Mutational Analysis of Saccharomyces cerevisiae Erf2 Reveals a Two-step Reaction Mechanism for Protein Palmitoylation by DHHC Enzymes*♦

DHHC protein acyltransferases (PATs) catalyze the palmitoylation of eukaryotic proteins through an enzymatic mechanism that remains largely unexplored. In this study we have combined genetic and biochemical approaches to examine the molecular mechanism of palmitate transfer of the yeast Ras PAT, which is composed of Erf2 and Erf4. The palmitoylation reaction consists of two steps; they are autopalmitoylation of the enzyme to create a palmitoyl-Erf2 intermediate followed by the transfer of the palmitoyl moiety to the Ras substrate. Palmitoyl-CoA serves as the palmitate donor. To elucidate the kinetic properties of the Erf2·Erf4 PAT, we have developed a coupled enzyme assay that monitors the turnover of the palmitoyl-enzyme species indirectly by measuring the rate of CoASH release. Mutational analysis indicates that the DHHC motif constitutes the catalytic core of the enzyme required for autopalmitoylation and palmitoyl transfer to the Ras2 substrate. In the absence of Ras2, the palmitoyl-Erf2·Erf4 complex undergoes a cycle of hydrolysis and re-palmitoylation, implying that in the presence of palmitoyl-CoA, the complex is autopalmitoylated and competent to transfer palmitate to a protein substrate.

Post-translational modifications extend the functional and regulatory repertoire of proteins. In the case of protein lipidation, the addition of prenyl, palmitoyl, myristoyl, and cholesterol moieties affects subcellular localization and protein trafficking, solubility, and degradation (1,2). Protein S-palmitoylation is the reversible thiol esterification of cysteine residues that is catalyzed by a family of DHHC motif-containing proteins called protein acyltransferases (PATs) 2 or simply palmitoyl transferases (3)(4)(5). Protein palmitoylation was first described over 30 years ago and has been implicated in key cellular processes including cell growth and proliferation, protein trafficking, protein turnover, and vesicle fusion (6,7). The reversible nature of protein palmitoylation makes it well suited to play a regulatory role, including in heterotrimeric and small GTP-binding proteins, receptors, and ion channels (8,9). PAT enzymes share a number of common features including four transmembrane domains and a canonical Asp-His-His-Cys sequence (DHHC motif) that has been hypothesized to be the catalytic center of the enzyme (5,7,10). Despite the central importance of protein palmitoylation in many physiological processes, the molecular mechanism of the palmitoylation reaction remains poorly understood.
There are 7 DHHC PATs in yeast and 23 in mammals (7). Based on in silico topology (TMPRED) computer simulations, the peptide loop containing the signature DHHC residues is predicted to immediately precede a transmembrane domain placing the DHHC motif in juxtaposition to the membrane (11). In addition to the DHHC motif and the transmembrane domains, most but not all PATs contain a cysteine-rich domain (CRD) (7). The exceptions include Akr1, Akr2, Pfa5 (yeast), and DHHC22 (mammals), which lack one or more of the conserved cysteines and histidines that are found in most PATs. The significance of this difference is not clear but may lie in the hypothesis that the canonical CRD forms a zinc finger domain (12,13). However, based on the high degree of conservation in the DHHC region, the sensitivity of the enzyme to neutral hydroxylamine, and the loss of activity when the cysteine residue is mutated, it has been hypothesized that the DHHC tetrapeptide is directly involved in catalysis (5,10). All DHHC enzymes tested to date appear to transfer palmitate via a palmitoyl enzyme intermediate (7,14). Using palmitoyl-CoA as the substrate, the PAT initiates the reaction forming palmitoylated enzyme, a process referred to as autopalmitoylation.
To investigate the enzymatic properties of these proteins, we have chosen the Saccharomyces cerevisiae Ras PAT, Erf2⅐Erf4. Erf2⅐Erf4 is a heterodimeric PAT that palmitoylates yeast Ras2 on Cys-318, adjacent to the farnesylated cysteine, Cys-319. In the absence of protein substrate, PATs undergo autopalmitoylation when incubated with palmitoyl-CoA (7,14). In this study we describe assays capable of measuring the rates of different steps in the palmitoyl transferase reaction. Mutants of the CRD appear to be deficient in the autopalmitoylation reaction, and this explains their inability to palmitoylate Ras2 in vivo. We observe that the palmitoyl-PAT species can undergo hydrolysis, either liberating free palmitate or when present, transferring the palmitoyl moiety to the substrate. Our data suggest that in the absence of Ras2, Erf2⅐Erf4 loses the palmitate to water; however, in the presence of its substrate, the enzyme favors the substrate and transfers palmitate to Ras2.
Plasmid Construction-The construction of pESC-His 6 -Erf2-FLAG-Erf4 has been described previously (17). Deoxyoligonucleotides used to construct ERF2 alleles are listed in Table  1. His 6 -ERF2 alleles were constructed using the QuikChange II site-directed mutagenesis kit (Stratagene) as per the manufacturer's instructions. Isolates produced from the mutagenesis protocol were sequenced to confirm the ERF2 allele changes ( Table 2). Low-copy versions of these alleles lacking the His 6 tag were constructed into B642 (Table 2). Briefly, the mutants were amplified using PCR with ERF2 alleles described above as templates and deoxyoligonucleotides Erf2-F and Erf2-R as primers. The Erf2-F and Erf2-R deoxyoligonucleotides have sequences that overlap with the 5Ј and 3Ј ends of the ERF2 ORF, respectively. The ERF2 alleles were introduced into the B642 through homologous recombination by first deleting part of the ERF2 ORF with NcoI and NruI and then transforming the PCR product and the "gapped" plasmid simultaneously into RJY-1330 and selecting for Trpϩ colonies. Plasmids were rescued (18) and the DNA was sequenced to confirm the presence of the mutations (GeneWiz, South Plainfield, NJ).
Complementation Assay-The in vivo function of the mutants along with the wild type protein was investigated using our previously described complementation assay (3). Briefly, in this assay, cells contain a defective allele of RAS2 that is balanced by an episomal copy of RAS2 linked to URA3. Under these conditions, the loss of ERF2 gene is permissible as long as the cell maintains the RAS2/URA3-based episome. This is detected by their ability or inability of the strain to grow on medium supplemented with 5Ј-fluoroorotic acid (FOA) (21). Cells carrying ERF2 alleles were transformed into RJY1330 and plated on synthetic medium containing glucose and lacking tryptophan. Colonies were inoculated into liquid synthetic medium containing glucose or raffinose (both lacking tryptophan) and grown to an A 600 of between 0.8 and 1.2. The cell density was determined by monitoring absorbance using a spectrophotometer at 600 nm. 1:10 serial dilutions were made starting with 10,000 cells. The cells were spotted onto synthetic medium containing or lacking 5Ј-FOA as presented in the figures.
Protein Purification-Strain RJY1827 was transformed with pESC(His 6 )ERF2-(FLAG)ERF4 and pMA210 and grown in SC(ϪLeuϪHis) medium containing 2% (v/v) ethanol, 2% (v/v) glycerol at 30°C. The cultures were grown to 2 ϫ 10 7 cells/ml and then induced by adding galactose to a final concentration of 4%. Cells were induced for 17h (30°C) and then harvested by centrifugation at 3000 ϫ g for 15 min, the pellet was resuspended in breaking buffer (50 mM Tris pH 8, 500 ml NaCl, 1 mM EDTA, 1 mM DTT, 1ϫ Protease Inhibitor Cocktail (Roche), 8 l/ml PMSF), and the cells were lysed using glass beads (400 -600 mesh, Sigma) for 40 min with 1.5-min pulses. The resulting extract was spun at 3000 ϫ g for 15 min to remove cellular debris and unbroken cells followed by a crude membrane fraction (P100) by centrifugation (100,000 ϫ g) for 1.5 h at 4°C with a Beckman 50.2 Ti rotor. The supernatant was discarded, and the pellet was resuspended in Tris-buffered saline, pH 7.4, with the aid of a Dounce homogenizer. The resulting extract was adjusted to a final concentration of 0.8% dodecylmaltoside, 500 mM NaCl, and 1 mM ␤-mercaptoethanol. To solubilize the membranes, the extract was incubated at 4°C (1.5 h). To aid in purification, urea and imidazole were added to a final concentration of 2.4 M and 1 mM, respectively. Insoluble material was then removed by centrifugation (10,000 ϫ g) for 15 min at 4°C. The supernatant was incubated with nickel-nitrilotriacetic acid resin at 4°C for 1 h. The resin was washed once with Solution W (50 mM Tris⅐HCl, pH 8.5, 0.08% dodecylmaltoside, 5 mM ␤-mercaptoethanol) containing 300 mM NaCl and then twice with Solution W containing 150 mM NaCl. The protein was eluted with 50 mM Tris HCl, pH 8.5, 150 mM NaCl, 0.08% dodecylmaltoside, 5% glycerol, and 250 mM imidazole. Eluates were desalted, and the buffer was changed to SPB (50 mM sodium phosphate buffer, pH 6.8, 10% glycerol) using a column of G-25 resin. Fractions containing His 6 -Erf2/FLAG-Erf4 were pooled to obtain ϳ0.5 mg of purified Ras PAT per liter of culture.
The GST-mCherry-Ras2 substrate used in Erf2⅐Erf4 palmitoylation assays was purified as described earlier (17). Previously, using truncated constructs, we showed that the C-terminal 35 amino acids of Ras2 are sufficient for substrate recognition and catalysis (19). Therefore, we expressed and purified GST-mCherry-Ras2 CT35 as the substrate for in vitro assays with Erf2⅐Erf4.
Coupled PAT Assay-The production of NADH was monitored as a single cuvette platform and a 96-well platform. In the single cuvette platform the production of NADH was monitored using a PTI Double-Double Quantum Master 3 Fluorimeter (340-nm excitation/465-nm emission). The 200-l reaction contained 2 mM ␣-ketoglutarate, 0.25 mM NAD ϩ , 0.2 mM thiamine pyrophosphate (TPP), 3 g of Erf2⅐Erf4, 1 mM EDTA, 1 mM DTT, 30 milliunits of ␣-KDH, and 50 mM sodium phosphate, pH 6.8. The reaction was initiated by the addition of 25 M palmitoyl-CoA and monitored for 5 min at 30°C. The reac- For the 96-well platform, the production of NADH was monitored with a Biotek Mx fluorimeter (Biotek, Winooski, VT) using 340-nm excitation/465-nm emission. The 200-l reaction contained 2 mM 2-oxoglutarate (␣-ketoglutamic acid), 0.25 mM NAD ϩ , 0.2 mM thiamine pyrophosphate, 0.4 -1 g of purified His-Erf2/FLAG-Erf4, 1 mM EDTA, 1 mM dithiothreitol, 32 milliunits 2-oxogluarate dehydrogenase (␣-ketoglutarate dehydrogenase), and 50 mM sodium phosphate, pH 6.8. The reaction was initiated by the addition of different concentrations of palmitoyl-CoA and monitored for 30 min at 30°C. The first 10 min of the reaction was analyzed to determine the initial rates of CoASH release. The PAT specific activity was determined from a standard curve of NADH production with different CoASH amounts. In these reactions CoASH was added to the standard PAT reaction mixture (without Erf2⅐Erf4 complex or palmitoyl-CoA), and the reaction was allowed to proceed to equilibrium before fluorescence was measured (excitation, 340 nm/emission, 465 nm).
To validate the assay, a negative control was used in which the enzyme was boiled. Boiled enzyme produced significantly less fluorescence (Ͻ20%), suggesting no significant contribution from sample heterogeneity or nonspecific fluorescence.
Radioactivity-based PAT Assay-Palmitoylation assays were performed as previously described (17) with the following modifications. The reaction consisted of 50 mM sodium phosphate Buffer, pH 7.4, 1 mM ␤-mercaptoethanol, 10 pmol of purified His-Erf2/FLAG-Erf4, 1 Ci of 30 Ci/mmol (666 nM final concentration) [9,10-3 H]palmitoyl-CoA (American Radiolabeled Chemicals, St. Louis, MO) in a total volume of 50 l with or without 200 pmol of GST-mCherry-Ras2 substrate. For measuring the kinetics of steady-state product formation (palmitoyl-Erf2 or palmitoyl-Ras2), the reaction was allowed to proceed for different time periods ranging from 0 to 60 min at 30°C and stopped using SDS-PAGE gel loading buffer. The palmitoylated proteins were separated using SDS-PAGE, and autoradiography was performed to determine the labeled proteins in the gel.
For performing the kinetics of autopalmitoylation reaction and measuring the initial rates of the consumption of substrates and the release of products, 20 pmol of purified Erf2⅐Erf4 was mixed with 240 pmol of [ 3 H]palmitoyl CoA in the presence of 50 mM sodium phosphate, pH 7.4, with 0.5 mM DTT, and aliquots were taken at various intervals ranging from 5 s to 600 s. The components were separated using thin-layer chromatography on a silica G60 plate using n-butanol/water/acetic acid (50/30/20) as the mobile phase. Running the aliquots on a TLC plate effectively terminated the reaction (data not shown). Radioactivity located at the origin corresponded to [

RESULTS
The DHHC CRD Domain Is Necessary and Sufficient for the Autopalmitoylation of Erf2-The conserved DHHC-CRD domain of PATs encompass 51 residues within the predicted cytoplasmic loop between transmembrane spanning domains 2 and 3. To determine how the DHHC-CRD of Erf2 contributes to autopalmitoylation and Ras2 palmitoyl transfer activity of the Erf2⅐Erf4 PAT, FLAG epitope-tagged Erf2 (either N or C terminus) proteins (Fig. 1A) were expressed and purified with Erf4 as described earlier (17). The C-terminal epitope tag resulted in a protein with lower activity for autopalmitoylation and palmitoyl transfer to Ras2 (Fig. 1B), and therefore, all subsequent experiments were performed with the N-terminal FLAG-tagged protein. The presence or absence of the N-and C-terminal regions had no measurable effect on the autopalmitoylation activity of Erf2⅐Erf4; however, deletion of either region dramatically decreased palmitate transfer to Ras2 in vitro and in vivo (Fig. 1C).
Conserved Residues within the DHHC-CRD Are Required for Autopalmitoylation and Ras2 Transfer Activity-The retention of autopalmitoylation by Erf2 truncation proteins indicated that the catalytic residues for autopalmitoylation were located between amino acid residues 119 and 301. The regions of highest sequence identity included a DPG sequence element and DHHC-CRD domain between TM2 and TM3 as well as a TTXE motif in the C-terminal region after TM3 (Fig. 2). To identify additional residues required for function, 21 mutations were created and tested by the in vivo complementation assay (Fig.  2). Mutations C175S, C178S, R185A, H188A, C189S, C192S, C195S, D200A, H201A, H202A, C203S, C209S, or F218S resulted in loss of function, whereas others (K173A, K/R182A, W205A, and N214S) supported growth in medium containing 5Ј-FOA and palmitoylation-dependent growth in vivo. Not all mutations in consensus sequences produced non-functional alleles of Erf2. The proline in sequence DPG (Asp-Pro-Gly) and glutamate in sequence TTXE (Thr-Thr-Xaa-Glu) were found in all but one of the yeast and human DHHC PATs. However, alanine substitutions at Pro-132 and Glu-286 had no measurable effect on Erf2 function and were not pursued further. Erf2 CRD mutations comprising the atypical zinc finger (CX 2 CX 9 HCX 2 C) between residues Cys-175 and Cys-195, inclusive, resulted in a loss of function phenotype (Fig. 2). Purification of these mutant proteins was not successful, suggesting that the mutant proteins were structurally unstable (data not shown).
Kinetics of Hydrolysis and Turnover of Palmitoylated Erf2⅐Erf4-To study the role of the amino acid residues of the active site, thin layer chromatography was used to monitor the products of the autopalmitoylation reaction over time. Surprisingly, steady-state levels of the palmitoyl-Erf2 intermediate occurred by the earliest time point (5 s), suggesting burst kinetics (Fig. 3). As expected, the production of CoASH also showed an initial burst within the first 5 s and then progressed linearly at a reduced rate (ϳ0.04pmol/min/g of Erf2⅐Erf4). In contrast, palmitate release proceeded linearly throughout the time course at a rate similar to the post-steady-state release of CoASH (ϳ0.04pmol/min/g of Erf2⅐Erf4). To show that there is in fact a buildup to the steady-state level of palmitoyl-Erf2, we performed the reaction at 4°C (Fig. 3A, inset). By slowing down the reaction, we were able to observe the simultaneous increase in the formation of palmitoyl-Erf2 and the production of CoASH. Under these conditions, we were unable to detect the release of palmitate from palmitoyl-Erf2. Therefore, based on the TLC assay, the catalyzed formation of palmitoyl-Erf2, the release of CoASH, and the release of palmitate were consistent with the notion of a burst in the formation of autopalmitoylated Erf2 followed by a slower rate of hydrolysis of the thioester linkage and regeneration of the enzyme. The successive formation of CoASH in the reaction was due to the turnover of the palmitoyl-Erf2 species. The slope of this phase was similar to the rate of the palmitate released, indicating that during the initial burst in activity, the substrate was used for palmitoylating the enzyme. Any further release of palmitate originated from the regeneration of the enzyme.
To determine the number of active sites in the enzyme, the concentration of Erf2⅐Erf4 (10, 20, 30, and 40 pmol) was varied, and the production of CoASH was measured during the linear phase of the reaction (Ͼ25 s) (Fig. 3B). The number of active sites was determined by extrapolating the catalytic rates for CoASH released from each concentration of Erf2⅐Erf4 to time 0. As a control, we performed the same experiment using 40 pmol of the catalytically inactive Erf2C203S/Erf4 (shown at the bottom of the graph). Plotting the pmol of CoASH produced as a function of the pmol of Erf2⅐Erf4 initially added to the reaction resulted in a linear relationship with a slope of 0.65 (Fig. 3B,  inset). Assuming the purified enzyme is 65% active, this is consistent with one active site per Erf2⅐Erf4 dimer.
A Fluorescence-based Assay to Measure the Autopalmitoylation of DHHC PATs-Current methods to measure palmitoylation use [ 3 H]palmitoyl-CoA and monitor the steady-state palmitoylation of the PAT followed by transfer to the respective protein substrate. Problems of sensitivity, cost, and inability to measure the turnover of palmitate on the enzyme have made studies of the autopalmitoylation reaction difficult. To this end we have developed a method to measure the autopalmitoylation reaction continuously by coupling the release of CoASH (Fig. 4A) to the reduction of NAD ϩ to NADH by ␣-ketoglutarate dehydrogenase (␣-KDH) and monitoring the reaction by NADH fluorescence emission at 465 nm (22). When NAD ϩ and ␣-ketoglutarate are present in excess over CoASH, the reaction proceeds with approximately zero-order kinetics (23). This reaction indirectly measures the turnover of the palmitoylated enzyme by monitoring the production of CoASH from palmitoyl-CoA and does not require the addition of the in vivo target substrate. As a proof of principle, we monitored the Erf2⅐Erf4dependent production of NADH continuously. In this assay, the accumulation of NADH is linear with time (Ͻ10 min, data not shown) and is dependent on active Erf2⅐Erf4 enzyme, as is demonstrated by the reduction of the NADH fluorescent signal in the Erf2 (C203)/Erf4 and boiled samples (Fig. 4B). There is a B, specific autopalmitoylation activity of Erf2⅐Erf4 truncation mutants is shown. The specific activity of autopalmitoylation for the Erf2 species was determined using the [ 3 H]palmitoyl-CoA radioactivity assay and separating the products by SDS-PAGE (10% gel). Bands corresponding to auotpalmitoylated Erf2 and palmitoylated Ras2 were excised from the gel, and the amount of radioactivity was determined using En 3 Hance (PerkinElmer Life Sciences) as per the manufacturer's instructions. Black bars denote specific autopalmitoylation activity; gray bars denote specific palmitoylation activity for the Ras substrate. Enzyme activity is normalized to the FLAG:Erf2 control. C, Erf2 truncation mutant complementation assay is shown. Cells from strain RJY1330 harboring the Erf2 truncation mutants shown were spotted as a 1/10 dilution series (starting with 2 ϫ 10 4 cells) on synthetic medium lacking tryptophan (SC-Trp) and supplemented with FOA. The strains were incubated for 3 days at 30°C.
small amount of nonspecific activity in the Erf2 (C203S)/Erf4 sample that is absent from the boiled sample. As we are unable to detect activity from Erf2 (C203S)/Erf4 by other means (5), this small amount of activity most likely represents a contaminant.
We next examined the NADH formation rate catalyzed by the wild type enzyme after reaching steady-state using the coupled assay in a 96-well platform. This apparent rate provides a measure of the release of CoASH and the hydrolysis of the palmitoyl-Erf2 thioester. The specific activity increases with increasing concentrations of palmitoyl-CoA, approaching saturation at a substrate concentration of 90 M. Under the conditions used in the assay, concentrations greater than 90 M appear to be above the critical micelle concentration of palmitoyl-CoA (24) (data not shown).
The autopalmitoylation reaction exhibited pH dependence (Fig. 4C). Fitting the data to obtain the V max and K m at different pH values revealed that at the pH 7.4 the WT enzyme had a V max of 45 pmol/min/g, a K m of 53 M, and a k cat of 2 min Ϫ1 in the absence of its cognate Ras2 substrate. The dependence of V max and the K m and k cat on pH is shown in Table 3 and Fig. 4C, respectively. The presence of a sharp transition between pH values of 6.8 and 7.4 suggests the involvement of titratable groups with a pK a of ϳ7.2 (Fig. 4D). In Fig. 4E, the catalytic efficiency (log (k cat /K m )) is plotted as a function of pH. There is a transition in the data points at a pH of ϳ7.1, which is consistent with the pK a value determined in Fig. 4D. The presence of one ascending slope with a slope of 0.7 Ϯ 0.2 is consistent with the dependence of activity on one titratable group in the enzyme. The dependence does not appear to be due to the pK a values of the titratable phosphates of the CoASH moiety. These pK a values are less than the pH values used in the experiment.
In addition, we measured the rate of CoASH release and palmitate hydrolysis from Erf2 mutants. Erf2 mutants R185A, H201A, C203S, C209S, and F218S were co-expressed with Erf4 and purified, and their activities were monitored using the coupled fluorescence assay (Fig. 5). Surprisingly, rapid autopalmitoylation of C209S and R185A was observed despite the apparent lack of function in the complementation assay (Fig. 2). Table 4 shows the apparent rates calculated for the wild type (WT) enzyme and the different mutants. R185A and C209S showed nearly a 2-fold increase in the rate of release of CoASH and greater than a 4-fold increase for the catalytic efficiencies, respectively, when compared with the wild type enzyme. In contrast to these two residues, F218S showed a decrease in activity, consistent with our observations from the complementation assay (Fig. 2). Although C203S and H201A appear to have activity greater than the boiled sample, a result that most likely represents a contaminant in our enzyme preparations, these mutants possess very little activity using radioactive [ 3 H]palmitoyl-CoA as substrate (5). These particular mutants, therefore, represent the base-line activity measurement for the assay. These results indicate that mutants in the cysteine-rich domain have a direct effect on autopalmitoylation and, therefore, play a key role in the activity of the enzyme.
The Role of DHHC-CRD in the Formation of Autopalmitoylated Erf2⅐Erf4 and Transfer of Palmitate to Ras2-We next examined the role of conserved residues within the DHHC CRD in formation of palmitoyl-Erf2. Wild type and mutant Erf2 proteins were incubated with 50 M [ 3 H]palmitoyl-CoA for 0.5, 10, 30, and 60 min. In wild type enzyme (Fig. 6A), formation of palmitoyl-Erf2 reached a steady state within 30 s. The same was true for the C209S mutant. Mutants H201A, F218S, and R185A, however, did not show the initial burst in activity. H201A appeared to reach a comparable steady-state level as the wild type enzyme but with a delay. The delay in reaching steady state for F218S was greater than 60 min. The steady-state levels of palmitoyl-Erf2 for the R185A and C209S mutants was a fraction of that of the WT enzyme; ϳ10% for R185A and 50% for C209S. Autopalmitoylation of C203S could not be detected in this assay, which was consistent with our previous observations (5). In summary, mutants R185A and C209S appeared to have greater autopalmitoylation activity and lower steady-state levels of palmitoyl-Erf2. Mutant H201A under went delayed autopalmitoylation with a greatly reduced rate of palmitoyl-Erf2 thioester hydrolysis and, therefore, showed little to no turnover of the palmitoyl-enzyme intermediate. F218S showed both delayed autopalmitoylation and reduced steady-state levels of palmitoyl-Erf2.
We next investigated the palmitoylation of Ras2 using the Erf2 mutants R185A, H201A, C203S, C209S, and F218S. [ 3 H]Palmitoyl-CoA, GST-mCherry-Ras2CT35, and Erf2⅐Erf4 complex were combined for 0.5,10, 30, and 60 min. The reaction was terminated using SDS-gel loading buffer, and the products were separated using SDS-PAGE (Fig. 6B). Transfer of palmitate to Ras2 was detected for the wild type protein at the initial time (30s) and increased over time. After 60 min, mutants R185A, C209S, and F218S showed rates similar to WT for transferring palmitate to Ras2. We did not detect transfer using H201A or C203S. Except for the H201A and C203S mutant Erf2 proteins, the mutated enzymes were capable of palmitoylating Ras2. However, the kinetics of Ras2 palmitoylation differed quite significantly than what was observed with autopalmitoylation (Fig. 6A). Although the steady-state levels of autopalmitoylation varied drastically between mutants, these data demonstrated that the association of the Ras2 substrate with the palmitoyl-enzyme had a major influence on palmitate transfer.
Decreased Hydrolysis of Palmitoyl-Erf2 and Enhanced Palmitate Transfer in the Presence of Ras Substrate-We next examined the effect of transfer of palmitate to the substrate on the kinetics of the autopalmitoylation and hydrolysis of palmitoyl-Erf2 using the TLC assay (Fig. 7). The production of CoASH was determined indirectly from the decrease in the amount of [ 3 H]palmitoyl-CoA. It should be noted that it was not possible to independently monitor the formation of palmitoyl-Erf2 or palmitoyl-Ras2 as they both have an R f of 0 and are found at the origin on the TLC plate. In the absence of Ras2 substrate, there is a burst followed by a slower, linear increase of CoASH production. The release of palmitate lacked the burst but followed a linear rate similar to that of CoASH when Ras2 substrate was absent. Interestingly, when Ras2 was present, the production of CoASH was increased, suggesting that the activity of Erf2⅐Erf4 was increased. In addition, the rate of palmitate hydrolysis from palmitoyl-Erf2 decreased from 0.05 pmol/s to 0.01pmol/s.

DISCUSSION
Despite the importance of protein palmitoylation in a variety of cellular processes involved in cancer, cardiovascular disease, infectious disease, and neurological disorders, the enzymatic mechanism of DHHC PATs has remained poorly understood (25). In this study the S. cerevisiae Ras PAT, Erf2⅐Erf4 was used as a model system to elucidate the basic mechanism of palmitoyl transfer. Genetic and biochemical analyses of Erf2 mutations have allowed us to dissect the two steps of the palmitoylation reaction, namely autopalmitoylation and formation of a palmitoyl enzyme intermediate and transfer of palmitate to Ras. The following conclusions can be drawn. First, the highly conserved DHHC motif appears to play a central role in autopalmitoylation of the enzyme and palmitoyl transfer to substrate. The substitution of the DHHC cysteine with serine or alanine abolishes enzyme activity in several published systems (4,5,7,14,26,29). The initial burst in autopalmitoylation is abolished by mutating Cys-203 of the DHHC motif, consistent with it being the side chain thioesterfied during the autopalmitoylation reaction. In contrast, mutating Erf2H201A uncoupled autopalmitoylation from palmitate transfer. There is one catalytic site per enzyme molecule, and autopalmitoylation is regulated by one ionization group, which has a pK a of 7.2. As described in detail below, His-201 is an excellent candidate for the pH-sensitive group. Finally, we have demonstrated that in the absence a protein substrate, the palmitoyl-enzyme thioester is prone to hydrolysis and that when substrate is present, this rate of hydrolysis decreases 90%. The biochemical characterization of mammalian protein acyltransferases has lagged in comparison to the yeast system. This is partly due to the difficulty purifying the mammalian enzymes in sufficient quantities for kinetic analysis. The expression and purification method we developed CoASH amounts were determined indirectly by subtracting the number of palmitoyl-CoA counts from the total counts at each time point as described under "Experimental Procedures." The experiment was also performed while maintaining the reaction temperature at 4°C (inset). B, active site titration analysis was performed at room temperature (24°C) using 10, 20, 30, and 40 pmol of Erf2⅐Erf4. Time points were taken every 10 s for 60 s. As a control for nonspecific activity, 40 pmol of the inactive Erf2 C203S enzyme were assayed (bottom of the main graph). There is a nearly 1 to 1 correlation between the amount of Erf2⅐Erf4 used and the amount of CoASH produced (inset).
in yeast was an essential first step in being able to carry out these experiments. The mechanistic implications of these observations are discussed below.
Novel palmitoylation assays described herein have allowed us to probe the catalytic mechanism of protein palmitoylation. Sequence conservation among DHHC PATs and in vivo functional assays focused attention on the loop between transmembrane 2 and 3 of Erf2 as being necessary and sufficient for autopalmitoylation but not sufficient for transfer of palmitate to Ras. Differences, however, do exist among DHHC PATs. A small subset of DHHC proteins lack the conserved cysteine and histidine residues that constitute the CRD that have led to their annotation as zinc finger proteins. The non-CRD DHHC proteins include Akr1, Akr2, and Pfa5 in yeast and DHHC22 in humans. Mutating the conserved cysteines and histidines residues predicted to bind zinc in Erf2 might destabilize the protein, thus, interfering with purification efforts. If the CRD does bind zinc, it is not required for palmitoyl transferase activity in general. However, the binding of zinc may have an effect on the autopalmitoylation burst seen with Erf2⅐Erf4. Analysis of the non-CRD DHHC enzymes using the same approach that we have used for Erf2⅐Erf4 will provide an answer as to the role of zinc binding on the function of the CRD.
What has emerged from this study is that the PAT reaction occurs in two steps. In the first step palmitoyl-CoA undergoes nucleophilic attack by a deprotonated cysteine of the enzyme to create the palmitoyl-PAT product. We observed that the first step of the reaction appeared to go to completion within a few seconds. Although we have yet to analyze our reaction using stop-flow spectroscopy, we were able to slow the reaction and monitor it in presteady-state conditions. Intriguingly, the production of CoASH and palmitoyl-Erf2 increased immediately, whereas the hydrolysis of the palmitate from palmitoyl-Erf2 was undetectable. Once formed, palmitoyl-PAT either transfers the palmitoyl group to a protein substrate, such as Ras2, or reacts with water (hydrolysis) to release free fatty acid. The real time autopalmitoylation fluorescence assay has proven to be a reliable and sensitive tool for investigating the post steady-state rate of regeneration of the enzyme resulting from the hydrolysis of palmitate from palmitoyl-Erf2. The assay was used to address the chemistry of the autopalmitoylation reaction. The pK a for autopalmitoylation was found to be 7.2. Although the microenvironment can affect the pK a of an amino acid side chain by several units compared with the side chain of the amino acid in solution, the most likely candidate is a cysteine or histidine  Table 3. D, determination of the pK a for the autopalmitoylation reaction by plotting the log of k cat versus pH. k cat values were determined from the data presented in Table 3. The titration curve intersects the x axis at a pH of 7.2. E, graph plotting the multiphasic nature of the log of k cat /K m versus the pH values. The slope of the initial phase is 0.7. At pH 9.2, the activity of Erf2⅐Erf4 is reduced to background values.  (n ϭ 3). The values for the Erf2⅐Erf4 kinetic parameters, V max and K m , are given in Table 4. residue. In addition, we determined that the dependence on activity results from a single titratable group. We hypothesize that to generate the reactive cysteine nucleophile (thiolate anion) (8), as opposed to other nucleophiles (e.g. for example, aspartate (27)), that can attack the incoming palmitate donor, a basic deprotonating agent is required. This group will be sensitive to the pH environment as the efficiency of the reaction will be modulated by the extent of availability of the deprotonating agent. One possible model is that a histidine and cysteine work in concert to produce a nucleophile capable of attacking the incoming palmitoyl-CoA thioester. In this model, His-201 would deprotonate Cys-203 to create the critical nucleophile. Alternatively, the change in pH may deprotonate the cysteine directly to create the nucleophile. At steady state, the rate of the Erf2⅐Erf4 autopalmitoylation reaction was relatively slow in the pH range from 5.7 to 6.8. However, as the pH increased to 7.4, the activity of the enzyme significantly increases and was maintained through pH 8.6. The activity of the enzyme was undetectable above pH 9.2 (data not shown). This increase in activity of Erf2⅐Erf4 appeared to be due to an increase in hydrolysis rate. In addition, the apparent K m of the substrate is reduced with increasing pH, suggesting greater affinity of the palmitoyl-CoA for the enzyme within this pH range. It is conceivable that an intermediate of such nature requires association with various charged residues for stability. This phenomenon has also been observed with the citrate synthases where histidines stabilize the intermediate (27). In DHHC-CRD-containing PATs such as Erf2⅐Erf4, we speculate that this role is fulfilled by many of the conserved histidines in the CRD and that the optimal pH to produce the highest affinity is within the pH range of 7.4 -8.6. Structural studies of the active site will greatly aid in understanding the basis of the hydrolytic activity, but given that DHHC-PATs are integral membrane proteins, attempts to date to crystallize PATs have been unsuccessful.
The rapid burst formation of palmitoyl-Erf2 was not necessarily expected and implies that in the absence of protein substrate, enzyme exists in a continuously palmitoylated state. This can be altered by mutation. For example, mutations in Erf2 residues C209S and R185A had reduced steady-state levels of palmitoyl-Erf2 that appear to correlate with the function of the enzyme in vivo as determined by the growth characteristics of the yeast cells. However, with respect to CoASH production, our results indicated that both variants have higher initial    hydrolysis rates than the wild type enzyme, although they lack or are severely diminished for in vivo function. The reduction in the steady-state amounts of palmitoyl-Erf2 during autopalmitoylation for mutants C209S and R185A is also consistent with increased hydrolysis rates. One possible explanation is that these mutations in some way expose the active site, thereby increasing hydrolysis. In contrast, other mutants, such as H201A, F218S, and C203S, lacked the initial burst in palmitoyl-Erf2 and exhibited reduced or no hydrolysis. Finally, the presence of Ras2 also reduced the rate of hydrolysis of palmitoyl-Erf2. The Ras2 substrate might (a) compete with water at the same site on Erf2⅐Erf4, presumably the active site and (b) conform the enzyme to a more active state, possibly by stabilizing the enzyme. Consistent with this notion, Erf2 mutants R185A, C209S, and F218S, which demonstrate less autopalmitoylation activity based on their steady-state levels of palmitoyl-Erf2, showed similar rates of palmitate transfer to Ras2 as wild type Erf2⅐Erf4 (Fig. 6B). The formation of a stable palmitoyl-enzyme in palmitoylation mediated by PAT enzymes makes them different from the serine palmitoyl transferases, where the enzyme positions the serine and the palmitoyl-CoA to form the final product (28). With respect to multisubstrate binding, the PATs do resemble carnitine palmitoyl transferase II more so than the serine palmitoyl transferases (29). In carnitine palmitoyl transferases, the palmitoyl-CoA binds first, followed by the carnitine. In the case of Erf2⅐Erf4, autopalmitoylation is the first event followed by the binding and the transfer to the second substrate, Ras2, although we cannot rule out that the enzyme can bind the Ras2 substrate in the absence of autopalmitoylation. An active site histidine is also found in carnitine palmitoyl transferase-I (CPT-I) and CPT-II, but no acyl-enzyme intermediates are formed.
The transfer of the palmitate to Ras2 may also involve the His/Cys catalytic core. Being a slower, non-hydrolyzable mutant of Erf2, we propose that His-201 acts as the primary nucleophile generator in transfer. Even the addition of imidazole in trans (data not shown) could not rescue the loss of transfer in H201A, suggesting that this mutation is non-reversible. We speculate that the H201A mutation alters the active site in such a way that water or a deprotonated cysteine cannot attack, as evident from no hydrolysis, and that Ras2 cannot participate in the transfer. Along the same lines, an acid-base catalysis scheme would be employed whereby His-201 acts as the base to extract a proton from Cys-318 of the Ras2 C terminus. This would produce an acidic thiolate anion on Ras2, which could act as a nucleophile on the palmitoyl-enzyme thioester. Given two His/Cys dyads, the association of the substrate, Ras2, may be required to "neutralize" the His-201-containing dyad, so that it cannot participate or interfere with the formation of palmitoyl-Erf2. In support of this notion, even if some of the mutants of Erf2⅐Erf4 including R185A, C209S, and F218S autopalmitoylated with lower steady-state levels of palmitoyl-Erf2 or with a timing delay, they could still transfer to Ras2 to a relatively similar extent as the wild type protein. Clearly, there is more work needed to fully understand the molecular mechanism of these enzymes.
With the linking of molecular diseases to DHHC family members, the in vivo ramifications of these data are just now coming to light. For example, the mammalian DHHC9 protein was identified to possess the highest degree of homology to Erf2. Like Erf2, DHHC9 also appears to have four transmembrane spanning domains. It also requires an accessory protein, Gcp16, to form a functional Ras PAT. The DHHC9⅐Gcp16 complex palmitoylates farnesylated H-ras and N-ras as substrates but cannot utilize N-terminal-palmitoylated substrates, like GAP-43 (30). Mutations in the gene for DHHC9 are the first instance of a defect in a posttranslational modification enzyme causing X-linked mental retardation. For example, a frameshift and splice-site mutation as well as mutations in the highly conserved residues Arg-148 and Pro-150 of DHHC9 have been identified in 4 of 250 families with X-linked mental retardation (31). Presumably these mutations result in decreased palmitoylation, but the molecular mechanism remains to be determined. Using assays described in this study, autopalmitoylation, hydrolysis, and phospho-transfer all contribute to the function of DHHC PAT, and a better understanding of the underlying mechanisms will provide insights into diseases involving palmitoylation.