Association of NMT2 with the acyl-CoA carrier ACBD6 protects the N-myristoyltransferase reaction from palmitoyl-CoA[S]

The covalent attachment of a 14-carbon aliphatic tail on a glycine residue of nascent translated peptide chains is catalyzed in human cells by two N-myristoyltransferase (NMT) enzymes using the rare myristoyl-CoA (C14-CoA) molecule as fatty acid donor. Although, NMT enzymes can only transfer a myristate group, they lack specificity for C14-CoA and can also bind the far more abundant palmitoyl-CoA (C16-CoA) molecule. We determined that the acyl-CoA binding protein, acyl-CoA binding domain (ACBD)6, stimulated the NMT reaction of NMT2. This stimulatory effect required interaction between ACBD6 and NMT2, and was enhanced by binding of ACBD6 to its ligand, C18:2-CoA. ACBD6 also interacted with the second human NMT enzyme, NMT1. The presence of ACBD6 prevented competition of the NMT reaction by C16-CoA. Mutants of ACBD6 that were either deficient in ligand binding to the N-terminal ACBD or unable to interact with NMT2 did not stimulate activity of NMT2, nor could they protect the enzyme from utilizing the competitor C16-CoA. These results indicate that ACBD6 can locally sequester C16-CoA and prevent its access to the enzyme binding site via interaction with NMT2. Thus, the ligand binding properties of the NMT/ACBD6 complex can explain how the NMT reaction can proceed in the presence of the very abundant competitive substrate, C16-CoA.


Cloning and site-directed mutagenesis
Cloning of human ACBD1 and ACBD6 was previously reported ( 18,19 ). Human NMT2 isoform 1 (NP_004799) was cloned by PCR from NM_004808.2 into pBIND vector and human ACBD6 was cloned into the pACT vector (Check-Mate Mammalian twohybrid system; Promega). As previously described ( 37,38 ), the membrane targeting signal present in the fi rst 97 residues of NMT2 was removed to produce an active cytosolic form of the protein in Escherichia coli (also known as ecNMT2, see below). Sitedirected 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 verifi ed by full-length sequencing of the constructs. N-terminally tagged green fl uorescent protein (GFP) and hemagglutinin (HA) fusion of full-length ACBD6 and full-length NMT2 were obtained in the pAcGFP1-C1 and pIRESneo vector (Clontech Laboratories, Inc.), respectively .

Protein expression and purifi cation
All proteins and mutant forms were produced as hexahistidine fusion forms. ACBD1 and ACBD6 were expressed in the E. coli host BL21(DE3) cells (Novagen) and ecNMT2 was produced in Roset-taDE3 cells (Novagen). Proteins were purifi ed by affi nity metal chromatography with the addition of 0.2% Triton X-100 in all the buffers for the ecNMT2 protein. The purifi ed proteins were stored at Ϫ 80°C in Tris-HCl [50 mM (pH 8.0)], NaCl (0.1 M), EDTA (5 mM), and glycerol [10% (v/v), with 0.2% Triton X-100 for ec-NMT2]. Prior to isothermal titration calorimetry (ITC) measurements, proteins were dialyzed in the ITC buffer (see below).

Affi nity purifi cation MS
Affi nity purifi cation followed by LC-MS/MS was used to identify candidate protein interactions of ACBD6 following identical methods to those previously published ( 11,39 ). Briefl y, ACBD6 was cloned into the pcDNA4TO expression vector encoding a C-terminal 2×-StrepTag and then transiently expressed in HEK293 cells following reported methods ( 39 ). The cells were harvested and lysed, and the Strep-tagged protein was captured on StrepTactin Sepharose under native conditions, and then eluted with 1× desthiobiotin (IBA Technology, Gottingen Germany) as reported ( 11 ). This protein sample was processed by in-solution digestion and analyzed by LC-MS/MS peptide sequencing as reported ( 11,39 ).
MS data were searched using Protein Prospector software version 5.10.17 ( 40 ) against the Refseq NCBInr human database (downloaded July 29, 2015 from ftp://ftp.ncbi.nlm.nih.gov/ refseq/H_sapiens/mRNA_Prot/) containing 99,857 sequences, concatenated with 99,857 additional randomized decoy sequences for calculation of the false discovery rate (FDR) ( 41 ). The FDR was <1%, using a protein score of 22, a peptide score of 15, a protein expectation value of 0.01, and a peptide expectation value of 0.001. Modifi cations allowed were: fi xed carbamidomethylation of Cys and the following variable modifi cations: oxidation of Met, start-Met cleavage, oxidation of the N terminus, acetylation of the N terminus, and pyroglutamate formation from Gln.
Protein identifi cations from ACBD6 affi nity-purifi cation experiments were performed over four biological replicate experiments. Proteins were reported as specifi c interacting proteins if observed pathogen Chlamydia trachomatis , ACBD6 protein was associated with cytosolic lipid droplets and was transferred to the parasitophorous vacuole resulting in its removal from the cytosol ( 22 ). Inside the vacuole, ACBD6 regulates the activity of a bacterial acyltransferase essential for lipid metabolism of the pathogen. A related fungal fatty acyl-CoA binding protein (ACBP) from the human pathogen Cryptosporidium parvum is required by the fungus for growth in human cells, as confi rmed using chemical inhibition of acyl-CoA binding to this ACBD protein ( 23 ).
Co-translational N -myristoylation of the glycine +2 residue of nascent peptides, following removal of the initiator methionine by the human N -myristoyltransferase (NMT) enzymes (NMT1 and NMT2), can affect the association of the acyl-proteins with membranes (24)(25)(26). The addition of a 14-carbon long aliphatic tail can act as a lipid anchor and mediates oligomeric assembly and interaction with other proteins. Although amide linkage of the myristate to the glycine residue appears irreversible, a structural conformational property of some acyl-proteins, called the myristoylswitch, can remove the aliphatic tail from the lipid bilayer and dissociate the proteins from their membrane-bound location ( 25,27 ). The acyl-transferase activity of the NMT enzymes is specifi c toward myristoyl-CoA (C 14 -CoA) and is essential for the intracellular development of pathogens (28)(29)(30)(31)(32)(33)(34)(35)(36). Chemicals inhibiting myristoylation of proteins are potent drugs against parasitic protozoa and fungi. The catalytic reaction is a multistep process initiated by binding of C 14 -CoA to apo-NMT, which triggers a conformational change exposing the peptide binding site ( 26,28 ). C 14 -CoA is then hydrolyzed and CoA is released. The aliphatic tail is transferred to the glycine +2 residue of the peptide, which is then released. NMT proteins can bind palmitoyl-CoA (C 16 -CoA), but a 16-carbon chain cannot be transferred to the glycine +2 residue by the myristoyl-transferase ( 26,28,29,33 ). It is unclear how NMT enzymes are able to perform myristoylation in vivo using a very rare acyl-donor (C 14 -CoA) in the presence of a very abundant acyl-CoA competitor (C 16 -CoA). Thus, a process to prevent binding or trigger release of this competitor from the C 14 -CoA binding site of NMT seems essential to warrant its activity in vivo.
We determined that ACBD6 was associated with the two human NMT enzymes in human cells, with a stronger interaction with NMT2 than NMT1. The C-terminal domain of ACBD6 interacted with a C-terminal region of NMT2. Our results further show that ACBD6 protected the enzyme from competitive binding by C 16 -CoA and that lipid-bound ACBD6 stimulated NMT2 activity toward its substrate, C 14 -CoA. Together, these results establish a physiological role for the binding property of a member of the human family of acyl-CoA carriers, and determine how NMT enzymes can perform their function in vivo.

Materials
The No-Weigh format of disuccinimidyl suberate (DSS) was purchased from Thermo Fisher Scientific. Ellman's reagent 19 ). All experiments were performed in ammonium acetate (25 mM, pH 7.4), supplemented with 0.1% Triton X-100 to prevent precipitation of ecNMT2, at 30°C. The proteins were dialyzed in the same buffer and fresh 10 mM stocks of the ligands were prepared from powder with the dialyzing buffer. For NMT2 ( Fig.  6B ), measurements were performed with 28 injections of 10 l of 165 M ligand every 150 s. The chamber contained the protein at an initial concentration of 5.5 M. For ACBD6 ⌬ C measurements ( Fig. 6A ), protein was at a concentration of 30 M and was injected with 600 M of C 16 -CoA and 300 M of C 18:2 -CoA . For data presented in Fig. 6C , proteins were at a concentration of 10 M and C 16 -CoA at 100 M. Control experiments were run by injecting buffer into the cell containing the protein and by injecting the ligand into the cell containing buffer. Heat generated from control runs was subtracted from the data of the experimental set performed under the same conditions

NMT activity measurements
Real-time measurements of the released CoA from acyl-CoAs by purifi ed NMT2 were performed at 30°C in a Cary Avian UV-Vis spectrophotometer (50 Bio;Varian) in the presence of DTNB. Measurements were made in dual-beam mode against blank buffer and normalized by subtracting values obtained in reactions performed in the absence of the target peptide. Reactions were performed in 500 l of 20 mM sodium phosphate buffer at pH 8.0 with 1 mM EDTA and 100 M Hs pp60src#2-9 peptide, 5-100 M acyl-CoAs (C 14 -CoA, C 16 -CoA, C 18:2 -CoA), and 0.2 mM DTNB. Unless otherwise indicated, a 2:1 molar ratio of ACBD6/ NMT2 was used. Titration of NMT2 was performed to determine the optimal concentration of the partially purifi ed proteins to maintain linearity of the measurement over a period of 5 min (usually 2 M ACBD6/1 M NMT2). Kinetic calculations were performed with GraphPad Prism 6.

Interaction of NMT enzymes with ACBD6
ACBD6 protein is a modular protein with an N-terminal conserved ACBD and two ankyrin-repeat motifs present in the nonconserved C-terminal domain ( Fig. 1A ). Affi nity purifi cation MS was performed using a C-terminally Streptagged ACBD6 construct, transiently expressed in HEK293 cells, in four biological replicates. From these experiments, the most specifi c interacting proteins were NMT1 and NMT2, identifi ed with a total of 24 and 14 unique peptides, respectively (supplementary Table 1). NMT1 and NMT2 share high sequence identity (73%, ClustalW alignment), thus unique and shared peptide sequences were identifi ed to each of these proteins in the MS data (supplementary Table 2 ). Suffi cient unique peptide counts were obtained to both NMT1 and NMT2 in each experiment to confi rm the presence of both proteins at comparable levels in this semi-quantitative analysis (supplementary Table 3). Two proteins, CAMK2D and UGGT1, were also identifi ed at a lower specifi city threshold, but NMT1 and NMT2 were the only proteins that reproducibly interacted with ACBD6 in all four experiments. The specifi city of the association with ACBD6 was further confi rmed with a mammalian two-hybrid interaction assay, which established a stronger binding of ACBD6 with NMT2 than with NMT1, as compared with the related ACBD5 in at least three of four replicate experiments with more than one unique peptide in at least one experiment. These proteins were also compared with a list of high-frequency background proteins from >2,000 similar affi nity purifi cation MS experiments with Strep-tagged proteins ( 11 ), and excluded if the background frequency was >0.10. MS data for proteins meeting these criteria are reported in supplementary Table 1.

Protein interaction assays
Mammalian two-hybrid interaction assays were performed in HeLa and HEK293 cells grown in 96-well plates with the CheckMate system (Promega). Hela cells were transfected with TurboFect reagent (Thermo Scientifi c) and HEK293 cells with TransIT-LT1 transfection reagent (Mirus Bio). Luminescence was quantifi ed with the Dual-Luciferase reporter assay of Promega, according to the manufacturer's instructions. The homobifunctional N -hydroxysuccinimide ester reagent, DSS, reacts with primary amine groups and was used to detect interaction of purifi ed NMT2 and ACBD6 proteins. DSS was bought in individual sealed microtubes (Thermo Fisher Scientifi c) and stock solutions of 100 mM in DMSO (further diluted in DMSO as needed) were made fresh prior to each experiment. DSS was added in a 30-fold excess of the protein (mole/mole). Proteins were dialyzed in 10 mM potassium phosphate (pH 7.4) at 4°C. ACBD6 and NMT2 proteins were mixed at a 1:1 molar ratio (20 M each) in 29 l of 50 mM potassium phosphate (pH 7.4) at 37°C for 20 min. When acyl-CoAs were used, they were added at a fi nal concentration of 50 M. One microliter of DSS was then added to the mixture at a fi nal concentration of 1 mM, and subsequently incubated for 30 min at room temperature. Untreated control reactions were performed with addition of 1 l of DMSO and incubated as described for the DSS-treated samples. Reactions were stopped by quenching of DSS with 2 l Tris-HCl (1 M, pH 7.4) for at least 15 min at room temperature. Mixtures were then boiled in SDS-PAGE loading buffer for 4 min and separated on denaturing SDS-polyacrylamide Tris-glycine gels. Proteins were detected by staining with GelCode Blue reagent (Thermo Fisher Scientifi c) or were transferred on nitrocellulose membrane and detected with a HRP-conjugated anti-histidine antibody (INDIA-HisProbe-HRP antibody; Thermo Fisher Scientifi c).

Co-immunoprecipitation experiments
Protein samples were made from HEK293 cells grown in 12-well plates. For each sample, three wells were cotransfected using the TurboFect reagent (Thermo Scientifi c) with the pair HA-ACBD6/ GFP-NMT2 or HA-NMT2/GFP-ACBD6 for 48 h. Cells were lysed with lysis immunoprecipitation buffer (Thermo Scientifi c) and protease cocktail inhibitor (Thermo Scientifi c) in a fi nal volume of 400 l . Cleared lysates were stored at Ϫ 80°C. For co-immunoprecipitation, 200 g of each protein sample was precleared with 25 l protein G-magnetic beads (NEBiolabs) for 1 h at 4°C. Following removal of the beads, 5 g of mouse monoclonal anti-HA antibody (Sigma-Aldrich, H3663) was added and incubated for 1 hr at 4°C. Then, 25 l of fresh protein G-magnetic beads was added and incubated overnight at 4°C. Beads were then washed three times with ice-cold PBS and with lysis IP buffer. Complexes were eluted by suspending the beads in 60 l SDS-PAGE loading buffer and heating at 70°C for 10 min. After removal of the beads, 10 l of each sample was analyzed by Western blot with rabbit polyclonal anti-GFP (Sigma-Aldrich, G1544) and with HRP-conjugated mouse anti-HA antibodies (Sigma-Aldrich, H6533)

ITC assays
ITC measurements of the binding of acyl-CoAs and fatty acids were performed on a VP-ITC instrument (MicroCal, LLC) ( 18, shown). When ACBD6 and ecNMT2 were mixed at a 1:1 molar ratio, most of the ecNMT2 monomeric form disappeared, and a new band was detected at approximately the calculated mass of a heterodimeric ACBD6/ecNMT2 complex (80 kDa). Addition of the NMT2 substrate, C 14 -CoA, and of the ACBD6 ligand, C 18:1 -CoA, did not prevent formation of the complex ( Fig. 3A ). In contrast, the truncated recombinant ACBD6 ⌬ C mutant (21 kDa) did not form a complex with ecNMT2 ( Fig. 3B ).

Ligand binding to ACBD6 stimulates the NMT reaction
Activity of purifi ed ecNMT2 enzyme was measured by the real-time detection of the release of CoA from C 14 -CoA in the presence of a synthetic peptide matching the amino terminal myristoylation sequence of c-Src protein ( 24 ) (supplementary Fig. 1C, D). Addition of increasing concentrations of ACBD6 protein resulted in the stimulation of activity of ecNMT2 ( Fig. 4 ). No inhibitory effect was observed, even with a 20:1 ACBD6/ecNMT2 molar ratio. Under similar conditions, addition of purifi ed ACBD1 had only a weak stimulatory effect. Compared with ACBD6, two ACBD6 mutant forms, Y 114 A and FKKY-AAAA (dubbed FKKY mut ), with defects in acyl-CoA binding ( 19 ) failed to stimulate the reaction in the presence of C 14 -CoA, thus acyl-CoA binding of ACBD6 is required for the stimulatory effect on NMT2 activity ( Fig. 5 ). The molecule C 18:2 -CoA, a ligand of ACBD6, was not an acyl donor for the NMT reaction ( Figs. 4, 5 ). In the absence of ACBD6, addition of C 18:2 -CoA to C 14 -CoA had a weak inhibitory effect on the reaction ( Fig. 5 ). However, in the presence of ACBD6, addition of C 18:2 -CoA resulted in increased stimulation of ec-NMT2 activity compared with the activity obtained with only ACBD6. This stimulatory effect was not detected with the two mutant ACBD6 forms, Y 114 A and FKKY mut . Thus, compared with apo-ACBD6, acyl-CoA-bound ACBD6 appeared to be the form enhancing the processing of C 14 -CoA by NMT2, independent of its binding to C 14 -CoA.
Association of NMT2 with ACBD6 was further investigated by co-immunoprecipitation of wild-type and mutant constructs of GFP-tagged ACBD6 by HA-NMT2, and of the inversely tagged GFP-NMT2 by HA-ACBD6 ( Fig. 2A, B ). A C-terminally truncated form of ACBD6 lacking the two ankyrin-repeat motifs (ACBD6 ⌬ C, Fig. 1A ) did not coimmunoprecipitate with NMT2 ( Fig. 2A, B ). The C terminus of NMT-2 includes residues that bind the target peptide substrate ( Fig. 1A ), and the C terminus itself may assist in catalysis by deprotonating the ammonium of the acceptor glycine residue ( 28 ). An NMT2 form lacking the last 58 residues, including the signature motif PS00976 (NMT2 ⌬ C, Fig. 1A ), did not interact with full-length ACBD6 either by co-immunoprecipitation or mammalian two-hybrid assay ( Fig. 2C ; supplementary Fig. 1A). Deletion mutants of conserved motifs of the ACBD of ACBD6 were also designed ( Fig. 1 , regions H2-4), and these did not prevent interaction with NMT2 ( Fig. 2D ). Thus, ACBD6 interacts via its C-terminal domain containing the ankyrin-repeat motifs with NMT2, and the NMT2 interaction with ACBD6 is dependent upon its C-terminal region.

ACBD6 and NMT2 form hetero-dimeric complexes
Association of the two purifi ed proteins was investigated using covalent cross-linking with the reagent DSS, followed by analysis of the trapped complexes by SDS-PAGE. As previously reported ( 18,19 ), ACBD6 (32 kDa) behaved as a monomeric protein, and no high molecular mass species were detected after DSS treatment ( Fig. 3A ). To produce and purify NMT2, a recombinant form was expressed in E. coli without the N-terminal membrane localization sequence (also known as ecNMT2 protein) ( 37,38 ). Purifi ed ecNMT2 (supplementary Fig. 1B) was primarily monomeric (48 kDa), and few bands of very high molecular mass were formed after DSS treatment (>100 kDa, not Fig. 1. Acyl-CoA binding protein ACBD6 and NMT2. A: The predicted four helices (H1 to H4) of the ACBD and the two ankyrin-repeat motifs of ACBD6 are shown. Residues changed to alanine by site-directed mutagenesis in the mutated forms S 106 A, Y 114 A, and FKKY-AAAA (also known as FKKY mut ) are shown . The form ACBD6 ⌬ C lacked the ankyrin domains. Isoform 1 of NMT2 (NP_004799.1) is shown with the two C 14 -CoA:protein NMT conserved domains, pfam01233 and pfam02799, and the two signature motifs, PS00975 and PS00976. The recombinant form produced in E. coli (ecNMT2) lacking the fi rst 97 residues and the truncated form (NMT2 ⌬ C) lacking the last 58 residues are indicated. B: Luminescence interaction quantification assays were performed with the mammalian CheckMate system (Promega) in HEK293 cells. Compared with ACBD5, which did not interact with NMT proteins, association of ACBD6 with NMT1 and NMT2 was 25-and 250-fold stronger, respectively. Error bars represent the standard deviations of three measurements. We previously identifi ed ZNF23 in a yeast two-hybrid screen using ACBD6 as bait ( 62 ), but that interaction could not be confi rmed in mammalian cells (supplementary Fig. 1A).
domain of ACBD6 is dispensable for ligand binding ( 18 ) and we confi rmed that the truncated ACBD6 ⌬ C form binds C 18:2 -CoA ( Fig. 6A ). ACBD6 ⌬ C did not interact with ec-NMT2 ( Fig. 2 ), suggesting that the ankyrin-repeat motifs were involved in the contact to the C-terminal motif of

ACBD6 interaction is required for NMT2 activity stimulation
The requirement for protein interaction in the stimulatory effect of ACBD6 in the presence of its ligand on the NMT reaction was investigated. The carboxy-terminal and HA-tagged NMT2 ⌬ C (C). GFP proteins were detected with a rabbit polyclonal anti-GFP antibody and HA proteins with an HRP-conjugated mouse monoclonal anti-HA antibody. As shown in (A), samples were also analyzed with a rabbit peptide-raised anti-ACBD6 antibody. Control cells were obtained by transfection of the two empty vectors (GFP and HA). For each sample, the signal obtained with 10% of the lysate before immunoprecipitation (input) is shown on the left of the signal obtained after pull-down ( ␣ HA). Note that the polyclonal anti-GFP antibody nonspecifi cally cross-reacted with a band that did not correspond to GFP-NMT2, GFP-ACBD6, or GFP-ACBD6 ⌬ C. Duplicates of the co-immunoprecipitation of GFP-ACBD6 or GFP-ACBD6 ⌬ C by HA-NMT2 are shown and indicated as samples #1 and #2 in (A ). D: Densitometry quantifi cation of the GFP signal obtained by Western blot analysis of co-immunoprecipitation samples obtained in cells cotransfected with GFP-NMT2 and the indicated truncated forms of ACBD6. Data were collected from a single representative blot. ABCD6 with NMT2 was required for the ligand-mediated increase in NMT2 activity.

N-terminal ACBD6 domain prevents binding of C 16 -CoA to NMT2
Purifi ed ecNMT2 binds both C 14 -CoA and C 16 -CoA with similar affi nity ( Fig. 6B ). Binding of C 16 -CoA to ec-NMT2 resulted in cleavage of the ester bond and release of CoA ( Fig. 7A ). Confi rmation of the strong binding affi nity of NMT2 for this acyl-CoA competitor was obtained by the observation that 20 M of C 16 -CoA produced a stronger signal than 15 M of C 14 -CoA ( Fig. 7B ). Addition of ACBD6 resulted in the stimulation of the reaction with C 14 -CoA, but it decreased the processing of C 16 -CoA by ecNMT2 ( Fig. 7A, B ). The mutant S 106 A lacks one of the two phosphorylated serine residues of ACBD6 ( 42,43 ), and binds to C 16 -CoA with stronger affi nity than ACBD6. This mutant stimulated ecNMT2 activity with C 14 -CoA as ACBD6 ( Fig. 5 ), but it could prevent the processing of C 16 -CoA at much higher concentration than ACBD6 ( Fig. 7B ). In contrast, the ACBD6 mutant, Y 114 A, which did not bind C , could not prevent the processing of C 16 -CoA by NMT2, even at very low concentration ( Fig. 7B ).

NMT2 interaction to the C-terminal ACBD6 domain is required for competitor protection
As established above, binding of C 16 -CoA to ACBD6 prevented processing of the acyl-CoA competitor by ecNMT2. However, the effect of ACBD6 could not be accounted for by a model requiring exhaustion of free-competitor in the reaction mixture because as little as 1 M of ACBD6 was enough to reduce the processing of 100 M C 16 -CoA to NMT2 ( Fig. 2C ; supplementary Fig. 1A). Compared with full-length ACBD6, ACBD6 ⌬ C could not stimulate the NMT reaction and the ecNMT2 activity rate in the presence of the truncated form was similar to those obtained with the Y 114 A and FKKY mut mutants ( Fig. 5 ). Thus, interaction of

Displacement of C 16 -CoA by linoleoyl-CoA
To confi rm that protection of NMT2 from C 16 -CoA by ACBD6 was the result of C 16 -CoA binding to ACBD6, the ligand, C 18:2 -CoA, was added to the reaction. ACBD6 binds C 18:2 -CoA with greater affi nity than C 16 -CoA ( 18,19 ). At an equal concentration of C 18:2 -CoA and C 16 -CoA (5 M), ACBD6 could no longer prevent processing of C 16 -CoA by ecNMT2 and activity rate doubled ( Fig. 7C ). The mutant Y 114 A, unable to bind either acyl-CoA species ( 19 ), had no detectable effect on ecNMT2 activity, irrespective of the C 18:2 -CoA concentrations. The mutant S 106 A, which binds C 16 -CoA with stronger affi nity than ACBD6, limited use of C 16 -CoA by ecNMT2, even when a 10-fold excess of half ( Fig. 7B ). As seen for the C 18:2 -CoA stimulation of C 14 -CoA processing by ecNMT2, ACBD6/NMT2 complex formation was also necessary to prevent usage of C 16 -CoA by ecNMT2. The truncated ACBD6 ⌬ C form, which did not interact with ecNMT2 ( Fig. 2A and 2B ), but bound C 16 -CoA ( Fig. 6A , inset), could not prevent processing of C 16 -CoA by ecNMT2. In fact, that form was even less effi cient in protecting ecNMT2 than the mutant Y 114 A ( Fig. 7B ). These results suggested that in the hetero-dimeric complex, ACBD6-NMT2 formed through the interaction of their C-terminal domains, C 16 -CoA was prevented from accessing the NMT2 binding site through sequestration to the N-terminal ACBD of ACBD6.  Thus, in the absence of a process to modulate binding or release of the acyl-CoA competitor from the C 14 -CoA binding site, compounds such as C 16 -CoA would inhibit the action of NMT enzymes in vivo. This is illustrated in Fig. 8 , which provides a summary of data presented in this study, indicating an important role of acyl-CoA binding proteins in this process. Several other reports provide evidence that other cellular components modulate the NMT reaction to ensure that the correct acyl chain is linked to the protein substrate. Whereas NMT enzymes are highly specifi c in transferring a 14-carbon tail on nascent peptides, C 14:1 and C 14:2 , rather than C 14:0 , are the fatty acids used to modify proteins such as the ␣ -subunit of the G-protein photoreceptor in retinal photoreceptor cells (49)(50)(51). The acyl-CoA pools in the retina are not signifi cantly different as compared with other tissues; hence a selective system needs to account for the unique acylation taking place in these cells ( 49 ). It can be hypothesized that acyl-CoA carriers with stronger affi nity to unsaturated myristate control the acylation profi le observed in this tissue.
In our study, the stimulation of the NMT activity of NMT2 in the presence of apo-ACBD6 and C 14 -CoA could be explained by the formation of the ACBD6/NMT2 complex ( Fig. 8B ). We speculate that in the ACBD6/NMT2 complex, the entry of the C 16:0 acyl-chain to the binding C 18:2 -CoA was added to the reaction. Thus, selectivity of binding of the ACB motif of the acyl-CoA carrier ACBD6 modulates the NMT reaction by controlling availability of acyl-CoA to NMT2 and stimulates its activity via protein contact with their respective C-terminal domains.

DISCUSSION
The acyl-transferase activity of the NMT proteins is highly specifi c toward C 14 -CoA. On the other hand, a variety of glycine-containing nascent peptide chains can bind to the C-terminal domain of NMT ( 33,(44)(45)(46)(47), indicating that NMT is less selective toward its protein substrate. The catalytic reaction is a multistep process initiated by binding of C 14 -CoA to the N-terminal domain of apo-NMT, which triggers a conformational change exposing the peptide binding site ( 26,28,48 ). C 14 -CoA is then hydrolyzed and CoA is released. The aliphatic tail is transferred and covalently linked to the exposed glycine residue of the peptide substrate, releasing the acylpeptide. Whereas the transfer rates of acyl chains other than C14 are extremely slow compared with myristate ( 33,44,47 ), NMT proteins bind the very abundant C 16 -CoA molecule with the same affi nity as C 14 -CoA, a minor acyl-CoA in cells ( 26,28,29,33,49 ).  8. Proposed events leading to the protection of NMT2 activity by ACBD6. A: NMT2 will bind both C 16 -CoA (C 16:0 ) and C 14 -CoA (C 14:0 ), but can only use C 14:0 to acylate a peptide. The high abundance of C 16:0 in vivo effectively inhibits protein myristoylation. B: ACBD6 interacts with NMT2 through binding mediated by its ankyrin (ANK) motif to the C terminus of NMT2. The ACBD domain of ACBD6 sequesters C 16:0 -CoA. The complex formed by NMT2 and ACBD6 blocks access of C 16:0 -CoA to the acyltransferase site. ACBD6 prevents C 16 -CoA inhibition of the myristoylation reaction and stimulates the formation of myristoylated-peptide. A red plus sign in the speculated entry site of the ACBD6/NMT2 complex indicates stimulation. C: Protein myristoylation is further stimulated by the binding of linoleoyl-CoA (C 18:2 ) to ACBD6. In vivo phosphorylation of the serine 106 residue (S106) in the ACBD of ACBD6 plays a role in this process. Phosphoserine106 is shown as a red fi lled circle next to the red plus sign indicating increased stimulation compared with the scheme depicted in (B). may offer another strategy to affect the activity of the parasite NMT and its association to ACBD6 (59)(60)(61).
The NMT process in vivo is more complex than explored in this study. For example, the generation of the acyl-donor and acyl-acceptor of the reaction are localized processes that require movement of the fatty acyl molecules through membranes and binding to carriers such as FABP, their activation to acyl-CoAs by acyl-CoA synthetases, and their transport by ACBD proteins to the NMT enzyme . ACBD proteins are involved in a variety of cellular mechanisms through association with a multiplicity of proteins. However, to our knowledge, the role of the binding of the acyl-CoA ligand to the ACBD function has not been established in these processes. The modulation of the NMT reaction by ligand binding to ACBD6 in the enzymatic complex formed with NMT2 provides one of the fi rst examples of a physiological role for the acyl-CoA binding property of a member of the human ACBD family.
site of the acyltransferase subunit is reduced in favor of the C 14:0 acyl-chain. Mutant forms of ACBD6, which are unable to bind C 16 -CoA or are unable to interact with the enzyme, could not prevent NMT2 from using this acyl donor ( Fig. 7 ). These fi ndings suggested that protection of the NMT reaction was also the result of the local sequestration of C 16 -CoA by the acyl-CoA carrier subunit of the ACBD6/ NMT2 complex. The highly improved preference for C 14:0 -CoA relative to C 16:0 -CoA in this complex is essential, given that these two compounds are at the low and high end of the concentration range, respectively ( 49,52 ). Without this shift in specifi city triggered by ACBD6, it is unlikely that the minor C 14:0 -CoA molecule could ever access the NMT2 binding site. Because ACBD6 can also bind C 14 -CoA, the stimulatory effect could also indicate the channeling of C 14 -CoA from the binding site of ACBD6 to NMT2 in the complex. In addition, the formation of the ACBD6/NMT2 complex could also alter the binding affi nity of NMT2 toward C 16 -CoA and render its binding to NMT2 less favorable. A particular role should be considered for the serine 106 residue of ACBD6. This site is phosphorylated in vivo ( 42,43 ) and, as shown in Fig. 7C , this residue is essential in controlling access of C 16 -CoA to NMT2. The binding of C 16 -CoA to the ACBD motif of phosphoACBD6 might trigger a substantial conformational change in the complex leading to blocking the entry or channeling of C 16 -CoA to the NMT2 binding site. In vivo, ACBD6 and NMT2 are exposed to a variety of acyl-CoAs at different (local) concentrations ( 49,52 ). These acyl-CoAs that are bound to the ACBD6/NMT2 complex may further stimulate the NMT reaction toward C 14:0 -CoA ( Fig. 8C ). In support of this model, we established that activity of NMT2 toward C 14:0 -CoA was stimulated by binding of C 18:2 -CoA to ACBD6 ( Fig. 5 ). The role of serine 106 is highlighted, as the ACBD6 serine mutant prevented C 16 -CoA access to NMT2 independently of C 18:2 -CoA ( Fig. 7 ). Together, this suggests that even under condition of excess of C 16 -CoA, ACBD6 would protect the specifi city of the NMT reaction as long as the formation of the complex with NMT2 or phosphorylation of ACBD6 was not prevented.
A variety of pathogens rely on protein acylation processes. Because binding of the peptides does not affect acyl-CoA binding to the N-terminal motif ( 33,53,54 ), compounds that can recognize the C-terminal binding domain of the NMT proteins of invading pathogens seem to be good candidates to specifi cally disrupt myristoylation without impacting acylation of the proteins of the host (28)(29)(30)(31)(32)(34)(35)(36). As examples, the development of Plasmodium falciparum in erythroid cells is blocked by several chemicals inhibiting the P. falciparum NMT enzyme (35)(36)(37)55 ). The inhibition of myristoylation of the HIV Gag protein prevents its membrane association and blocks viral budding ( 56 ). Interestingly, both P. falciparum and Plasmodium vivax have an ACBD6 homolog which is expressed during development ( 57,58 ). As defi ned in Chlamydia-infected human cells ( 22 ), the host ACBD6 and the parasite ACBD6 homolog could cross the vacuole membrane in association with lipid droplets. Disruption of such a process