Polyisoprenylation of Ras in Vitro by a Farnesyl-Protein Transferase”

Farnesylation of Ras occurs in vivo on a Cys residue in the C-terminal sequence -Cys-Val-Leu-Ser (termed a CAAX box). This modification is required for Ras membrane localization and cell transforming activity. Using [3H]farnesyl-PPi as precursor and Escherichia coli-expressed Ras, forms of Ras having the CAAX sequence were radiolabeled upon incubation with the cytosolic fraction of bovine brain. Forms of Ras having a deletion of the CAAX sequence or a Cys to Ser substitution in this sequence were not substrates. Radioactivity incorporated into Ras by bovine brain cytosol was released by treatment with iodomethane but not with methanolic KOH indicating a thioether linkage. High pressure liquid chromatography analysis of the cleavage products on a C-18 column showed a major peak of radioactivity that co-eluted with a farnesol standard. The enzyme responsible for Ras farnesylation in bovine brain was approximately 190 kDa as estimated by gel filtration and required a divalent cation for activity. Nonradioactive farnesyl-PPi, geranylgeranyl-PPi, and Ras peptides having the C-terminal sequence -Cys-Val-Leu-Ser competed in the assay with IC50 values of 0.7, 1.4, and 1-3 microM, respectively. Farnesol and Ras peptides having the sequence -Ser-Val-Leu-Ser were not inhibitory. These results identify a farnesyl-protein transferase activity that may be responsible for the polyisoprenylation of Ras in intact cells.


Farnesylation
of Ras occurs in viva on a Cys residue in the C-terminal sequence -Cys-Val-Leu-Ser (termed a CAAX box). This modification is required for Ras membrane localization and cell transforming activity. Using [3H]farnesyl-PPi as precursor and Escherichiu co&expressed Ras, forms of Ras having the CAAX sequence were radiolabled upon incubation with the cytosolic fraction of bovine brain. Forms of Ras having a deletion of the CAAX sequence or a Cys to Ser substitution in this sequence were not substrates. Radioactivity incorporated into Ras by bovine brain cytosol was released by treatment with iodomethane but not with methanolic KOH indicating a thioether linkage. High pressure liquid chromatography analysis of the cleavage products on a C-18 column showed a major peak of radioactivity that co-eluted with a farnesol standard.
The enzyme responsible for Ras farnesylation in bovine brain was Cpproximately 190 kDa as estimated by gel filtration and required a divalent cation for activity.
Farnesol and Ras peptides having the sequence -Ser-Val-Leu-Ser were not inhibitory. These results identify a farnesyl-protein transferase activity that may be responsible for the polyisoprenylation of Ras in intact cells.
The ras oncogene encodes a 21-kDa GTP-binding protein (Ras) that must be localized in the plasma membrane in order to transform cells (1). Ras may undergo at least five posttranslational modifications prior to membrane localization (2)(3)(4)(5)(6)(7)(8). Most of these modifications occur at the Ras C terminus which has a consensus motif -Cys-Aaa-Aaa-Xaa (Aaa, aliphatic amino acid, Xaa, any amino acid). This sequence is required for membrane localization of Ras and is referred to as a CAAX box' (9).
All Ras proteins are modified by the isoprenoid farnesyl at the C-terminal Cys via a thioether linkage (3)(4)(5). Mutational studies have shown that the Cys of the CAAX box is required for all of the C-terminal processing steps suggesting that polyisoprenylation is the first step (3,9). Other proteins have been identified that are modified by farnesyl or geranylgeranyl moieties (10-13).
It has been speculated that a putative farnesyl-protein transferase using farnesyl-PP, In the Tris-buffered assays, the reactions were incubated for 1 h at 24 ". Reaction velocities in the Hepes-buffered assays were at least 4 times greater, and incubations were for 10 min at 24 "C.
Ras proteins used as substrates were expressed in E. coli and purified (17,18). Ras proteins having different CAAX box sequences were constructed with oligonucieotide linkers as described (17 titatively extracted into a Triton X-114 detergent phase as has been described previously for Ras modified in viuo (2, 3); Ras lacking incorporated radiolabel remained in the aqueous phase (data not shown). This result indicated that the in vitro modification conferred hydrophobic properties to Ras. The activity or activities responsible for radiolabel incorporation into Ras remained in the reticulocyte cytosol fraction upon centrifugation at 100,000 x g. Polyisoprenylation of Ras using ["HIFPP could also be detected in 100,000 X g supernatant fractions prepared from rat brain, heart, kidney, liver, lung, skeletal muscle, spleen, and testis; the highest activity was observed using the supernatant prepared from brain. Activity was not detected in the membrane fraction. Incorporation of ["HIFPP into Ras was also detected using a bovine brain cytosolic fraction (Fig. 2) A product identification was performed as described under "Materials and Methods" using Ras protein radiolabeled by ["HIFPP and bovine brain cytosol. Recovery of radioisotope following treatment with iodomethane was 65%. Recovery of ["HIFPP standard in a parallel reaction lacking Ras was 61% indicating that the 35-40% loss occurred during the extraction procedures. The reaction products were analyzed by HPLC using a C-18 column as shown in Fig. 3 The observation that FPP served as precursor for polyisoprenylation of Ras and the co-elution of the major iodomethane cleavage product with farnesol indicates that a farnesyl-protein transferase activity is present in the cytosolic fraction of rabbit reticulocytes, bovine brain, and various rat tissues. Equivalent farnesylation was observed using Ras complexed with either guanosine-5'-0- (3-thiotriphosphate) or guanosine-5'-0-(2-thiodiphoshate). As shown in Fig. 4, the farnesyl-protein transferase activity eluted from a Superose-12 gel filtration column as a single peak having an apparent size of 190 kDa. A smaller activity peak having a size of 75 kDa was observed when the protease inhibitors leupeptin and antipain were not included in the chromatography buffer.
To determine specificity of the farnesyl transferase activity, several isoprenoids were tested as competitors of the farnesylation reaction (Table I). In assays having 0.5 NM ["HIFPP and 1 pM Ras, nonradioactive FPP and geranylgeranyl-PPi potently inhibited the assay. We do not know whether Ras is polyisoprenylated when geranylgeranyl-PPi is the precursor; however, Ras is not significantly modified by geranylgeranyl in uiuo (5) or in vitro (this work) when ["Hlmevalonate is used as a precursor of polyisoprene biosynthesis. Geranyl-PPi and dimethylallyl-PPi were less potent as competitors.
Farnesol and geranylgeraniol were inactive up to 100 PM indicating that the pyrophosphate moiety is required for competition. Several peptides having the Harvey Ras CAAX sequence were also tested ( Table I). The smallest peptide, CVLS, was nearly as potent as the larger peptides indicating that the critical interactions between Ras and the farnesyl-protein transferase are confined to the 4-residue CAAX box region. The sulfhydryl moiety is apparently essential for competition because peptides having a Cys to Ser substitution were inactive. Peptides lacking the CAAX sequence, free Cys, S-(trans,trans)farnesyl-Cys, and unrelated peptides having a single Cys residue were not competitors when tested up to 100 pM (not shown).
Initial experiments with peptide SSGCVLS indicate that it is a substrate for farnesylation. The predicted product of the smallest active peptide, S-(tram, trans)farnesyl-CVLS, was approximately 13-fold less potent than peptide alone (Table I). CONCLUSIONS The results in this study have identified an activity in the cytosolic fraction of mammalian cells which utilizes FPP to attach a farnesyl moiety onto Ras. The substrate specificity and peptide competition experiments indicate that the 4residue CAAX box is the minimal sequence efficiently recognized by the enzyme. Although FPP was 2-fold more potent than geranylgeranyl-PPi in competition experiments, it is possible that the farnesyl-protein transferase activity may also be involved with the modification of proteins with geranylgeranyl.
If this were true, we speculate that different sequences within the CAAX box (13) influence whether a protein is a substrate for farnesylation or geranylgeranylation. Future studies with pure enzyme will help to resolve the specificity of the activity. We presume that the farnesylprotein transferase detected in cell extracts is responsible for