ACBD6 protein controls acyl chain availability and specificity of the N-myristoylation modification of proteins[S]

Members of the human acyl-CoA binding domain-containing (ACBD) family regulate processes as diverse as viral replication, stem-cell self-renewal, organelle organization, and protein acylation. These functions are defined by nonconserved motifs present downstream of the ACBD. The human ankyrin-repeat-containing ACBD6 protein supports the reaction catalyzed by the human and Plasmodium N-myristoyltransferase (NMT) enzymes. Likewise, the newly identified Plasmodium ACBD6 homologue regulates the activity of the NMT enzymes. The relatively low abundance of myristoyl-CoA in the cell limits myristoylation. Binding of myristoyl-CoA to NMT is competed by more abundant acyl-CoA species such as palmitoyl-CoA. ACBD6 also protects the Plasmodium NMT enzyme from lauryl-CoA and forces the utilization of the myristoyl-CoA substrate. The phosphorylation of two serine residues of the acyl-CoA binding domain of human ACBD6 improves ligand binding capacity, prevents competition by unbound acyl-CoAs, and further enhances the activity of NMT. Thus, ACBD6 proteins promote N-myristoylation in mammalian cells and in one of their intracellular parasites under unfavorable substrate-limiting conditions.

ACBD6 controls protein N-myristoylation 625 bind to the N-terminal acyl-CoA binding (ACB) domain, and various motifs present at the carboxyl terminal of the different members of this family appear to define independent and nonredundant physiological functions of the ACBD proteins. The requirement and role of the acyl-CoA ligand bound to the ACB domain in the interaction with other proteins is poorly understood. The phosphorylation of two serine residues of -helix 4 of the ACB domain of ACBD6 was detected in vivo, but the effect on its activity has not been determined (39,40).
We previously showed that ACBD6 stimulated the activity of NMT2 and prevented the competition of the NMT reaction by C 16 -CoA (26). An interaction between ACBD6 and NMT2 is required, and ligand binding to ACBD6 further enhanced its stimulatory effect on NMT2. Mutants of ACBD6 deficient in ligand binding did not stimulate the activity of NMT2. They also could not protect this enzyme from the competitor C 16 -CoA. In the NMT2/ACBD6 complex, the acyl-CoA carrier appears to prevent the access of C 16 -CoA and allows C 14 -CoA to reach the acyl-transferase binding site in the presence of more abundant acyl-CoA. ACBD6 also interacts with NMT1 and likely regulates its activity.
The effect of the phosphorylation of Ser 106 and Ser 108 on the acyl-CoA binding property of ACBD6 and on the myristoyltransferase reaction was determined. We established that human ACBD6 (hACBD6) enhanced the activity of the malaria parasite Plasmodium falciparum NMT (PfNMT) enzyme. Although C 12 -CoA was a stronger competitor of C 14 -CoA than C 16 -CoA for the parasite myristoyltransferase enzyme, hACBD6 also protected PfNMT from inhibition by the shorter acyl-CoA species. An ankyrin-repeat-containing acyl-CoA binding protein of Plasmodium, PfACBD6, was identified. Like the human homologue, PfACBD6 stimulates the activity of PfNMT. We propose that an essential role for ACBD6 proteins is to maintain substrate availability for the myristoyltransferase reaction and to provide specificity for C 14 -CoA in the presence of competing acyl chains that are +2 or 2 carbon atoms different in length.

Materials
O-Phospho-l-serine, acyl-CoAs, and fatty acids were from Sigma-Aldrich. All compounds used were reagent grade. The peptide GLYVSRLF (C-terminal amide; molecular mass of 953.16) was synthesized by YenZym Antibodies, LLC.
Cloning and site-directed mutagenesis P. falciparum 3D7 NMT (1.2 kbp; gene ID: 811708) and ACBD6 (1.1 kbp; gene ID: 810744) cDNAs were cloned by RT-PCR. Total RNA was isolated from a frozen parasite sample (gift of Elizabeth S. Egan) with the PureLink RNA Mini Kit (Invitrogen). PfNMT was cloned in the pET28 vector (Novagen) with a hexahistidine tag at the N-terminal end. Using the same cloning strategy for PfACBD6 resulted in abundant unfinished translation products. A predicted transmembrane-spanning segment at the N-terminal end of the protein was removed, and a truncated construct (Met1 to Leu23) carrying a hexahistidine tag at the carboxyl end was successfully produced. Site-directed mutagenesis experiments were performed with the QuikChange Lightning Site-Directed Mutagenesis Kit (Agilent Technologies) according to the manufacturer's instructions. Primers were designed with the QuikChange Primer Design Program. The presence of the intended nucleotide change(s) and the absence of unwarranted mutations were verified by full-length sequencing of the constructs. ACBD6, N-terminally tagged with GFP, mutants, and truncated forms, were made in the pAcGFP1-C1 vector (Clontech Laboratories, Inc.). To produce the ACBD6 forms phosphorylated on Ser 106 and Ser 108 , the codons AGC 106 and AGC 108 were changed to the stop codon TAG. The constructs were cotransformed with pKW2.EF.Sep (41) into the Escherichia coli strain BL21(DE3)serB (Addgene catalog no. 34929) (41,42). For mammalian two-hybrid experiments, human NMT2 (hNMT2) and PfNMT were cloned into the pBIND vector, and hACBD6 was cloned into the pACT vector (Check-Mate Mammalian Two-Hybrid System; Promega), as previously described (26). HEK293 cells grown in 48-well plates were cotransfected with 200 ng of the constructs and the reporter pLuc at a pACT-pBIND-pLUC ratio of 2.25:0.25:0.5. Measurements were performed after 48 h of transfection with the Dual-Glo Luciferase Assay System (Promega).

Protein expression and purification
Human ACBD1, ACBD3, ACBD6, and NMT2 were produced as previously described (26)(27)(28)43). Human ACBD4 and ACBD5 were cloned in pETM41 and produced as maltose binding protein fusion forms (gift of Joseph Costello) (19) in RosettaDE3 cells (Novagen). The production of ACBD6 forms phosphorylated on Ser 106 , Ser 108 , or on both of these serine residues was performed in E. coli grown in LB medium supplemented with 2 mM phosphoserine (41,42). PfNMT and PfACBD6 were expressed in RosettaDE3. Although a large portion of PfNMT formed aggregates, the addition of 1% CHAPS during lysis of the cells and to the soluble fraction throughout the purification procedure was successful in maintaining PfNMT in solution. PfACBD6 was maintained in solution by the addition of 1% Triton X-100 in all buffers during the extraction and purification steps. Human ACBD1, ACBD3, ACBD6, NMT2, PfNMT, and PfACBD6 were purified by affinity metal chromatography. Human ACBD4 and ACBD5 were purified with amylose resin (New England BioLabs, Inc.). Purified ACBD1-6 proteins were stored at 80°C in 50 mM Tris-HCl (pH 8.0), 0.1 M NaCl, 5 mM EDTA, and 10% glycerol (v/v). Human NMT2 was stored in the same buffer in the presence of 0.2% Triton X-100. PfNMT was stored in 50 mM Tris-HCl (pH 8.0), 0.1 M NaCl, 5 mM EDTA, 1% CHAPS, and 20% glycerol to prevent the loss of activity. PfACBD6 was stored in 50 mM Tris-HCl (pH 8.0), 0.1 M NaCl, 5 mM EDTA, 10% glycerol, and 1% Triton X-100. 2, and 4 M ACBD6 in 500 l. The purified complex was then exposed to 2.5 M C 24 -CoA. Following incubation for 20 min at 37°C, ACBD6 was then pulled down a second time with 60 l NTA 50% slurry and washed as above. The amount of [ 14 C]C 18:1 -CoA left in the ACBD6 bound-resin fraction was quantified with a scintillation counter.

N-Myristoyltransferase activity measurements
Real-time measurements of released CoA from acyl-CoAs by purified NMT forms were performed at 30°C in a Cary Varian Cary 50 UV-Vis spectrophotometer in the presence of 200 M Ellman's reagent as previously described (26). Detection and quantification of the formation of the acyl peptide in the presence of acyl-CoAs and the peptide GLYVSRLF was performed by separation on a Luna 5u C18 100A HPLC column (250 × 4.6 mm; Phenomenex) and monitoring of the absorbance of the tyrosine residues at 274 nm. Reactions were performed in 20 mM sodium phosphate (pH 8.0), 1 mM EDTA, 1% CHAPS, and 200 M peptide with concentrations of C 14 -CoA from 5 to 50 M. The volume of the reactions, the concentrations of C 14 -CoA and competing acyl-CoA, the NMT enzyme, and the ACBD proteins are indicated in the legend of each figure. Reactions were performed at 37°C and were stopped at various times by adding 1.08 vol ice-cold methanol/TFA 8%. Samples were vortexed and precipitated on ice for 20 min. Proteins were removed by centrifugation at 8,000 g for 10 min at 4°C. The supernatants containing the peptide and acyl peptide were collected and dried down under vacuum overnight at 20°C. The pellets were suspended in 50% methanol/TFA 4% (v/v; from 30 to 80 l). A 10 l sample was injected onto the C18 column, and separations were performed as described previously (10,44,45). An acetonitrile (ACN) gradient was generated in the presence of 0.1% TFA and HPLC-grade water. Samples were loaded onto the column equilibrated with 10% ACN/TFA 0.1%. The ACN concentration was increased to 68% in 30 min at a flow rate of 1 ml per minute and to 100% in 1 min at 1.5 ml/min. After 3 min at 100% ACN/TFA 0.1%, the ACN concentration was decreased to 10%, and the column was equilibrated for 10 min at 1 ml/min. The percentage of product formed, detected by absorbance of the peptide at 274 nm, is reported as the relative volume of the myristoyl peptide peak (34.5 min) relative to the sum of the volume of the peptide peak (18 min) and of the acyl peptide detected in each injected sample. In one experiment, the reaction was performed in the absence of added C 14 -CoA but in the presence of the bound C 14 -CoA/ABCD6 complex. ACBD6 (20 M) was incubated with 100 M C 14 -CoA for 20 min at 37°C. The bound complex was pulled down as described above, and the 50 l eluate was incubated with 60 g PfNMT and 200 M peptide for 2 h at 37°C in 1 ml buffer. A control reaction was performed with the eluate obtained from the pull down of C 14 -CoA incubated in the absence of ACBD6. Calculations and statistical analysis were performed with GraphPad Prism 7.

Cell culture and transfection
HeLa229 cells, obtained from ATCC (CCL-2.1), were maintained in MEM  (Invitrogen) containing 10% FBS and 2 mM glutamine. For microscopy studies, cells were grown and transfected with Turbofect (Thermo Fisher Scientific) on 12 mm round coverslips (Electron Microscopy Sciences) in 24-well plates (46). Imaging was performed with a Keyence microscope equipped with a 40× objective. The two aromatic residues Phe46 (-helix l) and Tyrl 14 (-helix 4), whose substitution to alanine significantly reduced the binding capacity of hACBD6 (27) and the two phosphorylated residues Ser 106 and Ser 108 are shown. Note that the two aromatic residues protrude in the inside of the pocket involved in binding of the acyl chain.

Ser
106 and Ser 108 of hACBD6 are not essential for ligand binding Two serine residues located in the acyl-CoA binding domain of hACBD6, Ser 106 and Ser 108 , are phosphorylated in vivo (26,39,40). Ser 106 , near the fourth  helix (H4; S108 to K120), is present in some ACBD forms of humans and plants but not in the Plasmodium proteins, including the homologue PfACBD6 (Fig. 1A, B). Ser 108 , at the beginning of H4, is not conserved, but most ACBD proteins carry a glutamate (Glu) residue at this position. Glu can structurally mimic the presence of a phosphoserine, which might indicate that the activity of unphosphorylated ACBD6 could be different from phospho-ACBD6. The acyl chain of the acyl-CoA ligand is bound in the cavity formed by H1, H2, and H4, and the CoA head group faces H2 in the proximity of H3 (Fig. 1C). Site-directed mutagenesis of conserved residues in the four  helices of ACBD6 identified two aromatic residues, Phe46 of H1 and Tyr114 of H4, whose substitution to alanine had the most influence in acyl-CoA binding (27). Based on the structure, these two residues protrude in the acyl-chain binding cavity, and Ser 106 and Ser 108 are located at the top of the pocket. The substitution of the two serine residues to alanine generated ACBD6 forms with similar binding affinity compared with the native ACBD6 ( Fig. 2A). However, a significant decrease in binding capacity was detected, which indicates that although neither Ser 106 nor Ser 108 are essential, they are implicated in binding. The substitution and truncation of the serine residues did not appear to affect the cellular location of GFP-tagged forms (Fig. 2B). Deleting an entire  helix, H4, also had no apparent effect. Thus, the phosphorylation of Ser 106 and Ser 108 , as well as ligand binding, do not appear to be essential for the cytosolic and nuclear localization of hACBD6 in the cell.

Serine phosphorylation enhances ligand binding capacity of hACBD6
Phospho-Ser 106 (P-Ser 106 ), P-Ser 108 , and P-Ser 106 /P-Ser 108 ACBD6 forms were produced in E. coli and purified (see Materials and Methods). The binding activity of the phosphorylated forms was determined by competition experiments in the presence of radiolabeled [ 14 C]C 18:1 -CoA (i.e., ligand) and unlabeled acyl-CoAs (i.e., competitors). Competition depends on the binding affinity of the protein for its ligand and competitor. As expected, the decreased binding of the radiolabeled acyl-CoA was a function of the relative concentration of ligand to competitor at a given protein concentration (Fig. 3). At ligand concentrations close to the observed K d of 10 M ( Fig. 2A), the binding of the ligand was sensitive to the presence of the competitor, and the addition of acyl-CoA of various chain lengths resulted in lower binding values (Fig. 3A). Competition was observed with assays performed with the unphosphorylated form hACBD6 and the double-mutant Ser 106/ Ser 108 -Ala. However, with the double-phosphorylated P-Ser 106 /P-Ser 108 form, binding of the ligand [ 14 C]C 18:1 -CoA was higher in the presence of the competitor (Fig. 3A). This phenomenon was observed for all five acyl-CoAs tested, even when unlabeled C 18:1 -CoA was added to [ 14 C]C 18:1 -CoA. In addition, each of the single phosphorylated forms, P-Ser 106 and P-Ser 108 , also showed increased ligand binding in the presence of competitors (Fig.3B). To render ligand binding competition of P-Ser 106 /P-Ser 108 dependent on the total concentration of acyl-CoAs present in the reaction, titration experiments were performed in the presence of a fixed concentration of the competitor, resulting in the strongest effect on binding, C 24 -CoA (Fig.3A), and with increasing concentrations of the ligand [ 14 C]C 18:1 -CoA. Limited amounts of protein were also used, and under such conditions no binding could be detected at the lowest ligand concentrations in the absence of the competitor (Fig. 3C,   Fig. 2. Phosphorylated ACBD6 serine residues are not essential. Ser 106 and Ser 108 of hACBD6 were changed to alanine residues to generate the single Ser 106 -A, Ser 108 -A, and double Ser 106/108 -A constructs. A: Binding activity was measured with 2 M protein and increasing concentrations of [ 14 C]C 18:1 -CoA (0-20 M). Error bars represent the standard deviations of four measurements. The binding parameters K d and B max are summarized in the graph bar on the right. B: The GFPtagged constructs of hACBD6, the double-mutant Ser 106/108 -A, a truncated form of -helix4 (H4), and a form truncated from Ser 105 to Ser 108 (Ser l05 to Ser 108 ) were transfected in HeLa cells. Cells were fixed after 24 h, and the nuclei were stained with the Hoechst dye (blue). Images of the GFP (lower) and merged signal of GFP with Hoechst staining (upper) are shown. Note that images of various GFP-positive cells detected on the slides were manipulated and combined to generate the panel. inset). However, the addition of the competitor increased binding of the ligand from the lowest concentration (0.5 M) to concentrations closest to the K d (Fig. 3C). Competition did occur when the concentration of ligands reached a K d of 10 M. To a lesser degree, this behavior was also observed with the unphosphorylated form under the limiting ligand concentration of 2.5 M (Fig. 3D).

Acyltransferase activity of NMT enzymes is stimulated by hACBD6
The interaction of NMT2 with hACBD6 enhances the myristoyl-CoA esterase (26) and myristoyltransferase activity of the enzyme (Fig. 4). hACBD6 stimulation of myristoyl peptide formation occurs even under very limiting conditions (Fig. 4B) and in the presence of the competitor palmitoyl-CoA (C 16 -CoA) (Fig.4C). hACBD6 increases both the rate and yield of production of the acyl peptide and protects binding of the substrate myristoyl-CoA (C 14 -CoA) to NMT from competition. To determine whether the role of hACBD6 was limited to the human myristoyltransferase enzyme, the activity of PfNMT was determined (Fig. 5). As observed for NMT2, hACBD6 stimulates PfNMT activity, and the increased formation of the myristoyl peptide in its presence was greater under substrate-limiting conditions (up to 40-fold) (Fig. 5C, D). The interaction was significantly weaker than with NMT2, but an hACBD6/PfNMT complex appears to have formed in the cells (supplemental Fig. S1A). An hACBD6 form with truncated ankyrin-repeat motifs (ACBD6C-ter) did not associate with either PfNMT or NMT2. A mutant form affected in ligand binding and carrying the 4 substitution in the ACB domain (FFKY-to-A) (27) also fails to interact with PfNMT. This mutant protein interacted with hNMT2, suggesting that complex formation with the Plasmodium enzyme might require the acyl-CoA-bound form of hACBD6.

ACBD6 compensates for lack of NMT specificity for the C 14 acyl chain
The abundant C 16 -CoA is a competitor of the myristoyltransferase reaction. hNMT2 cannot prevent binding of an acyl chain two carbon atoms longer than C 14 -CoA, but the palmitoyl chain is not a substrate of the acyl-transferase reaction, and the enzyme cannot release it (4-6, 10). hACBD6 protects hNMT2 from C 16 -CoA (Fig. 4C) (26). Although myristoyl peptide formation by PfNMT was also reduced in the presence of C 16 -CoA, the -2 carbon species lauroyl-CoA (C 12 -CoA) was far more inhibitory (Fig. 6A). Under a low C 14 -CoA concentration (10 M), a 2-fold C 12 -CoA excess reduced the production of the myristoyl peptide by half, whereas a 12-fold excess of C 16 -CoA was required (Fig. 6B, C). Measurements of the esterase activity of the enzyme confirmed that the two competitors bind and were processed by PfNMT (supplemental Fig. S1B). The presence of hACBD6 in the reaction protected the enzyme from the u sage of an acyl chain other than C 14 . hACBD6 successfully relieved the inhibition by C 16 -CoA and further increased product formation (Fig. 6B). C 14 -CoA processing in the presence of C 12 -CoA was also increased by hACBD6 (Fig. 6C).
A new acyl peptide peak was detected in the presence of C 12 -CoA, and its formation increased as the levels of the C 14 peptide peak dropped (Fig. 6A, Fig. 7A). The formation of the lauroyl peptide was differently affected by hACBD6 depending on the presence of the correct substrate C 14 -CoA. In the absence of C 14 -CoA, hACBD6 could not prevent PfNMT from processing C 12 -CoA, and the reaction was nearly independent of its presence (Fig. 7B). In the presence of C 14 -CoA, a 2-fold concentration excess of C 12 -CoA resulted in a 2-fold reduction of myristoyl peptide, but in  the presence of hACBD6, a 4-fold excess of C 12 -CoA was required to achieve a 50% reduction (Fig.7C). In the presence of hABCD6, the increased preference of PfNMT for the C 14 chain over the C 12 chain was also confirmed by the decrease in C 12 processing when the enzyme had the choice of acyl chains (Fig. 7C). Although when challenged with nonphysiological conditions (high concentration of C 12 -CoA in the absence of C 14 -CoA), hACBD6 had little to no detectable effect, it could slow the usage of the C 12 chain by the PfNMT enzyme in a mixture containing C 14 -CoA. These data established that ACBD6 can provide substrate specificity for the myristoyltransferase reaction and prevent the enzyme from processing the "wrong" but more abundant acyl-chain species under conditions that are likely encountered in vivo.

Plasmodium ankyrin-containing ACBD6 protein stimulates PfNMT activity
A search of PlasmoDB, a genome database for Plasmodium (plasmodb.org), retrieved five genes encoding putative acyl-CoA binding proteins (Fig. 8A). Three genes produced the shortest of the forms, homologous to the human ACBD1 (DBI) protein. The other two genes carry different motifs at the extended carboxyl-terminal end. PfACBP1 carries two conserved motifs implicated in export through the parasite membrane and localization in the erythrocyte: Plasmodium helical interspersed subtelomeric and Plasmodium ring-infected erythrocyte antigen N-terminal (47). On the basis of the presence of two ankyrin-repeat motifs downstream of the ACB domain, the unnamed product of the fifth gene was referred to as PfACBD6. A transmembrane-spanning segment motif was predicted at the N-terminal end and had to be removed for successful expression in E. coli (see Materials and Methods). A 60residue-long motif, rich in lysine (12) and asparagine (26), is inserted between the second and third -helix of the ACB domain (Fig. 8A). These poly-KN tracks are often found in Plasmodium proteins and reflect the AT-rich genome of the parasite (48,49).
PfACBD6 is a functional protein and binds acyl-CoA (Fig. 8B, C). In addition, it stimulates the NMT activity of PfNMT and counters inhibition by C 16 -CoA (Fig. 8D, supplemental Fig. 1B). Thus, the ankyrin-repeat containing acyl-CoA binding proteins of humans and Plasmodium regulate the NMT reaction.

Myristoyl-CoA bound to hACBD6 is a substrate of NMT
The ability of different members of the human acyl-CoA binding protein family to stimulate the myristoyltransferase reaction was determined. Of the five ACBD forms tested, hACBD6 was the most efficient in enhancing the production of the myristoyl peptide by PfNMT (Fig. 9A, B). ACBD1, ACBD3, ACBD4, and ACBD5 were able to stimulate PfNMT but to levels similar or lower than that of the ACBD6 mutant altered in acyl-CoA binding (FFKY-A form). The preference for the ACBD protein that interacts with the enzyme confirmed that complex formation improves Fig. 6. hACBD6 protects PfNMT from acyl-CoA competition. Measurements of the formation of the myristoyl peptide were performed in the presence of 10 µM C 14 -CoA and increasing concentrations of C 16 -CoA (closed symbols) and C 12 -CoA (open symbols) (A). Reactions were performed in 120 µl at 37°C for 20 min with 400 nM PfNMT. Activity values obtained with C 16 -CoA and C 14 -CoA or with C 12 -CoA and C 14 -CoA in the absence or presence of 10 µM hACBD6 are shown in panel B and panel C, respectively. Values obtained in the presence of the competitors are presented relative to values obtained from the reaction performed in their absence. Note that PfNMT is more sensitive to competition by C 12 -CoA than by C 16 -CoA. Error bars in the three plots represent the standard deviations of values obtained from three reactions. *P < 0.05 and **P < 0.005. the ability of ACBD6 to enhance the myristoyltransferase reaction. The role of the binding activity of hACBD6 in the mechanism leading to increased NMT activity was investigated. The substrate C 14 -CoA was incubated with hACBD6, and the C 14 -CoA-bound hACBD6 complex was isolated from the unbound C 14 -CoA left in the reaction. The purified C 14 -CoA/ACBD6 was then fed to PfNMT in a buffer mixture lacking C 14 -CoA. The formation of the myristoyl peptide under such reaction conditions established that the C 14 -CoA bound to hACBD6 was available to the NMT (Fig. 9C). The isolation procedure was also performed with C 14 -CoA in the absence of hACBD6, and no product was formed following the incubation with the enzyme (Fig. 9C).
The ability to deliver and transfer the C 14 -CoA substrate from the acyl-CoA carrier to NMT allows the myristoyltransferase reaction to proceed efficiently, even at concentrations below the binding affinity of the enzyme. This would account for the greater stimulatory effect of hABCD6 observed under substrate-limiting concentrations compared with saturating conditions (Fig. 5D).

Phosphorylation-dependent stimulation of myristoylation
To further investigate the ability of hACBD6 to increase the availability of the substrate to the NMT enzyme, displacement experiments of ACBD6-bound ligand by acyl-CoA competitors were performed. The complex hACBD6/ [ 14 C]C 18:1 -CoA was isolated, and the amount of [ 14 C]C 18:1 -CoA still bound to the protein following incubation with increasing concentrations of the competitor C 24 -CoA was quantified (Fig. 10A). Compared with the hACBD6 form, phosphorylation of Ser 106 and Ser 108 residues protected the bound ligand, and a higher concentration of competitor was required to achieve 50% displacement. This observation was confirmed when increasing amounts of the ACBD6ligand complex were exposed to a fixed concentration of competitor (Fig. 10B). Although the removal of the two serine residues did not influence the dissociation of the ligand, their phosphorylation resulted in significant resistance of ligand bound to the P-Ser 106 /P-Ser 108 ACBD6 form (Fig. 10C). The phosphorylated forms also further increased the activity of the PfNMT enzyme compared with ACBD6 (Fig. 11). Substitution of the two serines to glutamate residues (Ser 106 /Ser 108 -Glu) did not appear to produce a form that was as active as the phosphorylated serine forms.
Thus, the improved binding capacity of the phosphorylated hACBD6 form enhanced its ability to hold the bound ligand in a mixture of competing acyl-CoAs under similar conditions encountered in vivo. This property of hACBD6 gives C 14 -CoA a competitive advantage in the presence of more abundant acyl-CoAs and provides the NMT enzyme with the correct substrate in the unfavorable cellular environment.

DISCUSSION
The NMT enzymes lack selectivity for myristoyl-CoA and display a binding affinity (600 nM) that is significantly higher than the cellular concentration of this acyl-CoA, ranging from 5 to 200 nM. These findings imply that mechanisms exist to support the NMT reactions in the cell (2,3,50,51). The use of a relatively low abundant acyl chain under those unfavorable cellular conditions strongly suggests that the covalent linkage of a myristate to an N-terminal glycine residue confers properties to the acylated proteins that the more abundant palmitate chain cannot provide. In fact, some proteins are palmitoylated and myristoylated, on N-terminal cysteine and glycine, respectively, which indicate that these two acyl modifications are not equivalent (52). In vitro, acyl-CoA species that are either two carbon atoms longer or shorter (C 12 -C 16 ) bind with similar affinity than C 14 (0.6-1.4 M) and thus can compete for binding to NMT (10). The transfer of acyl chains from the enzyme to the protein other than C 14 is either blocked or very slow, which favors the addition of a myristate. However, occupation of the binding site by these acyl-CoA species would lead to the progressive inactivation of the enzyme because they could not be released once bound. The acyl-chain species bound to the enzyme also affect the interaction of NMT with the target proteins. The binding of acyl chains other than C 14 can significantly increase the K m binding affinity of NMT for the protein substrate (10). It has been established that, in vivo, the enzyme can preferentially transfer species other than C 14:0 . In the retina, the unsaturated C 14:1 chain is the major species detected in myristoylated proteins. In addition, the polyunsaturated C 14:2 Fig. 9. C 14 -CoA bound to hACBD6 is a substrate for the PfNMT enzyme. A: Measurements of the formation of the myristoyl peptide were performed in the absence or presence of 20 µM human ACBD1, ACBD3, ACBD4, ACBD5, and ACBD6 and the ACBD6 mutant FKKY-to-A. FKKY-to-A carries a substitution of the residues Phe 46 , Lys 73 , Lys 95 , and Tyr 114 to alanine and displayed low binding capacity (27). The chromatogram traces obtained at 274 nm at the time of the elution of the C 14 -peptide (blue peak; 34.5 min) from the C18 column are shown. Reactions were performed in 120 µl at 37°C for 2 h with 700 nM PfNMT in the presence of 10 µM C 14 -CoA. B: The amount of the myristoyl peptide obtained in the experiments shown in panel A was quantified. Error bars represent the standard deviations of at least three reactions. C: hACBD6 (20 µM) was incubated with 100 µM C 14 -CoA for 20 min at 37°C and then purified to remove the unbound C 14 -CoA. The hACBD6/ C 14 -CoA complex was incubated with 1.25 µM PfNMT and 200 µM peptide in a 1 ml reaction for 2 h at 37°C (bottom chromatogram). A control reaction was performed with the pull-down sample obtained from C 14 -CoA incubated in the absence of hACBD6 (top). The chromatogram traces obtained at 274 nm are shown, and the C 14 -peptide is shown as a blue peak. and the shorter-chain C 12 , expected to have a much slower transfer rate than C 14:0 , are also detected in retinal proteins at similar levels compared to C 14 (14). Thus, it appears that the NMT enzyme itself does not strictly control the acyl species used for N-myristoylation but that acyl-CoA availability in a given cell greatly affects the reaction (6). The finding that the acyl-CoA binding protein ACBD6 stimulated NMT activity and enhanced the specificity of the enzyme toward the myristate chain in the presence of other species suggests that ACBD6 is part of the cellular mechanism that regulates the N-myristoylation reaction.
Of the five ACBD proteins tested, ACBD6 was the most efficient in stimulating NMT activity, but other members of the ACBD family also were able to enhance the production of the myristoyl peptide in vitro. This provides evidence that systems affecting acyl-CoA availability affect NMT activity. ACBD6 appears to be the only member to also interact directly with NMT, and this unique property could account for its greater effect on myristoylation compared with the other ACBD members. The fact that both the human and Plasmodium ACBD6 proteins stimulated PfNMT also suggests that motifs that are located downstream from the conserved ACB domain are unique to the ankyrinrepeat motif-containing member of the ACBD family. Phosphorylation of the two serine residues of the ACB domain of ACBD6 has a significant positive effect on binding capacity and myristoyl peptide formation. Other ACBD proteins are phosphorylated in vivo, but although ACBD3 and ACBD5 have six and nine phosphorylated residues, respectively, none are located in the ACB domain (39,40). The only two ACBD6 residues known to be phosphorylated are near and part of the fourth -helix of the ACB domain. These serine residues are not conserved in the ACBD family, and Ser 108 appears to be substituted by the phosphomimic Glu residue in other members. The effect of the phosphorylation of these serine residues on the properties of ACBD6 suggests that the activity of this member of the family is regulated in vivo. Cell-cycle global profiling of human phosphoproteins detected a specific upregulation of Ser 106 and Ser 108 phosphorylation during the M phase (39,40). This finding strongly suggests that the function of the phosphorylated form of ACBD6 is increased during mitosis. Given the significant difference in NMT activity in the presence of phospho-ACBD6 compared with the unphosphorylated form, an increased need for protein myristoylation during that phase of the cell cycle could be possible.
Myristoylation of proteins occurs during translation after the removal of the initiator methionine on the nascent polypeptide. However, during apoptotic events, NMT enzymes can also modify proteolytically cleaved proteins after the exposure of an N-terminal glycine and a cryptic myristoylation motif (50,51,53). ACBD6 phosphorylation could also be increased during apoptosis to support the additional need for NMT activity. A variety of pathogens rely on protein acylation processes. The development of P. falciparum in erythroid cells is blocked by several chemicals that inhibit the action of the PfNMT enzyme (11,12,54,55). The host and parasite ACBD6 regulate PfNMT activity, and the disruption of this process may provide an opportunity to block the parasite growth. In conclusion, our results provide evidence that ACBD6 proteins have an important role in the myristoylation of proteins in eukaryotic cells and support the substrate availability and selectivity of the NMT enzymes. Fig. 11. Phosphorylation enhances the stimulation of NMT activity by hACBD6. Measurements of the formation of the myristoyl peptide were performed in the absence or presence of 6 µM hACBD6, P-Ser 106 , P-Ser 108 , P-Ser 106 /P-Ser 108 , and Ser 106/108 -Glu, which carries the substitution of the residues Ser 106 and Ser 108 to Glu residues. Reactions were performed with 200 nM PfNMT in the presence of 10 µM C 14 -CoA in 120 µl at 37°C for 3, 6, 9, 12, and 15 min. The rates of formation were calculated and are presented as the amounts of myristoyl peptide formed per minute. Error bars represent the standard deviations of three measurements.