives have a higher GTPase activity (lanes 1, 4, 5, 6 and 8) compared to ras 61L derivatives (lanes 2, 3, 7 and 9), as evidenced by increased amount of labeled GDP versus GTP in ras WT samples. This result is irrespective of localization to Golgi or plasma membranes, and regardless of C-terminal modi®cations. Lane 1: ras WT -CAAX; Lane 2: ras 61L -CAAX; Lane 3: ras 61L -3S; Lane 4: E1-3-ras WT -3S; Lane 5: E1(QI)-3-ras WT -SAAX; Lane 6: E1-11-ras WT -SAAX; Lane 7: E1(QI)-11-ras 61L -3S; Lane 8: E1-22-ras WT -SAAX; Lane 9: E1(QI)-22-ras 61L -SAAX ras anchored as an integral membrane protein KC Hart and DJ Donoghue distinct from their more obvious targeting function remains to be demonstrated. The observation that three of the transmembrane-anchored derivatives, E1(QI)-11-ras 61L -SAAX, E1(QI)-11-ras 61L -3S, and E1(QI)-22-ras 61L -SAAX, retain transforming activity despite their inability to be lipid-modi®ed strongly implies that downstream signaling molecules such as Raf-1 and the MAPK cascade are activated in these cells. This contradicts recent reports that posttranslational modi®cations of ras are required for activation of Raf-1 and B-Raf (Okada et al., 1996; Kikuchi and Williams, 1994 ). The precise activation state of eectors downstream from ras, and whether these molecules are activated in a similar fashion by integral membrane versions of ras, is currently under investigation. Transforming activity of transmembrane derivatives of ras demonstrates that ras does not require transient or reversible association with the membrane If the lipid modi®cations on ras proteins serve only a targeting function, as indicated by our results, this raises the question of why ras proteins evolved to contain C-terminal lipid modi®cations instead of a more permanent transmembrane anchor. Small GTPbinding proteins related to ras are now known to function in a variety of capacities within the cell, such as mediating vesicular transport in the process of endocytosis, sorting and tracking through the secretory pathway, and regulating the structure of the cytoskeleton (reviewed in Hall, 1994; Ferro-Novick and Novick, 1993; Pfeer, 1994; Zerial and Stenmark, 1993) . Some of these processes may require the regulatory GTP-binding proteins to be transiently or reversibly associated with lipid bilayers, and only lipid modi®cations would allow this dynamic association to occur. Our results suggest, in the case of ras, that a reversible association with the plasma membrane is not required for activation of transforming signal transduction pathways, since integral membrane versions of ras 61L are transforming. It would be interesting to examine the eects of anchoring other small GTP-binding proteins more permanently to speci®c membranes or compartments using targeting sequences such as the transmembrane domain from the E1 protein. ras proteins require a minimum spacing from the plasma membrane in order to function Transforming activity of ras 61L derivatives was dependent upon a minimum distance from the plasma membrane mediated by the length of the linker region. Altering the distance between ras and the plasma membrane does aect the ability of ras to activate signal transduction pathways, as demonstrated by the inactivity of 3-amino acid linker derivatives of ras 61L and activation of these derivatives in transformation assays by addition of 11-or 22-amino acid linker regions. The crystal structure of c-H-ras bound to GDP (residues 1 ± 171) or GTP (residues 1 ± 166) indicates that the extreme C-terminus forms an alpha-helical region that juts out from the globular catalytic region of ras (De Vos et al., 1988; Pai et al., 1989) , suggesting that the catalytic domain is well-removed from the membrane in normal ras proteins. One can imagine that the native C-terminus of ras constitutes a natural`linker' region, not unlike the arti®cial linker regions incorporated into our ras fusion proteins. This would explain the failure of the short 3amino acid linker at the N-terminus to allow for a functional fusion protein, whereas linkers of 11-or 22amino acids, corresponding more closely in length to the C-terminus of normal ras, do result in a transforming ras 61L derivative. Conformational energy analysis performed on the 18 C-terminal amino acids excluded from the crystal structure of the GDP-bound form of ras indicates that this region likely forms a helix ± turn ± helix or helical hairpin' motif, allowing the N-terminus and C-terminus of ras to be in close proximity (Brandt-Rauf et al., 1990) . Perhaps these two regions of the protein interact in some manner. However, it appears from our results that interactions between the Nterminus and lipid modi®cations at the C-terminus are not important for ras function. There could be one or more eectors of ras that bind to a site dierent from the identi®ed eector domain, and access to this site may be dependent upon orientation of the protein with respect to the membrane, or may require binding to sites at both the N-and Cterminus of ras proteins. Identi®cation of other membrane-associated factors involved in regulation or activation of ras proteins and their eectors will provide clues as to why ras proteins evolved their unique structural elements. Clearly the intrinsic GTPase activities of our fusion proteins are not altered when compared to normal versions of ras WT and ras 61L . GTP binding and hydrolysis are also apparently unaected by mutation of the CAAX box. Thus, it appears that neither the N-nor the C-terminus are important for this function of ras. Speci®c amino acids in the eector regions of ras (amino acids 32 ± 38 or Switch I, and amino acids 60 ± 76 or Switch II) are involved in association with guanine nucleotides and also in binding to regulators of GTPase and nucleotide exchange activities. Since GTPase activity is normal in the fusion proteins, the overall structure of the catalytic regions is not expected to be drastically altered. To our knowledge, this is the ®rst study examining the role of C-terminal lipid modi®cations in regulating GTPase activity of ras. One discrepancy between our results and those of Buss et al. (1989) is that anchoring of wild-type ras to the plasma membrane by the E1(QI) transmembrane domain does not result in transformation. Buss et al. (1989) found signi®cant focus forming ability of wildtype ras after addition of N-terminal myristylation. They speculated that perhaps their protein was binding to inappropriate cellular targets, which may not be promoted by a transmembrane anchor. It is also possible that interaction with nucleotide exchange factors was enhanced in myristylated ras WT derivatives, but this does not occur in E1(QI)-ras WT constructs. To further clarify the role of membrane association in regulating the inherent enzymatic activity of ras proteins, it would be interesting to see if there are any subtle eects on the association of integral membrane versions of ras proteins with regulators of GTPase and nucleotide exchange activities. The observation that Golgi-targeted derivatives of ras are unable to transform ®broblasts, while perhaps not unexpected, provides some insight into the requirements for signal transduction through the ras pathway. It is clear that some mediators of the transforming pathway initiated by oncogenic ras are only available at the plasma membrane. Precisely what these mediators are remains to be elucidated. However, one can conclude that simple juxtaposition of ras to a lipidrich environment is not sucient to allow it to signal. Examination of these Golgi-targeted ras derivatives and their ability to interact with or activate the normal cytosolic substrates such as Raf-1 will provide future insights into the function of ras. Recent evidence demonstrates that other important pathways regulating cytoskeletal structure and organization are also activated by ras. Rac1 regulates membrane ruing in response to growth factor stimulation , while RhoA functions in the regulation of actin stress ®ber and focal adhesion assembly in growth factor-stimulated cells ). It appears that activation of both the MAPK and Rac/Rho pathways is required for full transformation by ras (Prendergast et al., 1995; Khosravi-Far et al., 1995; Qiu et al., 1995a,b) . In light of this recent evidence, we are currently investigating activation of the Rac/Rho pathway in response to our integral membrane versions of ras. Full-length Harvey p21ras WT and p21ras 61L cloned into pcDNAI at the HindIII-EcoRI sites were generously provided by J Buss. Restriction sites in the coding sequences of ras WT and ras 61L used to make these clones are as follows: BsaHI at nt 28 and FspI at nt 215 were used in generation of clones with E1 or E1(QI) transmembrane anchors; A¯III at nt 470 was utilized to generate mutations in the C-terminus. Note that the mutation resulting in the Gln?Leu at codon 61 occurs at nt 182, and consists of a A?T transversion. All constructs containing N-terminal transmembrane domains and/or C-terminal mutations described in this paper were constructed using the following strategy. Pairs of complementary oligonucleotides were designed and synthesized such that, when annealed, overhangs for restriction sites were formed. More details about the making of the various fusion constructs are available upon request. All synthetic oligonucleotides were gel puri®ed as previously described (Xu et al., 1993) , and all DNA sequences derived from oligonucleotides were con®rmed by dideoxy nucleotide sequencing before use. NIH3T3 cells were maintained as previously described (Hart et al., 1994) . Cells were split at a density of 2610 5 cells per 60 mm plate and transfected the following day using the calcium phosphate precipitation protocol (Chen and Okayama, 1987) . Cells from each 60 mm plate were split 1:12 2 days later, and scored for foci 12 ± 14 days later. These assays were repeated at least 26 for each construct. NIH3T3 cells were split 1610 5 onto 60 mm plates containing glass coverslips and transfected the following day with 10 mg of plasmid DNA, as described above. Two days after transfection, the cells were ®xed with 3% paraformaldehyde/PBS and permeabilized with 0.5% Triton X-100/PBS. The intracellular location of ras fusion proteins was detected with rat monoclonal antiserum Ab-2 (Y13-238) directed against v-H-ras (Oncogene Science) and uorescein-conjugated goat a-rat secondary antibody (Boehringer Mannheim). For double-labeling experiments, cells were ®xed and permeabilized as described above. ras fusion proteins were detected as described above, then the coverslips were treated with monoclonal Ab 10E6, which detects the cis-Golgi of cells (kindly provided by WJ Brown and V Malhotra) and rhodamine-conjugated goat a-mouse secondary antibody (Boehringer Mannheim). NIH3T3 cells were split 2610 5 onto 60 mm plates and 2 days after transfection, monolayers were washed 26 with Tris-saline, incubated 15 min with DME minus Cys and Met, and then labeled for 2 h with 100 mCi each of [ 35 S]-Cys and [ 35 S]-Met. Cells were lysed in RIPA (1% Triton X-100, 0.15 M NaCl, 50 mM Tris-HCl pH 7.5, 0.1% SDS, 1% DOC, 10 mg/ml aprotinin), precleared with Protein A-Sepharose, and incubated with Ab-2 rat monoclonal antibody. Immunoprecipitates were collected with Protein A-Sepharose beads coated with rabbit a-rat IgG, washed 46 with RIPA, and resuspended in 26 sample buer (50 mM Tris pH 6.8, 2% SDS, 20% 2-mercaptoethanol, 10% glycerol). Proteins were separated by 15% SDS ± PAGE and detected by¯uorography. NIH3T3 cells were transfected as described for immunoprecipitation. Two days after transfection, monolayers were lysed with lysis buer (20 mM Tris-HCl pH 7.5, 0.125 M NaCl, 1 mM MgCl 2 , 1 mM CaCl 2 , 1% Triton X-100, 10 mg/ml aprotinin) and precleared with Protein A-Sepharose. One third of the total lysate was then subjected to immunoprecipitation using Ab-2. Immune complexes were collected on Protein A-Sepharose beads coated with rabbit a-rat IgG, washed 26 with lysis buer, 26 with RIPA, then incubated for 30 min on ice with 100 ml of 1.0610 77 M a-[ 32 P]GTP in RIPA. Beads were then washed 36 with RIPA, 16 with lysis buer, and incubated at 378C for 1 h in 100 ml of lysis buer. 2 ml of each sample was spotted onto PEI-cellulose plates (JT Baker, Phillipsburg, New Jersey) and chromatographed in 0.75 M KH 2 PO 4 , pH 3.4. Abbreviations E1, avian coronavirus E1 glycoprotein; ER, endoplasmic reticulum; IBV, infectious bronchitis virus. Melanie Webster and Patricia d'Avis for critical reading of the manuscript, and Laura Castrejon for editorial assistance. KCH would like to thank Scott Robertson, Ryan Dellinger, and Vincent Ollendor for advice and support. KCH gratefully acknowledges support from the Lucille P Markey Charitable Trust. This work was supported by grant 3RT-0242 from the U.C. Tobacco Related Disease Research Program and by grant CB-163 from the American Cancer Society. Proc. Natl. Acad. Sci. USA Proc. Natl. Acad. Sci. USA Proc. Natl. Acad. Sci. USA Proc. Natl. Acad. Sci. USA, 91 Proc. Natl. Acad. Sci. USA Proc. Natl. Acad. Sci. USA, 92 Proc. Natl. Acad. Sci. USA Cancer Investigation Rab GTPases in vesicular transport We thank Jan Buss for providing the p21ras WT and p21ras 61L plasmids that served as the starting point for construction of all clones described in this paper, and for helpful advice on GTPase assays. We thank WJ Brown and V Malhotra for the mAb 10E6 used for co-localization studies. We also thank April Meyer for technical support, ras anchored as an integral membrane protein KC Hart and DJ Donoghue