Cloning and Characterization of a Mammalian Prenyl Protein-specific Protease*

Proteins containing C-terminal “CAAX” sequence motifs undergo three sequential post-translational processing steps: modification of the cysteine with either a 15-carbon farnesyl or 20-carbon geranylgeranyl isoprenyl lipid, proteolysis of the C-terminal -AAX tripeptide, and methylation of the carboxyl group of the now C-terminal prenylcysteine. A putative prenyl protein protease in yeast, designated Rce1p, was recently identified. In this study, a portion of a putative human homologue of RCE1 (hRCE1) was identified in a human expressed sequence tag data base, and the corresponding cDNA was cloned. Expression of hRCE1 was detected in all tissues examined. Both yeast and human RCE1 proteins were produced in Sf9 insect cells by infection with a recombinant baculovirus; membrane preparations derived from the infected Sf9 cells exhibited a high level of prenyl protease activity. Recombinant hRCE1 so produced recognized both farnesylated and geranylgeranylated proteins as substrates, including farnesyl-Ki-Ras, farnesyl-N-Ras, farnesyl-Ha-Ras, and the farnesylated heterotrimeric G protein Gγ1 subunit, as well as geranylgeranyl-Ki-Ras and geranylgeranyl-Rap1b. The protease activity of hRCE1 activity was specific for prenylated proteins, because unprenylated peptides did not compete for enzyme activity. hRCE1 activity was also exquisitely sensitive to a prenyl peptide analogue that had been previously described as a potent inhibitor of the prenyl protease activity in mammalian tissues. These data indicate that both the yeast and the human RCE1 gene products are bona fide prenyl protein proteases and suggest that they play a major role in the processing of CAAX-type prenylated proteins.

A variety of proteins are modified with an isoprenoid lipid at a cysteine that is initially four residues from the C terminus (1)(2)(3). Such proteins contain the so-called CAAX motif, in which the "C" is the modified cysteine, the "A" residues are most commonly (but not always) aliphatic amino acids, and the "X" residue can be one of several amino acids. The X residue determines whether the protein is modified by the 15-carbon farnesyl lipid or the 20-carbon geranylgeranyl. If the X residue is a leucine, the protein will be geranylgeranylated; several other residues (e.g. Met, Ser, Ala, and Gln) direct farnesylation. Following prenylation of the protein, two additional processing steps occur (4,5). First, a specific protease cleaves the -AAX tripeptide from the protein, leaving the prenylated cysteine as the new C terminus. The carboxyl group of the prenylcysteine is then methylated by a specific methyltransferase.
It is well established that protein prenylation plays a vital role in the membrane localization and function of most prenylated proteins (6). The role that the proteolysis and methylation of prenyl proteins play in their function, however, is not as well understood. Studies on peptides have demonstrated that each processing step significantly increases the affinity of farnesylated peptides for membranes (7), although the effect of the final two steps is not as great for geranylgeranylated peptides. Proteolysis and methylation also increased the hydrophobicity of Ras proteins processed in an in vitro system (8). Prevention of the proteolysis of Ras in cells resulted in a decrease in membrane localization (9,10), and in yeast it resulted in at least a partial loss of Ras function (10).
The enzymes responsible for prenylation of CAAX-containing proteins, protein farnesyltransferase and protein geranylgeranyltransferase I, have been cloned and studied in detail (3). Additionally, a specific prenyl protein carboxymethyltransferase has been identified in yeast as the product of the STE14 gene (11), and a human homologue of the STE14 gene product has recently been described (12). These findings have left the prenyl protein protease as the only member of the processing pathway yet to be identified on a molecular level.
Recently, an elegant genetic screen in yeast resulted in the identification of two candidate genes for prenyl protein proteases (10). The first gene, AFC1/STE24, appeared to be primarily involved in the processing of the precursor to a-factor, a farnesylated yeast mating pheromone. AFC1/STE24 catalyzes two cleavage events on the a-factor precursor, the first being the C-terminal proteolysis and the second being a cleavage occurring near the N terminus of the peptide (10,13,14). Strong evidence was provided that the second candidate gene, RCE1, was involved in the processing of the yeast Ras proteins, in addition to the C-terminal processing of the a-factor precursor (10,15). Because RCE1 was linked to the processing of a prenyl protein, it seemed likely that the RCE1 gene product might have a broader range of substrates than AFC1/STE24. We report here the identification, expression, and preliminary characterization of a human homologue of RCE1, termed hRCE1. Evidence is provided that hRCE1 is in fact a prenyl protein protease and that it is involved in the processing of a variety of CAAX-type prenylated proteins, including all known forms of Ras. These findings open the door for molecular studies of this protease and will facilitate studies aimed at determining the roles of the proteolysis and methylation steps in the functions of CAAX-type prenyl proteins.

EXPERIMENTAL PROCEDURES
Materials-The gt11 human umbilical vein endothelial cell library (HUVEC) 1 (16) was provided by John York of this institution. The plasmid pRS315-RCE1 containing the cDNA for yeast RCE1 (10) was a gift from Jasper Rine (University of California, San Francisco); the pATH-STE14 plasmid containing the cDNA encoding the yeast methyltransferase STE14 (11) was a gift from Susan Michaelis (Johns Hopkins Medical Center); the QE31-N-Ras expression plasmid for N-Ras (17) was a gift from Robert Bishop (Schering-Plow Research Institute); the pTrcHis-Rap1b expression plasmid for hexahistidine-tagged Rap1b (18) was a gift from Guy James (University of Texas Health Sciences Center, San Antonio). The reduced farnesyl-peptide analogue RPI (19,20) was a gift from Robert Rando (Harvard Medical School). Oligonucleotide primers were synthesized at the Duke University DNA Core Facility. The Bac-2-Bac Baculovirus Expression System, the GeneTrapper cDNA Positive Selection System, and the pCMV.SPORT 2 human fetal brain cDNA library were purchased from Life Technologies, Inc. S-Adenosylmethionine (AdoMet) was purchased from Research Biochemicals International. [ 3 H-methyl]-S-adenosylmethionine ([ 3 H]Ado-Met) was purchased from New England Nuclear. Peptides were synhesized by Princeton Biomolecules.
Cloning of hRCE1-The deduced amino acid sequence of yeast RCE1 was used to conduct a text-based search of the human expressed sequence tag (EST) data base at the National Center for Biotechnology Information. The clone W96411 was identified as a likely homologue. Five additional clones that overlapped W96411 were present in the data base, representing 725 nucleotides of the cDNA sequence. The primer 5Ј-GGGCTTCAGGCTGGAGGGCATTTT-3Ј was chosen from this sequence for use in a GeneTrapper cDNA positive-selection cloning protocol. A partial clone containing the hRCE1 open reading frame but lacking an initiation codon was cloned from a pCMV.SPORT 2 human fetal brain cDNA plasmid library. A probe was generated from the XcmI-PstI fragment of that clone and used to screen a gt11 HUVEC library (16). That screen yielded a clone containing the likely initiation codon, because it contained an upstream stop codon in-frame with the coding sequence. The 5Ј end of the HUVEC cDNA containing the initiation codon was removed as a BssHII-XcmI fragment and ligated to the human fetal brain cDNA XcmI-PstI fragment to generate a hRCE1 cDNA containing the entire open reading frame. This BssHII-PstI fragment was subcloned into the vector pFASTBAC-1 to produce p-FAST-BAC-1-hRCE1. In a second construct, designated pFASTBAC-1-⌬hRCE1, the 5Ј-untranslated region of hRCE1 was replaced with a portion of the 5Ј-untranslated region from the baculovirus polyhedron gene (CCTATAAAT), and the codons specifying the first 22 amino acids of hRCE1 were removed (see Fig. 1).
Northern Blot Analysis-A 32 P-labeled probe was generated with random hexamer priming from the 1100-bp BssHII-PstI hRCE1 cDNA fragment and hybridized to a human multi-tissue poly(A) ϩ RNA blot (obtained from CLONTECH); hybridization and washing were performed as described previously (21). The process was repeated using a probe made to the cDNA of ␤-actin as a control. Expression of RCE1 was also examined using a mouse multiple tissue Northern blot (CLON-TECH). The blot was probed with a 223-bp Rce1 cDNA fragment that was amplified from a mouse liver cDNA library (CLONTECH) with oligonucleotide primers 5Ј-TTGCCCTTGTGACCTGACAGATGG-3Ј and 5Ј-GAGTGGGCAGGTGAACACAGCAGG-3Ј.
Production of Recombinant Proteins in Sf9 Cells-Recombinant baculoviruses were produced with the pFASTBAC-1 vector and the protocol provided in the Bac-2-Bac Baculovirus Expression System kit. Recombinant baculoviruses were prepared for the two human RCE1 constructs described above, for yeast RCE1, and for the yeast prenyl protein carboxymethyltransferase STE14.
For production of recombinant proteins, Sf9 cells in log phase growth were diluted to 1 ϫ 10 6 cells/ml and infected with recombinant baculoviruses at multiplicities of infection ranging from 2 to 10. Cells producing yeast or human RCE1 were harvested 72 h post-infection and resuspended in 50 mM Tris-HCl, pH 7.7. Cells expressing yeast STE14 were harvested 60 h post-infection, and resuspended in 5 mM NaHPO 4 , pH 7.0, containing 5 mM EDTA (11) and a mixture of protease inhibitors (22). In all cases, cells were disrupted by sonication, nuclei and debris were removed by centrifugation at 500 ϫ g for 5 min, and membranes were then pelleted by centrifugation at 200,000 ϫ g for 1.5 h. Membranes from RCE1 producing cells were resuspended in 50 mM Tris-HCl, whereas membranes from cells producing STE14 were resuspended in 5 mM NaHPO 4 containing 5 mM EDTA and the protease inhibitor mixture (22). Final protein concentrations of the membrane suspensions were 10 -25 mg/ml. The suspensions were flash-frozen in liquid nitrogen and stored at Ϫ80°C in multiple aliquots.
Farnesyl-Ki-Ras and geranylgeranyl-Ki-Ras were resolved from the unprenylated precursor and modifying enzyme by applying the prenylation reactions to a heptyl-Sepharose column, which was washed with buffer B (50 mM Hepes, 3 mM MgCl 2 , 1 mM dithiothreitol, 5 M GDP) containing 0.1% lubrol (removing unprenylated Ki-Ras and the prenyltransferase), and then eluting the purified prenyl-Ki-Ras with buffer B containing either 2% sodium cholate (farnesyl-Ki-Ras) or 4% sodium cholate (geranylgeranyl-Ki-Ras). The eluted prenyl-Ki-Ras preparations were diluted 20-fold in buffer B and applied to an S-Sepharose column. The column was washed with buffer B, and the prenyl-Ki-Ras was eluted with buffer B containing 0.1% octylglucoside and 750 mM NaCl. Purified prenyl-Ki-Ras stocks were flash-frozen in liquid nitrogen and stored at Ϫ80°C in multiple aliquots. The peptides CVIM and GSPCVLM were prenylated chemically as described previously (26) and purified by high pressure liquid chromatography.
Prenyl Protein Protease Assays-Protease reactions were initiated by the addition of membranes containing the recombinant prenyl protein protease to an assay mixture containing prenylated proteins in 100 mM Hepes, pH 7.4, and 5 mM MgCl 2 in a total volume of 50 l; reactions were conducted at 37°C. Following the proteolysis reaction, methylation of proteolysed prenylated protein was initiated by addition of 20 g of membranes containing STE14 to the reaction in 17.5 M [ 3 H]AdoMet (2000 Ci/mmol), 5 mM NaHPO 4 , pH 7.0, 62.5 mM EDTA, 1 mM phenylmethylsulfonyl fluoride, 10 mM 1,10-phenanthroline, and 300 M Ntosyl-L-phenylalanine chlorophenyl ketone in a total volume of 20 l. The addition of EDTA and protease inhibitors served both to quench the proteolysis reaction and to inhibit proteases present in the STE14 membranes. Stoichiometric methylation was achieved within 20 min at 37°C. Reactions were terminated by adding 0.5 ml 4% SDS along with 50 g of bovine brain cytosol as carrier protein (27), and 3 H-methylated protein was quantitated using a filter assay (23,25,27).
Protease assays were also performed directly on prenylated proteins following their prenylation. In these assays, a 45-l prenylation reaction was performed as described above, and the concentration of prenylated protein generated in the reaction was determined. The prenylated protein was diluted to the desired concentration in buffer A, and a source of membranes was added to the mixture. The assay then proceeded as described above.

RESULTS AND DISCUSSION
The deduced amino acid sequence of yeast RCE1 was used to search a human EST data base, and the entry W96411 was identified as a potential homologue. Five additional overlapping ESTs were identified giving a composite sequence of 725 amino acids. A full-length hRCE1 clone was assembled from a clone isolated from a human fetal brain cDNA library, which contained the 3Ј end of the cDNA, and a clone isolated from a HUVEC cDNA library, which contained the 5Ј end of the cDNA. The transcript was 1500 bp, not including polyadenylation, and had an open reading frame that encoded a 329-amino acid protein (Fig. 1). The strongest homology between the yeast and human proteins was found in a region from Arg 166 to Pro 270 of hRCE1 (42% identity and 68% similarity), suggesting that the active site of the enzyme could reside in this region. This region was also highly conserved in a potential homologue of RCE1 found in the Caenorhabditis elegans genome data base (CEF48F5). As reported for yeast RCE1 (10), the deduced amino acid sequence of hRCE1 contained multiple predicted transmembrane domains (not shown). Examination of hRCE1 expression by Northern blotting revealed a primary band of 1.8 kilobases, which was expressed in all tissues tested ( Fig. 2A). A faint 2.4-kilobase band was also evident in some tissues. Examination of RCE1 expression in mouse tissues, using a probe corresponding to a sequence found in the mouse data base, gave stronger evidence of the presence of larger transcripts, which may represent alternatively spliced variants of RCE1 (Fig. 2B).
High level expression of the yeast carboxymethyltransferase STE14 in Sf9 cells has allowed development of a coupled assay for prenyl protein proteolysis that utilizes proteins prenylated in vitro as substrates and quantitates C-terminal proteolysis by stoichiometric methylation. In preliminary experiments, we found that expression of the cDNA encoding yeast RCE1 in Sf9 cells by infection with a recombinant baculovirus resulted in a   3. Yeast and human RCE1 gene products possess prenyl protein protease activity. Analysis of prenyl protein protease activity utilizing a coupled proteolysis/methylation assay was performed as described under "Experimental Procedures." Protease assays were conducted with increasing amounts of membranes prepared from uninfected Sf9 cells (q) or from Sf9 cells infected with recombinant baculoviruses containing the cDNA for either yeast (f) or human RCE1 (ࡗ). Purified farnesyl-Ki-Ras (2 M) was used as substrate, and reactions were conducted for 15 min at 37°C.

FIG. 4. Characterization of hRCE1 activity.
A, kinetics of hRCE1 activity for prenyl-Ki-Ras. Steady-state saturation kinetics for hRCE1 were analyzed for purified farnesyl-Ki-Ras and geranylgeranyl-Ki-Ras. hRCE1 membrane protein (15 ng) was incubated with increasing levels of farnesyl-Ki-Ras (q) and geranylgeranyl-Ki-Ras (f) for 10 min at 37°C. B, screening of prenyl proteins as substrates for hRCE1. Ki-Ras, Ha-Ras, N-Ras, G ␥1 , and Rap1b were prenylated and used directly at 1 M in protease assays. f-, farnesyl; g-, geranylgeranyl. hRCE1 membrane protein (50 ng) was added, and reactions were allowed to proceed for 15 min at 37°C. C, competition of hRCE1 activity by prenylated peptides. Prenylated peptides (farnesyl-CVIM (q) and farnesyl-GSPCVLM (f)) or unprenylated peptides (CVIM (E) and GSPCVLM (Ⅺ)) were included in the standard protease reaction mixture with 0.5 M purified farnesyl-Ki-Ras. The reaction was initiated by adding 15 ng of hRCE1 membrane protein and allowed to proceed for 15 min at 37°C. Activity is expressed as the percentage of that observed for an untreated control. D, inhibition of hRCE1 activity by the farnesyl-peptide analogue RPI. RPI compound was diluted in 10% Me 2 SO and added to the standard protease reaction mixture containing 0.5 M purified farnesyl Ki-Ras. Final Me 2 SO concentration in the reaction was 2%. Reactions were initiated by addition of 15 ng of hRCE1 membrane protein, and reactions were allowed to proceed for 15 min at 37°C. Activity is expressed as the percentage of that observed for an untreated control. dramatic increase (Ͼ1000-fold) in prenyl protein protease activity in the cells (Fig. 3); subfractionation of cell extracts revealed that all of the activity was found in the membrane fraction (data not shown). It was this demonstration that the yeast RCE1 gene encoded an authentic prenyl protein protease that compelled us to search for and ultimately clone hRCE1. Although attempts to express full-length cDNA encoding hRCE1 in insect cells were unsuccessful, when a construct in which the first 22 codons of the hRCE1 were deleted (designated ⌬hRCE1) was expressed in Sf9 cells, a dramatic (Ͼ1000fold) increase in prenyl protein protease activity was observed in the membrane fraction (Fig. 3).
Recombinant hRCE1 produced by expression of the ⌬hRCE1 construct utilized both farnesylated and geranylgeranylated Ki-Ras as substrates, with both substrates exhibiting apparent K m values of approximately 0.5 M (Fig. 4A) and similar k cat values, indicating that the enzyme has similar catalytic efficiencies for farnesylated and geranylgeranylated substrates. Farnesyl-Ha-Ras, farnesyl-N-Ras, farnesyl-G ␥1 , and geranylgeranyl-Rap1b were also found to be substrates for hRCE1 (Fig. 4B), demonstrating that the enzyme can utilize a broad range of CAAX-type prenyl protein substrates. Competition studies were performed with peptides corresponding to the C termini of Ki-Ras (CVIM) and of mouse N-Ras (GSPCVLM) to determine whether the enzyme could recognize unprenylated as well as prenylated substrates. GSPCVLM and CVIM that contained farnesylated cysteine residues were able to compete for the processing of prenylated Ki-Ras, whereas the corresponding unprenylated peptides could not, demonstrating both that hRCE1 is specific for prenylated proteins and that it recognizes short prenylated peptides (Fig. 4C). Additionally, the activity of a previously identified inhibitor of prenyl protease activity, a reduced farnesyl-peptide analogue termed RPI (19,20), was examined. The RPI compound was indeed an extremely effective inhibitor of hRCE1, exhibiting an IC 50 of approximately 5 nM (Fig. 4D).
The first report of prenyl protein protease activity indicated that proteolysis (and methylation) occurs in a membrane compartment in cells (8). Subsequent in vitro studies have focused on membrane fractions in analysis of this processing step. Two distinct activities have been characterized. The first activity was tightly bound by membranes (28 -31) and could only be released by detergents (20,32,33). The second activity was loosely associated with membranes and could be solubilized by a freeze-thaw process (34). For the purpose of comparison, these activities have been described as microsomal and soluble (35). The microsomal enzyme exhibits an apparently higher affinity for the prenyl peptide CVIM (K m of 0.65 M) (20) than the soluble enzyme (K m of 32 M) (34). Based on competitiontype assays, the microsomal enzyme exhibited a 250-fold specificity for prenylated peptides over unprenylated peptides, whereas the soluble enzyme reportedly had only a 5-fold specificity (35). Additionally, only the microsomal enzyme was inhibited by the RPI compound (19,20,35). The activity described herein for hRCE1 is most similar to the abovementioned microsomal activity, based on its apparent K m for farnesyl-Ki-Ras (ϳ0.5 M), its specificity for prenylated proteins, and the potent inhibition observed with the RPI compound.
The variety of substrates that hRCE1 processed in vitro suggests that it plays a major role in the processing of prenylated proteins in cells. Indeed, in an independent study, fibroblasts prepared from mouse embryos in which the RCE1 gene was disrupted failed to process the Ras proteins, as well as other prenylated proteins (36). Identification of the hRCE1 cDNA and the ability to produce substantial quantities of enzyme in Sf9 cells will provide several major advantages for its study. First, the high level expression of hRCE1 will allow detailed examination of its activity, including screening for specific inhibitors, with minimal concerns of background. Additionally, the high level production system will provide a platform for the purification of the enzyme and will allow initiation of structural approaches to its study. Finally, in combination with the overexpression of yeast carboxymethyltransferase, the tools are now in place to generate large amounts of processing intermediates of prenyl proteins, which should be quite useful in examining the properties that each of the processing steps import to the functions of prenylated proteins.