Identification of a Ras Palmitoyltransferase in Saccharomyces cerevisiae *

Most Ras proteins are posttranslationally modified by a palmitoyl lipid moiety through a thioester linkage. However, the mechanism by which this occurs is not known. Here, evidence is presented that the Ras2 protein of Saccharomyces cerevisiae is palmitoylated by a Ras protein acyltransferase (Ras PAT) encoded by the ERF2 and ERF4 genes. Erf2p is a 41-kDa protein localized to the membrane of the endoplasmic reticulum and contains a conserved DHHC cysteine-rich domain (DHHC-CRD). Erf2p co-purifies with Erf4p (26 kDa) when it is expressed in yeast or in Escherichia coli. The Erf2p/Erf4p complex is required for Ras PAT activity, and mutations within conserved residues (Cys189, His201, and Cys203) of the Erf2p DHHC-CRD domain abolish Ras PAT activity. Furthermore, a palmitoyl-Erf2p intermediate is detected suggesting that Erf2p is directly involved in palmitate transfer.ERF2 and ERF4 are the first genes identified that encode a palmitoyltransferase for a Ras GTPase.

Most Ras proteins are posttranslationally modified by a palmitoyl lipid moiety through a thioester linkage. However, the mechanism by which this occurs is not known. Here, evidence is presented that the Ras2 protein of Saccharomyces cerevisiae is palmitoylated by a Ras protein acyltransferase (Ras PAT) encoded by the ERF2 and ERF4 genes. Erf2p is a 41-kDa protein localized to the membrane of the endoplasmic reticulum and contains a conserved DHHC cysteine-rich domain (DHHC-CRD). Erf2p co-purifies with Erf4p (26 kDa) when it is expressed in yeast or in Escherichia coli. The Erf2p/Erf4p complex is required for Ras PAT activity, and mutations within conserved residues (Cys 189 , His 201 , and Cys 203 ) of the Erf2p DHHC-CRD domain abolish Ras PAT activity. Furthermore, a palmitoyl-Erf2p intermediate is detected suggesting that Erf2p is directly involved in palmitate transfer. ERF2 and ERF4 are the first genes identified that encode a palmitoyltransferase for a Ras GTPase.
Dozens of cellular and viral proteins are posttranslationally modified with palmitate or other long-chain fatty acids through a reversible thioester linkage (1,2). Examples are cell surface receptors, viral glycoproteins, and signal transducers including Ras, heterotrimeric G proteins, and nonreceptor tyrosine kinases. Palmitoylation is almost exclusively a property of membrane proteins and can occur on intracellular membranes early in the secretory pathway or at the plasma membrane. Although this modification was first described over 30 years ago, the molecular mechanism of palmitate addition has not been elucidated and has been a matter of controversy. A palmitoyltransferase activity assayed using mammalian H-Ras as a substrate was purified and identified as thiolase A, an enzyme required for fatty acid ␤-oxidation (3,4). The localization of this enzyme in peroxisomes makes it an unlikely candidate for a physiological regulator of Ras palmitoylation. Palmitoyltransferase activities assayed using viral glycoproteins, the nonreceptor tyrosine kinase p59 fyn , or G-protein heterotrimer as substrates have been detergent-solubilized, but the instability of the activity has hampered purification and molecular identification (5)(6)(7). A candidate palmitoyltransferase for Drosoph-ila hedgehog was recently identified as skinny hedgehog/sightless (5,6). Hedgehog is modified by cholesterol at the C terminus and palmitoylated through an atypical cysteine amide linkage at the N terminus. The failure to identify a palmitoyltransferase that acylates through a conventional thioester linkage, coupled with the observation that spontaneous and efficient transfer of fatty acid from acyl-CoA to proteins occurs in vitro, has lead to the suggestion that proteins autoacylate in vivo (7)(8)(9).
Plasma membrane localization of Ras requires farnesylation of the CaaX box cysteine via a thioether linkage, -aaX proteolysis, and carboxylmethylation (see reviews in Refs. 10,11). With the exception of K-Ras-4b, human Ras proteins are also palmitoylated on one or more neighboring cysteines via a thioester linkage. Palmitoylation is required for efficient plasma membrane localization and transforming activity of oncogenic forms of Ras (12). Previously, we described palmitoylationdependent alleles of yeast RAS2 and a genetic screen designed to identify mutations that render cells inviable if Ras2p is not palmitoylated (13,14). Mutations in two genes, ERF2 and ERF4/SHR5, were identified that resulted in diminished palmitoylation of Ras2p and mislocalization of GFP-Ras2p (14,15). Erf2p is a 41-kDa integral membrane protein localized at the ER, 1 which contains a conserved Asp-His-His-Cys cysteinerich domain (DHHC-CRD) between residues 164 -228. The DHHC-CRD domain, also referred to as the NEW1 or zf-DHHC domain (PF01529) is found in a large family of membrane proteins ranging from unicellular eukaryotes to humans (16,17). Genes encoding DHHC-CRD proteins in yeast include ERF2, AKR1, AKR2, PSL10, YOL003c, YNL326c, and YDR459c. Erf4p is a 26-kDa peripheral ER membrane protein.
The palmitoylation defect observed in erf2⌬ and erf4⌬ strains could affect palmitoylation either directly or indirectly by perturbing Ras2p trafficking and thereby preventing efficient interaction with a Ras palmitoyltransferase. Until now, it has not been possible to distinguish between these possibilities. In this study, we utilize an in vitro palmitoylation assay to demonstrate that Erf2p and Erf4p constitute a protein acyltransferase (Ras PAT) responsible for palmitoylation of yeast Ras2p.
Expression Plasmid Construction and Purification of Protein Substrates-All GST fusions were constructed from the galactose-inducible vector pEG(KG) (18). GST-Ras2CCaaX represents the complete Ras2 protein (38 kDa) fused to GST. GST-(HV)CCaaX consists of the C-terminal 35 amino acid residues of Ras2p fused to GST, and GST-(HV)ACaaX is the same construct except that Cys 318 has been mutated to Ala. GST-CCaaX consists of the C-terminal 5 amino acid residues of Ras2p fused to GST. Expression and purification of GST fusion proteins in yeast were performed as follows. Strain RJY543 was co-transformed with the indicated GST-Ras2p vector and pMA210, a plasmid that expresses GAL4 under the control of the ADH1 promoter (19). The culture was grown to an A 600 of 0.4 -0.6 in synthetic media containing 2% ethanol,2% glycerol as the carbon source. The cells were induced by the addition of galactose (4% final concentration) and incubated overnight at 25°C. Cells were centrifuged (750 ϫ g) for 10 min and lysed in a solution of Y-PER ® (Pierce) containing 1 mM DTT, 1 mM EDTA, 0.1% Triton X-100, and 0.13 mM phenylmethylsulfonyl at room temperature (30 -40 min). The yeast lysate was centrifuged at 2,000 rpm (750 ϫ g) (10 min), and GSH-agarose (Pierce) beads were added to the supernatant. The GSH-agarose beads were eluted (1 h, room temperature) with 20 mM glutathione in 50 mM Tris-HCl (pH 7.4), 0.02% Triton X-100, 10% glycerol. As expected, GST-Ras2CCaaX fusions were membrane-associated due to the prenylation of the CaaX box cysteine. Analysis of the Ras substrate by SDS-PAGE revealed that the GST-Ras protein is the major band present in the preparation. The minor binds cross-react with anti-GST antibody and are either free GST or a proteolytic product of the GST-Ras fusion (data not shown).
H-Ras was expressed as an N-terminal His 6 -tagged fusion in Sf9 cells and purified from detergent extracts of membranes using nickel chelate affinity chromatography (20). The stoichiometry of palmitate on the purified substrate has not been determined, but a significant fraction is probably lost during purification due to the action of thioesterases and the presence of reducing agents in the buffers. Myristoylated G i␣1 was purified from bacteria expressing N-myristoyl transferase (21). G i␣1 was reconstituted with nonprenylated G ␤␥ subunits purified from Sf9 cells in the assay (22).
Partial Purification of Erf2p/Erf4p Complex-RJY543 yeast cells expressing GST-Erf4p along with FLAG-Erf2p or FLAG-Erf2p mutants (Erf2(H201A)p, Erf2(C189S)p, Erf2(C203S)p) were solubilized in Y-PER ® reagent as described above. E. coli cells expressing GST-Erf4p along with FLAG-Erf2p or FLAG-Erf2p mutants (Erf2(H201A)p, Erf2(C203S)p) were lysed by high pressure (25,000 p.s.i.) homogenization (Avestin) in 50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 1 mM DTT, 1 mM EDTA, and 0.13 mM phenylmethylsulfonyl fluoride (Buffer A). Extracts were centrifuged at 6000 rpm (2800 ϫ g) for 15 min to remove unbroken cells and debris. Triton X-100 (0.3% final) was added to the supernatants, and samples were incubated at room temp for 20 min. The cell lysates were then diluted with equal volumes of Buffer A to a final concentration of 0.15% Triton X-100. GST-Erf4p and FLAG-Erf2p (wild type and mutants) fusion proteins were affinity-purified from yeast and bacterial cell extracts by binding to GSH-agarose or anti-FLAG M2 antibody agarose beads (Sigma). The beads were washed three times with 50 mM Tris-HCl (pH 7.4). The volume of the wash each time was at least 10 times the bead volume. The beads were taken up in 50 mM Tris-HCl (pH 7.4), 0.02% Triton X-100, and 10% glycerol. The ratio of GST-Erf4p to FLAG-Erf2p partially purified from either yeast or bacteria depended on whether GSH-agarose or M2 antibody-agarose beads were used (an example is shown in Fig. 2).
Immunoblotting-Proteins bound to the GSH-agarose beads were analyzed for the presence of both GST and FLAG peptide fusion proteins. GST fusion proteins were detected by immunoblotting with rabbit anti-GST (Molecular Probes) antibody and peroxidase-conjugated goat anti-rabbit secondary antibody (Sigma). FLAG-tagged proteins were detected on Western blots by probing with anti-FLAG ® M5 monoclonal antibody (Sigma) and peroxidase-conjugated sheep anti-mouse secondary antibody (Amersham Biosciences). The presence of FLAG-Erf2p was determined by affinity purification on anti-FLAG antibody-agarose beads followed by an anti-FLAG immunoblot.
Ras PAT Assays-Ras PAT activity was assayed by measuring the incorporation of tritiated palmitate. [ 3 H]palmitoyl-CoA substrate was synthesized from [ 3 H]palmitic acid (NEN) and Coenzyme A using acyl-CoA synthetase (Sigma) and purified as described (23). The PAT assay (25 l of final volume) was performed by adding 1.5 g of GST-Ras(HV)CCaaX (2 M) to GST-Erf4p/FLAG-Erf2p GSH beads in 1 mM DTT,100 mM MES (pH 6.3). The reaction was started by the addition of 1 l of [ 3 H]palmitoyl-CoA (0.5 M), incubated for 15 min at 30°C, and terminated by the addition of 5 l of a 5ϫ solution of SDS gel loading buffer without DTT. Heat inactivation was performed by boiling (100°C, 15 min) the GSH beads containing GST-Erf4p/FLAG-Erf2p prior to the addition of substrates. The assays were analyzed by SDS-PAGE using Bis-Tris gels, pH 6.4, (Nu-PAGE ® ) and subjected to fluorography as described (13). Quantitation of Ras PAT activity was done by excising Ras protein bands from the gel, solubilizing them in Soluene S-350 (Packard), and counting in a scintillation counter.
The following modifications to the Ras PAT assay were performed to determine the fold purification of the enzyme. Cells were lysed by three rounds of high pressure homogenization (Avestin Emulsiflex-C5) (yeast, 30,000 p.s.i.; bacteria 25,000 p.s.i.) in 50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 1 mM DTT, 1 mM EDTA, and 0.13 mM phenylmethylsulfonyl fluoride (Buffer A). The extracts were centrifuged at 3000 rpm (700 ϫ g) (yeast) or 6000 rpm (2800 ϫ g) (bacteria) for 15 min to remove unbroken cells and debris. Triton X-100 (0.3% final) was added to the supernatants, and samples were incubated at room temp for 20 min. The cell lysates were then diluted with equal volumes of Buffer A to a final concentration of 0.15% Triton X-100. GSH-agarose was added, and samples were incubated overnight at 4°C. The beads were washed three times with 50 mM Tris-HCl (pH 7.4), and the beads were taken up in 50 mM Tris-HCl (pH 7.4), 0.02% Triton X-100, and 10% glycerol. Ras PAT activity present on the GSH beads or in the yeast and bacterial extracts were quantitated in a 30-l assay containing 100 mM Tris-HCl, pH 8.0, 0.4 M [ 3 H]palmitoyl-CoA, and 0.75 g of GST-Ras(HV)CCaaX (0.83 M). Protein concentrations were determined by Bradford assay (cell extracts) or modified Bradford dye-binding assay for GSH beadbound protein (24). Reactions were terminated by the addition of 5 l of a 5ϫ solution of SDS gel loading buffer without DTT. Heat inactivation was performed by boiling the GSH beads (100°C, 15 min) containing GST-Erf4p/FLAG-Erf2p prior to substrate addition.

RESULTS
Deletion of either ERF2 or ERF4 results in a nonviable strain when a palmitoylation-dependent Ras2 allele is the only Ras gene expressed (13) (Fig. 1A, lower). The DHHC-CRD is required for Erf2p function; mutation of Cys 189 to Ser or His 201 to Ala abolishes Erf2p function. The phenotypes of single and double deletions of ERF2 and ERF4 mutant strains are indistinguishable suggesting that they function at the same step. Consistent with this prediction, FLAG-tagged Erf2p co-purifies with GST-Erf4p on a GSH-agarose affinity column (Fig. 1B) and GST-Erf4p co-purifies with FLAG-Erf2p using anti-FLAG antibody-agarose beads (data not shown). The level of Erf2p that is detected is reduced when isolated from erf4⌬ cells, suggesting either that Erf4p is involved in the stability of Erf2p or our ability to extract Erf2p (Fig. 1B). Of the two loss-offunction alleles described above, Erf2(C189S)p, fails to interact with Erf4p, and the level of mutant protein detected is reduced. The other non-functional mutant protein, Erf2(H201A)p, is still able to interact with Erf4p (Fig. 1B).
To examine if the Erf2p and Erf4p complex is directly involved in the palmitoylation of Ras, we performed an in vitro palmitoylation assay on the GSH-agarose-enriched Erf2p/ Erf4p complex. A prenylated Ras substrate protein, Ras2-(HV)CCaaX was purified from yeast as a GST fusion protein.
The CaaX box of GST-Ras2(HV)CCaaX is prenylated, aaXproteolyzed, and carboxyl-methylated (25). Incubation of Ras2p with bead-bound Erf2p/Erf4p in the presence of [ 3 H]palmitoyl-CoA led to incorporation of the label onto the Ras2p substrate. (Fig. 1C). As expected of a protein-mediated reaction, incorporation of [ 3 H]palmitate is prevented by heating the beads (100°C, 15 min) prior to adding the substrates. In the presence of Erf2p/Erf4p, 0.14 pmol/min of palmitate was incorporated into the Ras2p substrate. This represents an acceleration of the spontaneous rate of ϳ160-fold.
Consistent with radiolabeling studies in vivo (14), palmitoylation in vitro requires both Erf2p and Erf4p. Removal of either from the assay abolished Ras2p palmitoylation. The DHHC-CRD domain of Erf2p is also important for the palmitoylation reaction. Mutating Cys 189 or His 201 residues within the conserved DHHC-CRD of Erf2p abolished Ras PAT activity (Fig.  1C). The loss of Ras PAT activity observed with the Erf2(C189S)p mutant can be attributed to its absence from the complex (Fig. 1B). However, the loss of Ras PAT activity in the Erf2(H201A)p suggests that the conserved histidine of the DHHC signature sequence might play a more direct role catalyzing palmitate transfer. Mutation of Cys 318 , the residue palmitoylated on Ras in vivo, to Ala also abolished palmitoylation in the in vitro Ras PAT assay ( Table I). Palmitoylation of Ras2p is sensitive to treatment with hydroxylamine (1 M, pH 7.6, 30 min) as expected if the modification is a thioester linkage (data not shown).
The relative enrichment of the GST-Erf4p/FLAG-Erf2p complex from yeast cell extracts using GSH affinity beads was assessed. As seen in Fig. 2 GST-Erf4p is readily detected following staining with Coomassie Blue, but FLAG-Erf2p is below the detection limit. However, it is clear from the anti-FLAG immunoblot that Erf2p is present. Presumably not all GST-Erf4p is complexed with FLAG-Erf2p under these conditions. We often see a slower migrating species in Erf2p immunoblots (Fig. 2, lane 5). The electrophoretic mobility of this slower migrating form is not sensitive to hydroxylamine and therefore does not represent a palmitoylated species. The cause of the slower migrating species that presumably arises by posttranslational modification is under investigation. To further assess the purity of the Erf2p/Erf4p complex we also performed immunoblots for Sec61p, the major integral membrane protein of the ER and Ras2p. Neither appeared to co-purify with Erf2p/ Erf4p, demonstrating that the GST-Erf4p/FLAG-Erf2p complex comprises the major proteins on the GSH beads (data not shown). It was not possible to detect Ras PAT activity above background in a strain expressing endogenous levels of ERF2 and ERF4. However overexpression of GST-Erf4p and FLAG-Erf2p from high copy inducible plasmids results in detectable Ras PAT activity with a specific activity of 37 pmol/min/mg. Partial purification by GSH affinity beads increases the specific activity to 1300 pmol/min/mg, representing a 35-fold purification. The specific activity of the partially purified complex is an underestimate because GST-Erf4p is present in large excess over Erf2p. These data are consistent with the conclusion that the Erf2p/Erf4p complex itself constitutes the Ras PAT activity.
It is still formally possible that PAT activity is co-purifying as a minor component of our yeast affinity purification. To rule this out, we partially purified the Erf2p/Erf4p complex from E. coli expressing an operon fusion of FLAG-Erf2p and GST-Erf4p driven by the P tac promoter (Fig 3A). Proteins are not modified with thioester-linked fatty acids in bacteria. As seen in Fig. 3B, Erf2p/Erf4p isolated from E. coli is able to carry out palmitoylation of Ras2p. The Erf2p/Erf4p activity purified from extracts exhibits the same heat sensitivity as the activity isolated from yeast. No Ras PAT activity is detected in E. coli not expressing ERF2 and ERF4. Total extracts expressing the GST-Erf4p/FLAG-Erf2 operon fusion have Ras PAT activity of 4.5 pmol/min/mg. Purification by GSH-agarose affinity chromatography increases the specific activity to 340 pmol/min/mg or a 76-fold purification. It is not clear at this time if the difference in specific activity between Ras PAT isolated from bacteria compared with that isolated from yeast is significant. FIG. 1. ERF2 and ERF4 are required for the viability of palmitoylation-dependent yeast strains. A, plate assay demonstrating the requirement of functional ERF2 and ERF4 alleles for growth of RJY1277 expressing palmitoylation-dependent Ras2. Mutating residues, C189S or H201A, within the conserved DHHC-CRD of ERF2 leads to a loss in viability comparable to a complete deletion of ERF2. Viability was assessed on rich medium plates containing 2% glucose (YEP 2% Glu) (top) or rich plates containing 2% galactose and 0.1% 5-FOA (5.75 mM) (YEP 2% Gal FOA) (bottom). A detailed description of this assay can be found in Bartels et al. (14). B, immunoblot analysis of GST-Erf4p and FLAG-Erf2p from galactose-induced cultures of RJY543. Cells were solubilized in YPER reagent (Pierce) and the GST fusion proteins were partially purified on GSH-agarose beads (Pierce). Samples were then immunoblotted with either anti-␣-GST to detect GST-Erf4p (top panel) or anti-␣-FLAG antibodies to detect FLAG-Erf2p (middle panel). Expression of FLAG-Erf2p was determined by affinity purification on anti-FLAG antibody linked to agarose beads followed by an anti-FLAG immunoblot (bottom panel). C, the Ras PAT activity of partially purified extracts of ERF2 and ERF4 wild type and mutant strains was assayed. PAT assays (25 l of final volume) were performed as described under "Experimental Procedures" using partially purified Erf2p/Erf4p from YPER-solubilized yeast, and the products analyzed by SDS-PAGE using Bis-Tris gels, pH 6.4 (Nu-PAGE) and fluorography as described (13) (top). The Ras protein bands were excised from the gel and counted in a scintillation counter (bottom). a The PAT activity is the mean Ϯ standard error of triplicate assays for one experiment. All substrates were assayed with the same preparation of Erf2p/Erf4p partially purified from yeast (see Fig. 2).
b Palmitoylated cysteine residue is underlined. c The CaaX box is shown intact for clarity although we presume the actual substrates have undergone prenylation, aaX proteolysis, and carboxyl methylation.
The importance of the signature 200 DHHC 203 motif of the DHHC-CRD is also evident in experiments with bacterially expressed Erf2p/Erf4p. FLAG-Erf2(H201A)p and FLAG-Erf2(C203S)p can be co-purified with GST-Erf4p, but no Ras PAT activity above background levels is detected (Fig. 3).
Erf2p appears to be directly involved in the transfer of palmitate to Ras based on the fact that wild type FLAG-Erf2p becomes labeled in the presence of [ 3 H]palmitate (Fig 4). Incubating Erf2p/Erf4p partially purified from yeast with [ 3 H]palmitate in the absence of the Ras2(HV)CaaX substrate results in the formation of an acyl-enzyme intermediate (Fig.  4). Appearance of the tritium-labeled Erf2p was sensitive to heat inactivation, indicating that palmitoylation requires a native conformation of Erf2p. It was shown in Fig. 3 that the DHHC signature in the DHHC-CRD domain of Erf2p is required for Ras PAT activity. This domain is also involved in the formation of the palmitoyl-Erf2p intermediate. Although the Erf4p-Erf2(H201A)p mutant does not display Ras PAT activity, it still forms the [ 3 H]palmitoylated Erf2(H201A)p acyl-enzyme intermediate (Fig. 4). This suggests that His 201 may play a role in the transfer of palmitate from the acylated Erf2p to Ras. The conserved cysteine of the DHHC motif (Erf2C203Sp) behaves similarly to the Erf2(H201A)p mutant, i.e. stable expression, co-purification with Erf4p, and loss of Ras PAT activity. However, unlike Erf2(H201A)p, the Erf2(C203S)p mutant protein does not form the palmitoylated enzyme intermediate (Fig 4) suggesting that Cys 203 may be the site of palmitate attachment. Together, these results suggest that the Ras PAT reaction involves the formation of a [ 3 H]palmitoylated Erf2p acylenzyme intermediate prior to transfer of labeled palmitate to the Ras substrate.
To begin to evaluate the protein substrate specificity of the Erf2p/Erf4p Ras PAT, full-length GST-Ras2 was compared with GST fused to the final 28 amino acid residues of the Ras2 hypervariable (HV) region. The HV domain is required for palmitoylation of Ras in vivo. 2 Erf2p/Erf4p Ras PAT palmitoylated GST(HV)CCIIS to levels similar to full-length GST-Ras2CCIIS (Table I). Consistent with the importance of the hypervariable domain for substrate recognition, in vitro palmitoylation of GST-CCaaX is reduced 10-fold compared with GST-(HV)CCaaX. Next, we examined whether yeast Ras PAT is able to palmitoylate mammalian H-Ras, which like yeast Ras, is farnesylated on a C-terminal CaaX box and palmitoylated on two adjacent cysteine residues. H-Ras was indeed palmitoylated by yeast Ras PAT but at levels much lower (5%) than the proposed natural substrate, Ras2p. The reduction in labeling could be due to differences between the H-Ras and yeast Ras2 hypervariable domains that we show above is required for Ras PAT activity. Alternatively, it could be due to incomplete farnesylation of the H-Ras purified from Sf9 insect  [ 3 H]palmitoyl-CoA was added to wild type or Erf2p mutant (H201A or C203S) Erf2p/Erf4p complexes, and Ras PAT assays were carried out as described in Fig. 1. The GST-Ras2(HV)CaaX was added to the Ras PAT reactions as indicated. A heat pretreatment (100°C, 15 min) was performed to determine whether labeling was protein-mediated. Samples were resolved by SDS-PAGE Bis-Tris gels, pH 6.4 (Nu-PAGE), fixed, and fluorography was performed. The dried gel was exposed to film for 1 week. The asterisk indicates the migration position of FLAG-Erf2p. The arrow on the right indicates the migration position of GST-Ras2(HV)CCaaX. cells or to residual palmitate remaining on the purified H-Ras. We also examined G i␣1 , a mammalian G-protein ␣ subunit, which is normally palmitoylated on a cysteine residue adjacent to a myristoylated N-terminal glycine. This substrate was produced in a strain expressing N-myristoyltransferase, resulting in a purified preparation that is stoichiometrically N-myristoylated (21) but has not been palmitoylated in vivo. Ras PAT was able to palmitoylate the G i␣1 , but again, the level was ϳ5% that of yeast Ras2p. These results suggest that the Ras PAT is capable of palmitoylating other substrates but exhibits a strong preference for specific protein substrates.
The lipid substrate specificity of Ras PAT was investigated by adding increasing amounts of unlabeled acyl-CoA competitors to the radioactive Ras PAT reaction. As expected, addition of unlabelled palmitoyl-CoA was the most effective competitor, achieving 90% inhibition at 8 M, a 16-fold excess (Fig. 5). Decanoyl (C10:0) was ineffective at similar concentrations. Lauryl (12:0) and myristoyl (14:0) were more effective, but maximal inhibition required palmitoyl or longer acyl chain lengths. Both saturated and unsaturated fatty acyl-CoAs were effective inhibitors. Little difference was observed between C16:0 and C16:1, or C18:0 and C18:1. The lack of complete acyl-CoA substrate specificity of Ras PAT is consistent with the finding that thioester-linked fatty acids on protein are heterogeneous in vivo (1). DISCUSSION Genetic studies have revealed that Erf2p and Erf4p are required for efficient palmitoylation and plasma membrane association of Ras in yeast (14,15). In this study, we establish that Erf2p/Erf4p can directly mediate palmitate transfer to yeast Ras using palmitoyl-CoA as a donor, thereby acting as a palmitoyltransferase. Erf2p and Erf4p copurify as a complex of unknown stoichiometry. Erf2p appears to have a direct role in palmitate transfer based on its ability to form an acyl-interme-diate that is dependent upon the DHHC signature motif in the DHHC-CRD. It is important to note that the putative metazoan homologs of Erf2p were identified based on the presence of the highly conserved DHHC-CRD motif that we have shown is required for Ras PAT activity, leading us to suspect that other members of the DHHC-CRD family too are palmitoyltransferases (14). Consistent with this prediction, preliminary results with two other members of the yeast Erf2p family, Akr1p and YOL003c, indicate that these two DHHC-CRD proteins can also palmitoylate Ras2p, albeit at a lower efficiency than Erf2p. 3 Presumably, Akr1p and YOL003c palmitoylate other substrates preferentially in vivo. It will be interesting to examine whether metazoan DHHC-CRD proteins also possess PAT activity. For the most part, there is little functional information known about these other DHHC-CRD proteins (17). However, recently a c-Abl-interacting protein containing a DHHC-CRD motif, Aph2, has been identified by a two hybrid screen (25). Co-expression of Aph2 and c-Abl results in induction of apoptosis in COS-7 cells.
Erf4p is necessary for stable expression or solubilization of Erf2p from yeast cells, suggesting that it may act as a chaperone for Erf2p. Whether it has other functions in mediating palmitate transfer is under investigation. Erf4p does not have any obvious protein domains that predict its function, and although putative Erf4p homologs can be found in other fungal genomes, it has not been possible to identify metazoan homologs based on sequence identity. It is not known at this time whether other DHHC-CRD proteins require association with auxiliary proteins similar to the Erf2p interaction with Erf4p.
At present it is not clear whether Er2p/Erf4p acts as a typical enzyme or palmitoylates Ras through an unconventional mechanism (26). Resolving this issue in the present study is not possible due to several factors, including limitations in the assay (subsaturating concentrations of Ras substrate) and the difficulty of estimating the amount of active Erf2p/Erf4p in the GST-pulldown. Current efforts are focused on obtaining a purified preparation of Erf2p/Erf4p that is amenable to a more detailed mechanistic analysis.
The Erf2p/Erf4p-dependent Ras PAT is distinct in several ways from the recently proposed Hedgehog palmitoyltransferase, skinny hedgehog/sightless (ski) (5). Ski shares a short region of sequence homology with membrane-bound acyltransferases, including acyl-coenzyme A:cholesterol acyltransferase and diacylglycerol acyltransferase, whereas there is no significant homology between Erf2p or Erf4p and this class of enzymes. The mechanism of palmitate addition by Ski is unclear. It is predicted to occur through a thioester linkage to the N-terminal cysteine residue, followed by a rearrangement through a cyclic intermediate to form an amide (5). Ras PAT modifies Cys, preferably in the context of a C-terminal-prenylated cysteine, via thioester linkage to cysteine. The substrate hedgehog is an extracellular protein that is modified either in the lumen of the secretory system or on the surface of the cell, whereas Ras is modified by the Erf2p/Erf4p complex, which is localized to the ER (17). The Erf2p/Erf4p-dependent Ras PAT may therefore represent a new class of enzymes required for the palmitoylation of cellular proteins. Erf2p/Erf4p exhibits a preference for yeast Ras substrates. Based on these results, we propose that other members of the Erf2p DHHC-CRD family carry out thioacylation of other lipid-modified proteins.