DHHC9 and GCP16 constitute a human protein fatty acyltransferase with specificity for H- and N-Ras.

Covalent lipid modifications mediate the membrane attachment and biological activity of Ras proteins. All Ras isoforms are farnesylated and carboxyl-methylated at the terminal cysteine; H-Ras and N-Ras are further modified by palmitoylation. Yeast Ras is palmitoylated by the DHHC cysteine-rich domain-containing protein Erf2 in a complex with Erf4. Here we report that H- and N-Ras are palmitoylated by a human protein palmitoyltransferase encoded by the ZDHHC9 and GCP16 genes. DHHC9 is an integral membrane protein that contains a DHHC cysteine-rich domain. GCP16 encodes a Golgi-localized membrane protein that has limited sequence similarity to yeast Erf4. DHHC9 and GCP16 co-distribute in the Golgi apparatus, a location consistent with the site of mammalian Ras palmitoylation in vivo. Like yeast Erf2.Erf4, DHHC9 and GCP16 form a protein complex, and DHHC9 requires GCP16 for protein fatty acyltransferase activity and protein stability. Purified DHHC9.GCP16 exhibits substrate specificity, palmitoylating H- and N-Ras but not myristoylated G (alphai1) or GAP-43, proteins with N-terminal palmitoylation motifs. Hence, DHHC9.GCP16 displays the properties of a functional human ortholog of the yeast Ras palmitoyltransferase.

The Ras family of proteins (H-Ras, N-Ras, and K-Ras) are membrane-localized GTPases ϳ21 kDa in size (1,2). Ras proteins are highly conserved with differences among the isoforms occurring in the C-terminal hypervariable domain of about 25 amino acids. Lipid modifications localize Ras proteins to the cytoplasmic face of the plasma membrane (PM) 1 or Golgi apparatus (reviewed in Refs. [3][4][5][6]. An essential motif for localiz-ing Ras is the C-terminal -CaaX motif where "C" represents cysteine, "a" residues are typically aliphatic, and the "X" residue is methionine in N-and K-Ras and serine in H-Ras. The -CaaX motif of Ras is prenylated at the cysteine with a C15 farnesyl group by a cytoplasmic farnesyltransferase. This is followed by -aaX proteolysis and carboxylmethylation of the farnesylated cysteine catalyzed by enzymes localized in the endoplasmic reticulum. N-Ras and H-Ras have, respectively, one and two additional cysteine residues immediately upstream of the prenylated cysteine that are palmitoylated, whereas K-Ras has a polybasic domain. Ras signals from endomembranes as well as from the PM (7,8), and Ras localization is critical for its function in signal transduction pathways. Nonprenylated forms of Ras proteins are cytoplasmic, fail to signal, and cannot transform cells (9). Palmitoylation of Ras increases the affinity of farnesylated Ras for membranes and contributes to its biological activity (10,11). Palmitoylation of Ras is a dynamic modification that dictates the distribution of H-and N-Ras on the Golgi and at the PM (12). Studies of H-and N-Ras trafficking in mammalian cells indicate that palmitoylation occurs early in the secretory pathway and facilitates transport to the PM (13,14).
The molecular identity of the protein fatty acyltransferases (PATs) that modify palmitoylated proteins has only recently been discovered (15). Genetic and biochemical studies in Saccharomyces cerevisiae revealed that a complex of Erf2 (effect on Ras function 2) and Erf4 palmitoylates Ras2 (16), and the yeast protein Akr1 palmitoylates Yck2 (yeast casein kinase 2) (17). Erf2 and Akr1 share a common domain, the Asp-His-His-Cys (DHHC), within a cysteine-rich domain (CRD). In Akr1, the DHHC motif is replaced with DHYC. The CRD domain is essential for PAT activity. In both enzymes, mutations in the first His and the Cys of the DHHC motif abolish activity (16,17). The finding that these two otherwise unrelated proteins share a common protein domain that is essential for PAT activity has lead to the hypothesis that the DHHC-CRD is a protein fatty acyltransferase domain (15)(16)(17).
We sought to identify a human counterpart of the yeast Erf2⅐Erf4 complex that would palmitoylate mammalian Ras proteins. Here we describe a protein complex composed of DHHC9 and a Golgi-localized protein designated GCP16 (25) that harbors palmitoyltransferase activity for H-and N-Ras.

EXPERIMENTAL PROCEDURES
Construction of Expression Plasmids-Standard molecular biological techniques were used to manipulate DNA. All constructs derived from PCR were cloned directly into pCR2.1-TOPO vector (Invitrogen) and verified by DNA sequencing. The pCDNA3.1-Myc-His vectors are from Invitrogen. The pEGFP-N and -C vectors that fuse the green fluorescent protein (GFP) coding region to the 3Ј or 5Ј ends, respectively, of the insert were purchased from Clontech. All of the point mutations were created using a QuikChange mutagenesis kit (Stratagene) and verified by DNA sequencing. DHHC9 (accession number BC006200) and GCP16 (accession number BC001227) were obtained from the Image consortium (Image numbers 2964425 and 3456384, respectively). The GCP16 cDNA was PCR-amplified from the Image clone and subcloned from the TOPO vector into the XhoI/EcoRI sites of pEGFPC3, resulting in GFP-GCP16. For FLAG-GCP16, PCR primers were designed to add a FLAG tag to the N terminus and maintain the stop codon at the C terminus of GCP16 prior to subcloning as a NotI/EcoRI fragment into pCDNA3.1C(Ϫ). The same FLAG-GCP16 fragment was subcloned into the NotI/EcoRI sites of pBlueBac4.5 to generate high titer FLAG-GCP16 baculovirus using the Bac-N-Blue transfection kit (Invitrogen). The DHHC9 cDNA was PCR-amplified from the Image clone and subcloned from pCR2.1TOPO into the EcoRI/BamHI sites of either pCDNA3.1C(Ϫ) (DHHC9-Myc-His) or pEGFPN1 (DHHC9-GFP). The stop codon was mutated to alanine to prevent termination of translation prior to the C-terminal tags. The DHHC9-Myc-His cDNA was subcloned into the EcoRI/NotI sites of pVL1393. High titer DHHC9-Myc-His baculovirus was generated using the Bac-N-Blue transfection kit.
Antibodies-Myc-peptide fusion proteins were detected using monoclonal (mouse ascites) or polyclonal (Alexis) antibodies to the human Myc peptide. FLAG-tagged peptides were detected by probing with ANTI-FLAG® M2 monoclonal antibody (Sigma). Rabbit polyclonal GFP antibodies were generated and affinity-purified as described (26). For confocal microscopy, rabbit polyclonal antibodies to giantin (Covance) and TRAP-␣ (translocon associated protein-␣) (27) were used to detect the Golgi apparatus and ER, respectively, both at a 1:1000 dilution. The TRAP-␣ antibody was a kind gift of Dr. Ramanujan Hegde (NICHD, National Institutes of Health). Secondary antibodies used for immunofluorescence were Alexa 546-conjugated (red) goat anti-mouse, Alexa 488-conjugated (green) goat anti-rabbit (Molecular Probes), and Cy3 (red) (Novus) antibodies, all at a dilution of 1:1000.
Northern Blotting-Northern blotting was carried out using a human multiple tissue Northern blot (BD Bioscience Clontech) containing poly(A) mRNA (1 g/lane). The membranes were prehybridized using ExpressHyb solution (BD Bioscience Clontech) with continuous shaking for 30 min at 60°C. Hybridization was performed at 65°C for 2 h with a [␣-32 P]dCTP labeled DNA fragment in ExpressHyb solution. The probes used were a 1.0-kb EcoRI/BamHI cDNA fragment corresponding to the open reading frame of DHHC9, a 0.5-kb NotI/EcoRI cDNA fragment corresponding to the open reading frame of GCP16, and the human ␤-actin cDNA control probe (BD Bioscience Clontech). The probes were prepared using the Prime-a-Gene kit (Promega) according to the manufacturer's protocol.
Cell Culture and Transient Transfection-HEK-293 cells (ATCC) were cultured in high glucose Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum, 2 mM glutamine, 150 units/ml penicillin, and 50 units/ml streptomycin. The cells were seeded at ϳ3 ϫ 10 5 cells/well of a six-well plate or 1 ϫ 10 6 cells/plate for 100-mm dishes and allowed to attach overnight at 37°C in 5% CO 2 . The next day, the cells were transfected using Lipofectamine 2000 reagent (Invitrogen) in Opti-MEM® medium (Invitrogen) without serum according to the manufacturer's protocol. For confocal microscopy, the cells were seeded into six-well plates containing coverslips coated with poly-L-lysine, transfected, and analyzed 24 -48 h later.
Immunofluorescence and Confocal Laser Microscopy-For immunofluorescence, the cells (48 h post-transfection) were washed with PBS and fixed in a freshly prepared solution of 4% (v/v) paraformaldehyde and 5% (w/v) sucrose for 10 min at room temperature. The cells were washed twice with PBS, permeabilized with 1% Triton X-100 in blocking buffer (1% bovine serum albumin w/v, 1% normal goat serum v/v in PBS) for 10 min, and washed once in blocking buffer. Nonspecific staining was blocked by incubating cells for 45 min at room temperature in blocking buffer. For analysis of ER localization, the cells were washed in blocking buffer containing 0.1% saponin and blocked for 1 h at room temperature with blocking buffer containing 0.1% saponin and 50 g/ml RNase A. All of the antibodies were diluted in blocking buffer. The cells were incubated in the appropriate primary antibody for 1 h, washed three times in blocking buffer, and incubated for 1 h with appropriate secondary antibodies. After incubation in secondary antibody, the cells were washed three times in PBS, mounted onto slides with a drop of Vectashield mounting medium (Vector Laboratories), and immediately visualized using a Zeiss LSM 510 confocal microscope with a 63ϫ oil immersion objective. The confocal images were assembled as montages using Adobe PhotoShop 7.0 and adjusted only for brightness and contrast.
Expression and Purification of Protein Substrates-Recombinant baculoviruses expressing human H-Ras or N-Ras as N-terminal His 6tagged fusions were generated using the Bac-N-Blue Expression System (Invitrogen) according to the manufacturer's instructions. The N-Ras/pVL1393 transfer vector was a kind gift of Dr. Andrew Chan (Genentech). The transfer vector encoding H-Ras was prepared as follows. A BamHI-XhoI fragment encoding H-Ras was excised from H-Ras/pcDNA3.1ϩ (Guthrie Institute) and subcloned into pBlueBacHis2A digested with the same enzymes. H-Ras was purified from detergent extracts of Sf9 membranes using nickel chelate affinity chromatography (28). The stoichiometry of palmitate on the purified H-Ras has not been determined, but a significant fraction is probably lost during purification because of the action of thioesterases and the presence of reducing agents in the buffers. N-Ras was purified from the soluble fraction of Sf9 cells using nickel chelate affinity chromatography. Myristoylated G␣ i1 was purified from bacteria expressing N-myristoyl transferase (29). To form heterotrimeric G i , G␣ i1 was reconstituted with nonprenylated G ␤␥C68S subunits purified from Sf9 cells (30). Recombinant GAP-43 was expressed in Escherichia coli and purified using calmodulin-Sepharose affinity chromatography (31). The GAP-43 expression vector (32) (33). The PAT assay (final volume, 50 l) was performed by mixing DHHC9⅐GCP16 in 50 mM Tris (pH 7.4), 125 mM NaCl, 1 mM DTT, 10% glycerol, and 0.1% dodecyl maltoside with 10 l of Ras substrate in 50 mM Tris (pH 7.4), 1 mM EDTA, and 1 mM DTT. The reaction was initiated by the addition of 30 l of reaction mix (167 mM Mes, pH 6.4, 1 mM DTT, 1.7 mM [ 3 H]palmitate-CoA). The reaction was terminated by the addition of 12.5 l of a 5ϫ solution of SDS gel loading buffer with DTT (final concentration, 5 mM), and boiled for 1 min before resolution on 13% SDS-PAGE. The gels were soaked for 20 min in a solution containing 1 M sodium salicylate and 15% methanol, dried, and subjected to fluorography. Alternatively, Ras protein bands were excised from the gel, solubilized in Soluene S-350 (Packard) for 3 h at 50°C, and quantitated by liquid scintillation spectroscopy.
Preparation of DHHC9⅐GCP16 Protein Complex-All of the buffers contain the protease inhibitors 10 g/ml aprotinin, 10 g/ml leupeptin, 10 g/ml lima bean, 10 g/ml pepstatin A, and 1 mM phenylmethylsulfonyl fluoride, and all of the steps were performed at 4°C unless indicated. Sf9 cells were coinoculated with baculovirus expressing DHHC9 or GCP16 and incubated for 72 h at 27°C. The cell pellet was suspended in 200 ml of cavitation buffer (50 mM Tris, pH 7.4, 150 mM NaCl, 1 mM EDTA, 10 mM ␤-mercaptoethanol) and incubated for 30 min under pressure at 500 psi. The lysate was centrifuged at 800 rpm for 10 min to remove nuclei and unbroken cells. The postnuclear supernatant was centrifuged at 100,000 ϫ g for 30 min generating P100 and S100 fractions. P100 membranes were suspended in 100 ml of extraction buffer (50 mM Tris, pH 7.4, 125 mM NaCl, 10 mM ␤-mercaptoethanol, 10% glycerol, 1% dodecyl-maltoside), subjected to Dounce homogenization, and incubated for 1 h with end-over-end rotation. The extract was cleared at 100,000 ϫ g for 30 min, diluted 2-fold in extraction buffer without dodecyl-maltoside, and loaded onto 5 ml of Ni 2ϩ -nitrilotriacetic acid-agarose resin (Qiagen) pre-equilibrated with extraction buffer containing 0.1% dodecyl-maltoside. The resin was washed in 20 volumes of wash buffer (50 mM Tris, pH 7.4, 25 mM NaCl, 10 mM ␤-mercaptoethanol, 10% glycerol, 0.1% dodecyl-maltoside, 5 mM imidazole, pH 7.4), and the complex was batch-eluted (3 ϫ 5 ml) in wash buffer containing 200 mM imidazole. The first two elutions were pooled and diluted 2-fold in 50 mM Tris (pH 7.4), 10 mM ␤-mercaptoethanol, 10% glycerol, 0.1% dodecyl-maltoside resulting in a final concentration of 12.5 mM NaCl and 100 mM imidazole and were further purified on a Mono Q HR 5.5 ion exchange column using a fast protein liquid chromatography system (Amersham Biosciences). The column was washed with five volumes of Mono Q buffer (50 mM Tris, pH 7.4, 1 mM EDTA, 1 mM DTT, 10% glycerol, 0.1% dodecyl-maltoside) and eluted with a 30-ml gradient of NaCl (0 -1 M) in Mono Q buffer. Fractions containing DHHC9⅐GCP16 were identified by Western blot, pooled, and diluted 2-fold in buffer containing 50 mM Tris, pH 7.4, 10% glycerol, 0.1% dodecyl-maltoside, resulting in a final DTT concentration of 0.5 mM. The MonoQ pool was incubated with 150 l of ANTI-FLAG® affinity gel (Sigma) for 2 h by end-over-end rotation. The resin was transferred to a column, washed with buffer containing 50 mM Tris, pH 7.4, 125 mM NaCl, 0.5 mM DTT, 10% glycerol, 0.1% dodecyl-maltoside, and eluted in the same buffer containing 0.2 mg/ml FLAG peptide (Sigma). The purity of the enzyme preparation was determined by silver stain as described (34). The concentration of enzyme was determined by extrapolation from a linear curve with known concentrations of bovine serum albumin using Sypro Ruby protein gel stain (Molecular Probes) and quantitation with a Storm TM 860 (Amersham Biosciences).

Identification of a Candidate Mammalian Ras PAT-A
search of mammalian genomic data bases with the DHHC-CRD sequence of yeast Erf2 uncovered approximately 20 DHHC proteins. Of these, DHHC9 was chosen as the best candidate based on expect value (2 e Ϫ31 ), percent identity, and predicted structural similarities to Erf2. DHHC9 is a 364-amino acid protein (Fig. 1A) that shares 70% sequence identity with Erf2 in the DHHC-CRD (Fig. 1B) and 31% identity overall. Like Erf2, the sequence predicts a protein with four transmembrane spans and a DHHC domain in the cytoplasmic loop between the predicted second and third transmembrane spans (Fig. 1A). In yeast, Erf2 requires a second subunit, Erf4, for Ras PAT activity (16). Initial attempts to identify a mammalian homolog of S. cerevisiae Erf4 were unsuccessful. However, by aligning Erf4 sequences from S. cerevisiae and Candida albicans, it was possible to identify a putative Aspergillus homolog (tigrblast. tigr.org/er-blast/index.cgi?projectϭafu1). A BLAST search with the Aspergillus sequence yielded a candidate Erf4 sequence in Schizosaccharomyces pombe. The S. pombe sequence was used as a query in iterative searches with PSI-BLAST (35) to identify candidate mouse and human Erf4 homologs. A sequence alignment of the fungal, mouse, and human sequences is shown in Fig. 1C. The mammalian candidate for an Erf4 homolog is identical to GCP16 (Golgi-complex associated protein of 16 kDa), a palmitoylated protein that interacts with the Golgiassociated protein GCP170 (25). We obtained full-length cDNAs for DHHC9 and GCP16 to investigate whether they encoded functional orthologs of Erf2 and Erf4.
Tissue Distribution of DHHC9 and GCP16 -To determine whether the DHHC9 cDNA represents an authentic transcript, a Northern blot of poly(A) mRNA from various human tissues was hybridized with a probe derived from the DHHC9 cDNA. An abundant transcript (2.3 kb) was identified in kidney, skeletal muscle, brain, lung, and liver and to a lesser extent in placenta, heart, colon, and small intestine (Fig. 2A). The DHHC9 transcript was absent from thymus, spleen, and peripheral blood leukocytes. A previous study reported that GCP16 mRNA is also expressed ubiquitously (25). We detected a transcript of similar size (1.9 kb) in all tissues except colon and thymus (Fig. 2B). There is a high degree of overlap in the expression of transcripts for DHHC9 and GCP16 (Fig. 2C). A notable exception is in peripheral blood cells where GC16 is found but DHHC9 is not.
Subcellular Distribution of DHHC9 and GCP16 in HEK-293 Cells-Both yeast Erf2 and Erf4 behave as integral membrane proteins (36,37). To determine whether DHHC9 and GCP16 have similar properties, the nature of their membrane association was assessed. When expressed in HEK-293 cells, DHHC9 and GCP16 were resistant to extraction from membranes with salt or high pH but were extracted with the detergent dodecyl maltoside (data not shown). This behavior is characteristic of integral membrane proteins and similar to that of yeast Erf2⅐Erf4 (36,37).
To address whether DHHC9 and GCP16 were associated with the same membrane compartment, we performed subcellular localization experiments using confocal microscopy and GFP and Myc-tagged proteins. GFP-GCP16 and DHHC9-Myc codistributed on internal membranes in HEK-293 cells (Fig.  3A). Endogenous GCP16 has been previously shown to localize to the Golgi (25), consistent with our findings with the transfected protein (data not shown). DHHC9-Myc codistributed with the Golgi-specific marker giantin in HEK-293 cells expressing DHHC9 and GCP16 (Fig. 3B). Thus, both GCP16 and DHHC9 are localized at the Golgi. Studies of Ras palmitoylation, and trafficking in tissue culture cells indicates that mammalian Ras is palmitoylated on the Golgi (12)(13)(14). DHHC9⅐GCP16 therefore fulfills the requirement for a Golgilocalized Ras PAT.
DHHC9-Myc immunofluorescence was also observed on internal membranes that extended to the periphery of the cell and coincided with that of an ER marker, TRAP-␣ (27) (Fig.  3C). Thus, DHHC9 may also be associated with the endoplasmic reticulum.
Coimmunoprecipitation of DHHC9 and GCP16 -If DHHC9 and GCP16 represent functional homologs of yeast Erf2⅐Erf4, then they are expected to form a stable protein complex. To investigate whether DHHC9 and GCP16 interact, coimmunoprecipitation of these proteins was performed on lysates from HEK-293 cells transfected with DHHC9-GFP and FLAG-GCP16 (Fig. 4). Immunoprecipitation of DHHC9 with anti-GFP polyclonal antibody was sufficient to pull down GCP16 (Fig. 4A,  lane 3), whereas no immunoreactive protein was observed following immunoprecipitation of lysates from cells expressing DHHC9 alone (Fig. 4A, lane 1) or GCP16 alone (Fig. 4A, lane 2). Conversely, immunoprecipitation of GCP16 with anti-FLAG antibody precipitated DHHC9 from lysates from cells transfected with both DHHC9 and GCP16 (Fig. 4C, lane 3) but not from cells expressing GCP16 alone (Fig. 4C, lane 2) or DHHC9 alone (Fig. 4C, lane 1). To demonstrate immunoprecipitation of DHHC9 and GCP16, nitrocellulose membranes were stripped and reprobed with anti-GFP (Fig. 4B) or anti-FLAG (Fig. 4D) antibodies. DHHC9-GFP and FLAG-GCP16 were present in the appropriate immunoprecipitates.
The DHHC9⅐GCP16 Complex Is Required for Protein Acyltransferase Activity-To determine whether the DHHC9⅐ GCP16 complex has PAT activity, we expressed DHHC9-Myc-His and FLAG-GCP16 in insect cells using recombinant baculoviruses and partially purified the complex from detergent extracts using nickel affinity chromatography. As controls, DHHC9-Myc-His was expressed alone, and DHHC9-Myc-His(C169S) was expressed with FLAG-GCP16. Cysteine 169 of DHHC9 corresponds to the residue within the DHHC motif of the DHHC9-CRD. Mutation of the corresponding cysteine in all DHHC-CRD PATs characterized to date inactivates the enzyme (16,17,19,23,24). Similar to mammalian cells, FLAG-GCP16 copurified with DHHC9-Myc-His when expressed in insect cells (Fig. 5A, lane 2). GCP16 also copurified with the DHHS mutant of DHHC9 (Fig. 5A, lane 3). DHHC9 expressed alone appeared to be less stable than when it was expressed with GCP16 because a significant pool of proteolyzed DHHC9 was eluted from the nickel resin in addition to full-length protein (Fig. 5A, lane 1).
We next assayed nickel elutions for PAT activity using H-Ras as a substrate (Fig. 5B). Radioactive palmitate was incorporated both into DHHC9 and H-Ras. Heat inactivation of DHHC9 essentially eliminated Ras palmitoylation, showing a dependence of the reaction on the integrity of DHHC9 and eliminating the possibility that Ras was autoacylating in the reaction Fig. 5B, lanes 6 and 8). Both autoacylation of DHHC9 (Fig. 5B, lanes 5 and 7) and palmitoylation of H-Ras (Fig. 5B,  lane 7) were dependent upon the presence of GCP16 in the complex. Although the amount of full-length DHHC9 in the assay was ϳ2-fold less than that in the DHHC9⅐GCP16 complex, this does not account for the dramatic reduction in Ras palmitoylation (Fig. 5B, lane 3) and the loss of DHHC9 autoacylation (Fig. 5B, lanes 1 and 3). Finally, both DHHC9 autoacylation and palmitoylation of H-Ras were not observed when the conserved cysteine (Cys 169 ) of the DHHC motif is mutated to a serine (Fig. 5B, lanes 9 and 11).

Purified DHHC9⅐GCP16 Is a Protein Acyltransferase with Specificity for H-and N-Ras-
The DHHC9-Myc-His⅐FLAG-GCP16 complex was purified from insect cells to apparent homogeneity using sequential chromatography over nickelagarose, anion exchange, and FLAG immunoaffinity columns. The final elution from the FLAG affinity resin was analyzed by silver staining. Two prominent bands consistent with the molecular weights of DHHC9-Myc-His and FLAG-GCP16 can be observed (Fig. 6A). The identities of the silver-stained bands as DHHC9-Myc-His and FLAG-GCP16 were confirmed by Western blot (Fig. 6C). The PAT activity of the preparation was documented by autoacylation of DHHC9-Myc-His (Fig. 6B).
To begin to address the substrate specificity of DHHC9⅐GCP16, we assayed its activity for proteins with Nand C-terminal palmitoylation motifs. N-Ras and H-Ras are palmitoylated at one or two cysteine residues, respectively, that lie immediately upstream of the C-terminal farnesylated cysteine. The DHHC9⅐GCP16 complex palmitoylated both Hand N-Ras (Fig. 7, lanes 3 and 7). Palmitate incorporation into H-Ras in vivo is dependent upon cysteine residues 181 and 184 (38). Similarly, H-Ras (C181S,C184S) was not palmitoylated in vitro by DHHC9⅐GCP16 (Fig. 7, lane 5). In vitro palmitoylation of Ras was also lost when the DHHC9⅐GCP16 complex was heat denatured prior to the assay (Fig. 7, lanes 4 and 8). Two other palmitoylated proteins, G␣ i1 and GAP-43 that are normally modified near the N terminus were assayed. In cells, G␣ i1 is N-myristoylated prior to palmitoylation at the adjacent cysteine; GAP-43 is solely modified with palmitate at two cysteine residues near the N terminus (39). Purified DHHC9⅐GCP16 did not palmitoylate G␣ i1 in the presence or absence of G ␤␥ subunits or GAP-43 in vitro (Fig. 7, lanes 1 and 9, and data not  shown). These results suggest that DHHC9⅐GCP16 is able to distinguish different palmitoylation motifs in target proteins.
Palmitoylation of Ras by DHHC9⅐GCP16 Is Catalytic-Using purified DHHC9⅐GCP16, we performed a time course and substrate concentration curves using N-Ras as a substrate to ascertain the rate of palmitate transfer and to demonstrate enzyme turnover (Fig. 8). N-Ras purified from the soluble fraction of insect cells was used as a substrate to avoid the complications of detergent effects on enzyme activity. The transfer of palmitate to N-Ras occurred as early as 1 min, was linear up to 8 min, and reached saturation by 10 min (Fig. 8A). In the absence of enzyme, there was no significant palmitate incorporation into N-Ras (Fig. 8A). To determine whether palmitoylation was catalytic, PAT assays were performed on increasing Nonetheless, it appears that palmitoylation of Ras by DHHC9⅐GCP16 is catalytic. The observed maximum was 2.8 pmol of palmitate transferred to substrate by 0.12 pmol of enzyme in 8 min. This yields a turnover number of approximately three/min. DISCUSSION In this study we have identified and characterized a human Ras PAT localized in the Golgi apparatus. Similar to the yeast Ras PAT (16), the active enzyme required two components: a DHHC-CRD-containing protein DHHC9 and GCP16, a Golgilocalized protein that had previously been linked to protein trafficking through the secretory pathway. DHHC9 and GCP16 copurified following detergent extraction from cell membranes and palmitoylated both H-and N-Ras. DHHC9 expressed and purified in the absence of GCP16 was inactive as a PAT. DHHC9⅐GCP16 exhibited substrate selectivity, palmitoylating Ras proteins that are modified near a farnesylated cysteine, but showed no activity toward substrates palmitoylated near the N terminus, N-myristoylated G␣ i1 , or GAP-43. Kinetic analysis of DHHC9⅐GCP16 demonstrated catalytic activity of the complex toward N-Ras. The molecular and biochemical properties of DHHC9⅐GCP16 underscore its functional and structural homology to the yeast Ras PAT, Erf2⅐Erf4.
The requirement of DHHC9 PAT activity for a second protein distinguishes it from other DHHC-CRD PATs. Purified preparations of yeast Akr1 (17) and mammalian DHHC15 (23) and HIP14 (24) are sufficient for palmitate transfer to their respective protein substrates. By contrast, DHHC9, like Erf2 (16), does not autoacylate, nor does it palmitoylate Ras in the absence of GCP16. How these partner proteins impart PAT activity to their corresponding DHHC-CRD proteins remains to be determined. One possibility is that GCP16 and Erf4 stabilize their respective DHHC-CRD proteins, permitting them to adopt a conformation necessary for activity. Consistent with this idea, DHHC9 was more prone to proteolysis when expressed in insect cells in the absence of GCP16 (Fig. 5). However, it is interesting to speculate that partner proteins like GCP16 and Erf4 have a more active role in the reaction cycle and may be a general requirement for all DHHC-CRD PATs. There are indications that the human and Ras PATs palmitoylate their substrates with a faster kinetics than do DHHC-CRD PATs that function alone. The yeast DHHC-CRD protein PfaIII (YNL326) functions as a PAT for the N-myristoylated protein Vac8. 2 Whereas we measured a turnover number of three/min for DHHC/GCP16 in this study, the turnover number for PfaIII was estimated at two/h. 2 More extensive kinetic analysis of the DHHC-CRD PATs will be important for elucidating the requirement for and function of binding partners.
GCP16 was previously identified as a palmitoylated protein that interacts with the Golgi autoantigen GCP170 (25). GCP170 is a member of the golgin family, high molecular weight proteins with cytoplasmic coiled-coil domains that are localized to the cytoplasmic face of the Golgi cisternae (40). The golgins GM130 and giantin are structural components of the Golgi apparatus linked functionally to the tethering of transport vesicles. GCP16 is a 137-amino acid protein with no obvious structural features other than a short predicted coiled-coil domain. Ohta et al. (25) found that GCP16 is palmitoylated and that fatty acylation accounts for the behavior of GCP16 as an integral membrane protein and its association with the Golgi complex. Overexpression of GCP16 in COS cells impaired trafficking of vesicular stomatitis virus glycoprotein to the cell surface and modestly inhibited secretion of a soluble form of dipeptidyl peptidase IV without having an obvious effect on Golgi morphology (25). This suggests a potential role for GCP16 in vesicular transport through the secretory pathway. Our finding that GCP16 is required for palmitoylation of Ras proteins in vitro predicts its direct involvement in the trafficking of palmitoylated Ras proteins to the cell surface in vivo. Palmitoylation of Ras is required for exit from the ER in yeast and from the Golgi in mammalian cells for subsequent trafficking to the PM (13,14,36). How the role of GCP16 as a subunit of a Ras PAT relates to its interaction with GCP170 and the trafficking of other proteins is unclear at present but hints at an intimate relationship between the palmitoylation apparatus and the machinery involved in protein transport.
The sequence similarity between GCP16 and yeast Erf4 is limited (Fig. 1C). However, it is clear that they have similar functions. GCP16 is palmitoylated at cysteines 69 and 72 in vivo (25). Interestingly, we did not observe any palmitoylation of GCP16 when the DHHC9⅐GCP16 complex was incubated with [ 3 H]palmitoyl-CoA. We presume that GCP16 is palmito- ylated in insect cells because palmitoylation is required for membrane association (25). Accordingly, the palmitoylated cysteine residues in GCP16 may not be available for palmitoylation in vitro. Alternatively, GCP16 may be a substrate for another PAT. In yeast, Erf4 is membrane-bound even when Erf2 is deleted. 3 It is likely that additional Ras PATs are present in mammalian cells in addition to DHHC9⅐GCP16. The absence of DHHC9 mRNA expression in peripheral blood cells, thymus, and spleen suggest that other enzymes are palmitoylating H-and N-Ras in those tissues. HIP14 is a possible candidate. Ducker et al. (19) demonstrated that membrane preparations from cells overproducing HIP14 have increased PAT activity for a farnesylated peptide derived from the N-Ras sequence. Interestingly, HIP14 acts as an oncogene when expressed in NIH-3T3 cells. Its transforming ability is dependent upon the cysteine in its DHHC motif, suggesting PAT activity as a possible mechanism. It is unknown whether palmitoylation of Ras or another protein target is important for HIP14 oncogenic activity. A separate study of HIP14 reported PAT activity for SNAP-25, PSD-95, GAD65, synaptotagmin, and huntingtin but not Lck or H-Ras (24). The discrepancy between the two studies demonstrates a need for more detailed kinetic analysis of HIP14 to establish its substrate specificity (41).
Phylogenetic analysis of DHHC-CRDs places DHHC9 in a subgroup with five other mammalian DHHC-CRD proteins (DHHC18, 14, 5, 8, and 19) (23). 4 It will be of obvious interest to determine whether any of these proteins have Ras PAT activity and whether they require GCP16 or another binding partner. Heterologous expression of DHHC18 and H-Ras in HEK-293 cells resulted in increased metabolic radiolabeling of H-Ras with palmitate, whereas DHHC2, 3, 7, and 15 and HIP14 (DHHC17) overexpression had no effect (23). A similar assay was a good predictor of PAT activity for PSD-95. All 23 murine DHHC-CRD proteins were screened for their ability to increase palmitoylation of PSD-95 in a cotransfection and metabolic radiolabeling assay. DHHC2, 3, 7, and 15, which represent another subfamily of the DHHC-CRDs, gave the most robust radiolabeling of exogenously expressed PSD-95 in COS and HEK-293 cells. The authors went on to show that DHHC15 and DHHC2, but not DHHC18, palmitoylated PSD-95 in vitro (23). The selective palmitoylation of PSD-95 by a subset of the DHHC-CRDs supports the hypothesis that DHHC-CRD proteins have substrate specificity.
A recent study highlights the importance of palmitate turnover on Ras in the regulation of its localization and activity (12). The authors propose that palmitoylated Ras isoforms undergo a constitutive cycle of deacylation and reacylation. The deacylated form rapidly exchanges between a cytoplasm and membranes. Palmitoylation of Ras in the Golgi stabilizes its association with Golgi membranes, allowing it to signal from the Golgi or to be packaged into vesicles for transport to the PM. The localization of DHHC9⅐GCP16 in the Golgi makes it a bona fide candidate for a mammalian Ras PAT in vivo and an important regulator of Ras signaling.