Biochemical studies of Saccharomyces cerevisiae myristoyl-coenzyme A:protein N-myristoyltransferase mutants.

Saccharomyces cerevisiae myristoyl-CoA:protein N-myristoyltransferase (Nmt1p) is an essential 455-residue, monomeric enzyme that catalyzes the transfer of myristate from myristoyl-CoA to the NH2-terminal Gly residue of cellular proteins. Nmt1p has an ordered Bi Bi reaction mechanism with binding of myristoyl-CoA occurring before binding of peptide substrates. To define residues important for function, the polymerase chain reaction was used to generate random mutations in the NMT1 gene. A colony color sectoring assay was used to screen a library of 52,000 transformants for nmt1 alleles encoding enzymes with reduced activity. nmt1 alleles were identified that produced temperature-sensitive (ts) growth arrest due to substitutions affecting eight residues conserved in orthologous Nmts: Asn102, Ala202, Cys217, Ser328, Val395, Asn404, Leu420, and Asn426. Ala202 → Thr, Cys217 → Arg, Ser328 → Pro, Asn404 → Tyr, and Asn426 → Ile produced the most severe ts phenotype. Their effects on the functional properties of the enzyme's myristoyl-CoA and peptide binding sites were defined by purifying each mutant from Escherichia coli and conducting in vitro kinetic analyses with acyl-CoA and peptide substrates and with two competitive inhibitors: S-(2-oxo)pentadecyl-CoA, a nonhydrolyzable myristoyl-CoA analog, and SC-58272, a peptidomimetic derived from the NH2-terminal sequence of an Nmt1p substrate (ADP-ribosylation factor-2, Arf2p). None of the substitutions affect the enzyme's acyl chain length selectivity. When compared with wild type Nmt1p, Cys217 → Arg produces 3- and 6-fold increases in Ki for SC-58272 at 24 and 37°C but no change in Ki for S-(2-oxo)pentadecyl-CoA, indicating that the substitution selectively affects Nmt1p's peptide binding site. Asn426 → Ile selectively perturbs the myristoyl-CoA binding site, resulting in the most pronounced reduction in affinity for S-(2-oxo)pentadecyl-CoA (12- and 20-fold). Ala202 → Thr, which confers the most severe ts phenotype, provides an example of a substitution that affects both sites, producing 3- and 6-fold increases in the Ki for S-(2-oxo)pentadecyl-CoA and 6- and 9-fold increases in the Ki for SC-58272 at 24 and 37°C. An N-myristoylation-dependent change in the electrophoretic mobility of Arf1p was used to assay the effects of the mutants on cellular levels of protein N-myristoylation under a variety of growth conditions. The ts growth arrest produced by nmt1 alleles correlates with a reduction in myristoyl-Arf1p to ≤50% of total cellular Arf1p.

Saccharomyces cerevisiae Nmt1p has been used as a model for examining the enzyme's kinetic mechanism, substrate specificities, and biological functions. The NMT1 gene is essential for vegetative growth (11,12). It encodes a 455-residue cytoplasmic protein (13). Nmt1p has no known co-factor requirements (14). The enzyme's mechanism is ordered Bi Bi (15). Apoenzyme binds myristoyl-CoA, forming a high affinity myristoyl-CoA⅐Nmt1p complex (K d ϭ 15 nM; Ref. 16). There is heterotropic cooperativity between the enzyme's acyl-CoA and peptide binding sites; formation of the binary complex allows synthesis of a peptide binding site (16). After assembly of a ternary myristoyl-CoA⅐Nmt1p⅐peptide complex, C14:0 is transferred from CoA to the peptide substrate, and the products are released (CoA followed by myristoylpeptide). Nmt1p's acyl-CoA and peptide specificities have been defined in vitro using purified enzyme, Ͼ300 fatty acid analogs with systematic alterations in chain length, polarity, conformation, and steric bulk, plus Ͼ100 octapeptides representing variations in the NH 2terminal sequences of known N-myristoylproteins (reviewed in Ref. 1).
The primary structures of six orthologous Nmts have been determined (11,(17)(18)(19). The 450 -529-residue proteins contain 105 absolutely conserved amino acids. Searches of current protein data bases with these primary structures has failed to disclose any discernible homology to other sequence entries.
Only a modest amount of information exists about Nmt1p's structure/activity relationships. Deletion analyses suggest that the minimal domain required for myristoyltransferase activity spans Ile 59 -Phe 96 through Gly 451 -Leu 455 (20). Two distinct genetic selections have identified two residues important for function. An allele with a single amino acid substitution, Gly 451 3 Asp (nmt1-451D) was recovered during a screen for mutations that cause temperature-sensitive (ts) fatty acid auxotrophy (12,21). nmt1-451D produces growth arrest at various stages of the cell cycle 1 h after shifting from 24 to 37°C and produces lethality after an 8 -12-h incubation at nonpermissive temperatures (12,22). The growth arrest and lethality can be rescued at 37°C by adding myristate but not shorter or longer chain saturated fatty acids. Another allele with a single Leu 99 3 Pro substitution (nmt1-99P) is associated with undermyristoylation of Gpa1p, thereby reducing this ␣ subunit's affinity for the ␤␥ polypeptides of a heterotrimeric G protein involved in the mating response (23,24). The resulting free ␤␥ subunits produce constitutive activation of the mating pathway and growth arrest. Leu 99 3 Pro results in less global alterations in protein N-myristoylation than does Gly 451 3 Asp (24). Gly 451 and Leu 99 are conserved in all six known Nmts. Introduction of the Gly 3 Asp substitution in Cryptococcus neoformans and Candida albicans Nmts also produces temperature-sensitive growth arrest and myristic acid auxotrophy in these fungal pathogens (5,6).
Site-directed mutagenesis has been used to replace each of human Nmt's four conserved His residues with Asn and each of its two conserved Cys residues with Ser (25). Site-directed mutagenesis of absolutely conserved amino acids can be very useful for identifying residues critical for substrate recognition, binding, and/or catalysis. In the case of Nmt, designing such experiments is hindered by the high percentage of conserved amino acids, by the lack of homology to other proteins, and by the lack of a rapid screening assay for disabling mutations. In this report, we describe how random PCR mutagenesis and an in vivo screen can be used to recognize mutations of conserved residues in Nmt1p that affect the functional properties of its myristoyl-CoA and/or peptide binding sites.

Random PCR Mutagenesis and Colony Color Sectoring Assay
The entire open reading frame (ORF) of NMT1 plus three overlapping subdomains of the ORF were amplified by PCR (Fig. 1A). PCR was performed using the conditions described in Fig. 1B plus the following thermocycling protocol: 94°C for 1 min (denaturation), 55°C for 2 min (annealing), and 72°C for 3 min (extension) for a total of 30 cycles.
Each PCR fragment was used to replace a segment of the NMT1 ORF by in vivo recombination with a gapped plasmid (26). Replacement involved the following steps: (i) the PCR product was purified by agarose gel electrophoresis and the GeneClean kit (BIO 101); (ii) pBB361 (NMT1 LEU2) was digested with either NcoI and BalI, NcoI and AvrII, SphI and MluI, or BalI alone (Fig. 1A), and the linearized, gapped plasmids were purified as in the first step; (iii) S. cerevisiae strain YB523 (nmt1⌬, ura3-52, ade2, ade3, leu2, pBB290) was co-transformed with the PCR fragment and gapped plasmids (27); (iv) cells were plated on SC minus leucine and incubated for 3 days at 30°C.
A colony color sectoring assay was performed as outlined in Fig. 2. Leucine prototrophs were replated on YPD and incubated at 30°C for 5 days. Individual red colonies were restreaked on YPD and incubated for 5 days at 30°C. Colonies that continued to exhibit a nonsectoring (red) phenotype due to retention of the 2 pBB290 episome (NMT1 ADE3 URA3), were replica plated onto SC/5-FOA minus leucine and incubated at 24°C and 35-40°C to identify nmt1 mutations that produced a ts growth phenotype.
Total cellular nucleic acids were extracted (28) from ts isolates and used to transform Escherichia coli strain JS5. Plasmids were purified from bacterial colonies that grew on Luria broth (LB) supplemented with kanamycin (50 g/ml). The nucleotide sequence of each plasmid's nmt1 insert was determined using a panel of 14 oligonucleotide primers (11,12), a DyeDeoxy terminator cycle sequencing kit, and a model 373 Automated DNA Sequencer (Applied Biosystems). The primers cover both strands of NMT1's ORF at intervals of 200 base pairs. 400 -500 base pairs of sequence were obtained per primer, allowing 200 -300base pair overlaps between each primer-driven reaction.

Site-directed Mutagenesis
Twenty-three of the ts nmt1 alleles contained more than one amino acid substitution. These multiply mutated protein sequences were aligned with one another and with the sequences of orthologous Nmts using the algorithm included in GeneWorks (version 2.4). The alignments identified eight residues that were conserved in five or six of the six known Nmts and mutated in the ts nmt1 alleles. Site-directed mutagenesis was used to introduce each of the eight mutations (individually) into NMT1's ORF: Asn 102 3 Thr (AAC 3 ACC); Leu 171 3 Ser (TTG 3 TCG); Lys 389 3 Ile (AAA 3 ATA); Val 395 3 Asp (GTT 3 GAT); Leu 408 3 Ser (TTG 3 TCG); Phe 413 3 Ser (TTC 3 TCC); Asp 417 3 Val (GAC 3 GTC); and Leu 420 3 Ser (TTG 3 TCG). The entire ORF of each site-directed mutant nmt1 allele was sequenced to confirm that only the desired nucleotide substitution was present.
One hundred-milliliter cultures of the various transformants were grown at 24°C in YPD to an A 600 ϭ 0.8. Twenty-five-milliliter aliquots were removed and incubated at 24 or 37°C for 2 h. Cells were subsequently harvested by centrifugation at 1,600 ϫ g for 10 min at 4°C and washed twice with 15 ml of phosphate-buffered saline. The cell pellet was resuspended in 0.5 ml of lysis buffer (2% SDS, 80 mM Tris, pH 6.8, 200 M Pefabloc SC, 2 M leupeptin, 2 M pepstatin; all protease inhibitors from Boehringer Mannheim). Cells were disrupted by vortexing with 0.5 ml of 425-600-m glass beads (Sigma; vortexing was in four cycles, 1 min/cycle). The mixture was boiled for 10 min. Cellular debris were removed by centrifugation at 10,000 ϫ g for 5 min. The protein concentration in the cleared lysates was determined using the BCA assay kit (Pierce). Equal masses of protein from each sample (100 g) were reduced, denatured, fractionated by SDS-PAGE (32), and transferred to polyvinylidene difluoride membranes (Amersham Corp.). Western blots were probed with a previously characterized rabbit anti-C. albicans Nmt sera that recognizes the orthologous S. cerevisiae acyltransferase (Ref. 19; diluted 1:5000 in Blotto). Antigen-antibody complexes were visualized using an enhanced chemiluminescence (ECL) kit and the protocol recommended by its manufacturer (Amersham).
Expression and Purification of 6XHis-Nmts-1.5-liter cultures of E. coli strain JM101, containing a 6XHis-nmt1p expression plasmid, were grown in LB/streptomycin (30 g/ml) and 100 M isopropyl-␤-D-thiogalactopyranoside at 24°C for ϳ12 h until they reached an A 600 ϭ 1. Bacteria were harvested by centrifugation at 5,000 ϫ g for 20 min at 4°C and resuspended in 50 ml of buffer A (0.3 M NaCl, 5 mM ␤-mercaptoethanol, 1 mM Pefabloc SC, 2 M leupeptin, 2 M pepstatin, 25 mM sodium phosphate buffer, pH 7.0). Cells were lysed with French press at 3,200 p.s.i. The lysates were centrifuged at 20,000 ϫ g for 30 min at 4°C. Each supernatant was collected and mixed with 5 ml of Ni 2ϩ -NTAagarose (Qiagen; the resin was washed three times with buffer A before use). The resulting suspension was incubated on ice for 2 h with intermittent shaking and then subjected to centrifugation at 3,000 ϫ g for 5 min at 4°C. The supernatant was removed and discarded. The Ni 2ϩ -NTA-agarose was washed four times with ice-cold buffer A (30 ml/wash) and loaded into an empty 10-ml Poly-Prep column (Bio-Rad). The column was washed with 10 ml of buffer A containing 40 mM imidazole. Nmt activity was eluted by sequentially washing the column with 3 ml of buffer A containing 60 mM, 80 mM, 100 mM, 120 mM, and finally 150 mM imidazole. Wild type and mutant enzymes eluted with buffer A, 120 mM imidazole and were used on the day of their purification for kinetic studies (see below). Wild type 6XHis-Nmt1p and each mutant 6XHis-nmt1p were purified on at least three separate occasions. The purity of each preparation was verified by SDS-PAGE followed by Coomassie staining.
Nmt1p without an NH 2 -terminal 6XHis tag was also expressed in E. coli and purified to apparent homogeneity using a protocol described by Rudnick et al. (33).

Kinetic Studies of Wild Type and Mutant Nmts
Assessment of Peptide Substrate Specificity-A coupled in vitro Nmt assay system was employed (33,34). [ 3 H]Myristoyl-CoA was generated first using [ 3 H]myristate, CoA, and Pseudomonas acyl-CoA synthetase (Boehringer Mannheim). A purified wild type or mutant Nmt was then added together with a peptide substrate. Following a 10-min incubation at 24 or 37°C, [ 3 H]myristoylpeptide was purified from the reaction mixture by reverse phase high pressure liquid chromatography using a 4.6 m (internal diameter) ϫ 150-mm C4 column (Vydac) and a linear gradient from H 2 O containing 0.1% trifluoroacetic acid and 0.05% triethylamine to acetonitrile, 0.1% trifluoroacetic acid. The amount of [ 3 H]myristoylpeptide produced was quantitated with an in-line scintillation counter (Packard).
Defining the Affinity of Mutant Nmts for a Peptidomimetic Inhibitor, SC-58272-A series of K m Ј values were defined for the Ppz2p octapeptide substrate (10 -20 M) at 24 and 37°C using 230 nM [ 3 H]myristoyl-CoA, 10 -1000 nM of the peptidomimetic inhibitor SC-58272 (7), and 0.1-250 g/ml of Nmt1p, 6XHis-Nmt1p, or a 6XHis-nmt1p. The K m Ј at each inhibitor concentration and the type of inhibition were determined using double reciprocal plots. A K i was obtained from a plot of the K m Ј values versus the inhibitor concentration.
Assessment of Acyl-CoA Substrate Specificity-The acyl chain length selectivities of Nmt1p, 6XHis-Nmt1p, and each 6XHis-nmt1p were compared at 24  varying amounts of the competitive inhibitor S-(2-oxo)pentadecyl-CoA (0.005-1 M), and purified Nmt1p, 6XHis-Nmt1p, or a 6XHis-nmt1p. The K m Ј at each inhibitor concentration and type of inhibition were determined using double reciprocal plots. K i values were calculated from a plot of K m Ј versus the inhibitor concentration.
General Comments about the Kinetic Analyses-Each 6XHis-tagged enzyme preparation was studied at each temperature using purified Nmt1p without a 6XHis tag as a reference control. Analysis of myristoylpeptide production by the Nmts at 24 and 37°C indicated that the enzymes were stable over the course of the 10-min incubation period. The kinetic studies involved over 3,000 datapoints, each obtained either in duplicate or triplicate. The mean percentage difference between duplicate datapoints was 9.9 Ϯ 9.4 (S.D.). The mean percentage difference between triplicate datapoints was 12.8 Ϯ 11.6.

Gel Mobility Shift Assay for Assessing the Degree of N-Myristoylation of Arf1p in Strains Containing
Wild Type or Mutant Nmts pBB290 in YB523 (nmt1⌬) was replaced with low copy centromeric pRS315-based plasmids containing NMT1 or one of the ts nmt1 alleles. One hundred-milliliter cultures of the various strains were grown at 24°C in YPD to an A 600 ϭ 0.8. Twenty-five-milliliter aliquots were removed and incubated at 24 or 37°C for 2 h. Cell lysates were prepared using the same protocol employed for measurement of steady-state Nmt levels. Western blots were probed with rabbit antiserum R23 raised against a peptide (SNSLKNST) encompassing the C-terminal eight residues of Arf1p (kindly supplied by Richard Kahn, Emory University; final dilution ϭ 1:2000 in Blotto. Note that R23 does not cross-react with Arf2p, whose C-terminal sequence is SNNLKNQS). 2 Antigenantibody complexes were visualized as described above.

Random PCR Mutagenesis and a Colony Color Sectoring Assay Yields a Series of nmt1 Alleles Encoding Temperature-sensitive Mutants
Specific regions of NMT1 were targeted for random mutagenesis using in vivo recombination to repair a gapped plasmid with PCR fragments (Figs. 1 and 2). A colony color sectoring assay (35,36) was used to identify nmt1 alleles encoding mutant enzymes with reduced activity. This assay (Fig. 2) takes advantage of the fact that two mutations within the S. cerevisiae purine nucleotide biosynthetic pathway, when present in different combinations, give rise to colonies with different colors. Mutations in ADE2 cause accumulation of the chromophore phosphoribosylaminoimidazole, resulting in red colonies. Strains with ADE3 mutations do not accumulate the chromophore and grow as white colonies. ADE3 mutations also affect an enzymatic step upstream of Ade2p. Therefore, ade2 ade3 strains grow as white colonies. Introduction of a low copy centromeric plasmid containing NMT1 and ADE3 into an ade2 ade3 nmt1⌬ strain produces cells that form red colonies (Fig.  2). Since NMT1 is essential, if these cells lose this plasmid they die. Viable colonies containing the episomal copy of NMT1 and ADE3 remain red. If a second episome is introduced containing products from the PCR mutagenesis, one of two color phenotypes will be observed. If the mutation does not reduce Nmt1p's activity below the point required to acylate essential cellular N-myristoylproteins at levels compatible with viability, then there will be no selective pressure to retain the plasmid; i.e. during the course of cell division, 5-20% of newly formed daughter cells will lose the NMT1 ADE3 plasmid, and colonies will appear as red circles containing white sectors (sectoring phenotype). If a mutation inactivates NMT1, then the colony will be forced to retain the NMT1 ADE3 plasmid and will 2 Further details concerning this gel shift assay will be published elsewhere, including its validation using isogenic strains with NMT1 or nmt1-451D plus various combinations of wild type or null alleles of ARF1 and ARF2 (Lodge, J., Jackson-Machelski, E., Devadas, B., Kishore, N., Freeman, S., McWherter, C., Sikorski, J., and Gordon, J., (1997) Microbiology, in press). remain red (nonsectoring phenotype). Replica plating nonsectoring red colonies onto 5-FOA-containing plates will force the nmt1⌬ ade2 ade3 leu2⌬ host strain to remove the URA3-containing NMT1 ADE3 episome, thereby revealing the phenotype produced by the remaining LEU2 episome with its mutant nmt1 allele (Fig. 2).
Three overlapping regions of NMT1's ORF (Met 1 3 Leu 345 , Ala 202 3 Asn 397 , and Lys 389 3 Leu 455 ) or the entire ORF were replaced by DNA fragments generated using standard conditions for the polymerase chain reaction or conditions designed to increase the error rate of Taq polymerase. A total of 52,000 transformants were screened; 2122 transformants (4%) had a nonsectoring phenotype, 1093 of the nonsectoring colonies (51%) had a lethal phenotype when incubated at 24°C on synthetic complete media containing 5-FOA but lacking leucine, and 35 nonsectoring colonies had a ts phenotype based on differences in their growth in this media at 24 and 35 or 40°C (Fig. 1B).
Polymerase chain reactions 1-3, performed under standard conditions, produced 47 nonsectoring transformants with a lethal phenotype and two with a ts phenotype (Fig. 1, A and B). Plasmids were recovered from 38 of the transformants with a lethal phenotype: 25 of the 38 did not contain any NMT1 sequences and presumably arose from self-ligation of the gapped plasmids. Sequence analysis of the remaining 13 plasmids revealed that they encoded truncated enzymes. Nine contained insertions or deletions that produced frameshifts, resulting in the introduction of three to nine amino acid substitutions followed by a stop at Val 214 , Val 310 , Leu 398 , Lys 399 , Ser 400 , Leu 411 , or Leu 420 ; four contained single base changes that produced a stop codon at Glu 317 , Ser 331 , Gln 402 , and Leu 406 . The two plasmids that produced a ts phenotype contained nmt1 alleles with single amino acid substitutions affecting conserved residues: Ser 328 3 Pro (nmt1-328P) and Asn 404 3 Tyr (nmt1-404Y ) (Fig. 3, A and B).
The number of nonsectoring ts colonies obtained with a given set of primers was affected by the PCR conditions employed and by the size of the region of NMT1 targeted for replacement by in vivo repair of the gapped plasmid. Manipulating these two parameters (reactions 4 -8 in Fig. 1B) yielded 33 additional plasmids that conferred a ts phenotype. Three plasmids had nmt1 alleles with single amino acid substitutions affecting absolutely conserved residues: Ala 202 3 Thr (nmt1-202T ), Cys 217 3 Arg (nmt1-217R), and Asn 426 3 Ile (nmt1-426I) (Fig.  3, A and B). Nucleotide sequence analysis of the remaining plasmids revealed 24 unique nmt1 alleles, all of which contained multiple amino acid substitutions (range ϭ 2-6 substitutions). A total of 64 amino acids were altered in these 24 alleles. Of these 64 residues, 6 were mutated in at least two different alleles: Ala 202 , Lys 389 , Val 395 , Leu 408 , Asp 417 , and Asn 426 . All of these amino acids are conserved (Fig. 3B). Two of these substitutions, Ala 202 3 Thr and Asn 426 3 Ile, were also represented in other ts nmt1 alleles with single amino acid changes (see above). Moreover, mutations that affected four of the six residues did not always result in the same amino acid substitution: Ala 202 was replaced with either Thr or Gly, Lys 389 with Asn or Ile, Val 395 with Asp, Ala, or Gly, and Asp 417 with Asn or Val. Together, these findings suggested that the PCR mutagenesis of NMT1 was comprehensive.
The remaining 1046 nonsectoring colonies with lethal phenotypes were generated by using modified PCR conditions designed to increase the error rate of Taq polymerase or by targeting the entire NMT1 ORF. There was no obvious difference in the frequency of nonsectoring colonies with lethal phenotypes when the modified PCR conditions were used to target Met 1 3 Leu 345 , Ala 202 3 Asn 397 , or Lys 389 3 Leu 455 (reactions 4 -6 in Fig. 1B). Our analyses of the 35 nonsectoring colonies with a ts phenotype and the 47 nonsectoring colonies with a lethal phenotype suggested that the majority of the nmt1 alleles that would be recovered from these 1046 colonies would contain multiple amino acid substitutions and/or stop codons. Therefore, they were not analyzed further.

Steady-state Levels of Mutant Nmts at Permissive and
Nonpermissive Temperatures nmt1⌬ cells, containing episomal copies of one of the eight ts nmt1 alleles with single amino acid substitutions, were incubated in YPD broth at the permissive temperature (24°C) until they reached mid-log phase. The temperature was then raised to 37-40°C for 2 h. Western blots of total cellular proteins were prepared and probed with a previously characterized rabbit anti-Nmt sera (19). The steady-state levels of six of the mutant acyltransferases were equivalent to each other and to Nmt1p (Fig. 4B). The concentration of nmt420Sp at 24°C was equal to Nmt1p but was 50% that of the wild type enzyme at 37-40°C. nmt328Pp was ϳ3-fold less abundant than Nmt1p at permissive and nonpermissive temperatures (Fig. 4B). Thus, with the exception of nmt1-328P and nmt1-420S, differences in temperature sensitivity produced by the mutant nmt1 alleles could not be simply ascribed to differences in the steady-state levels of their protein products.

In Vitro Kinetic Studies of Purified Wild Type and Mutant Nmts
As noted above, five of the eight ts nmt1 alleles produced growth arrest on YPD at Յ39°C. To define the effects of their Ala 202 3 Thr, Cys 217 3 Arg, Ser 328 3 Pro, Asn 404 3 Tyr, and Asn 426 3 Ile substitutions on substrate recognition and/or catalysis, each mutant was expressed in E. coli, a bacterium with no endogenous Nmt activity (37). Nmt1p and nmt451Dp were used as reference controls. A genetically engineered NH 2 -terminal tag of six histidine residues (6XHis) allowed rapid purification of wild type and mutant enzymes by Ni 2ϩ -NTA-agarose affinity chromatography. The purification protocol produced a 100-1000-fold increase in specific activity over what was observed in unfractionated bacterial lysates and yielded apparently homogeneous preparations of each enzyme based on Coomassie staining of SDS-polyacrylamide gels (data not shown). the specificity of the acyltransferase reaction at 24 or 37°C (range of lauroyltransferase activity ϭ 2-15% of myristoyltransferase activity; palmitoyltransferase activity ϭ 0 -8%; n ϭ 2 independent preparations of each mutant, each assayed on two separate occasions in triplicate).

Effects of the Amino Acid Substitutions on the Functional
Having established that all of the mutants "retained" their character as myristoyltransferases, we examined the effects of the amino acid substitutions on the enzyme's affinity for S-(2oxo)pentadecyl-CoA (40). The rationale was as follows. None of the mutants could be purified in sufficient quantity or were sufficiently stable at 24 or 37°C to permit thermodynamic analysis of myristoyl-CoA binding by isothermal titration calorimetry (ITC), as has been done with wild type apo-Nmt1p (16). For an ordered reaction such as the one catalyzed by Nmt1p, a competitive inhibitor for the first of the two substrates, obtained at any concentration of the second substrate, is a dissociation constant for E ϩ I. S-(2-Oxo)pentadecyl-CoA is a competitive Nmt inhibitor containing a single methylene insertion between the CoA sulfur and the fatty acid carbonyl carbon of myristoyl-CoA. Determination of its K i required less enzyme and could be performed more rapidly than ITC analysis of myristoyl-CoA binding. Moreover, ITC has established that purified apo-Nmt1p perceives S-(2-oxo)pentadecyl-CoA as a close approximation of myristoyl-CoA: the enthalpy of binding of the nonhydrolyzable inhibitor (Ϫ25.8 kcal/mol) is equivalent to that of myristoyl-CoA (Ϫ24.8 kcal/mol) (Ref. 16). 3 The K i for S-(2-oxo)pentadecyl-CoA was defined at 24 and 37°C using the purified mutant Nmts, varying amounts of [ 3 H]myristoyl-CoA, and an Nmt1p substrate (GNSGSKQH-NH 2 ; Ref. 22) derived from the NH 2 -terminal sequence of Ppz2p, a protein phosphatase that suppresses the cell lysis defect produced by a protein kinase C null allele (41,42). The concentration of peptide was 2-100-fold over its K m for each mutant Nmt (see below). For Nmt1p, the K i is 5 nM (Table I), a value identical to the K d obtained from ITC (16). The addition of an NH 2 -terminal 6XHis tag to Nmt1p has no appreciable effect on its affinity for S-(2-oxo)pentadecyl-CoA (K i ϭ 4 -7 nM). The Cys 217 3 Arg substitution in nmt217Rp does not produce any detectable alteration in affinity for the inhibitor (5 nM at 24 and 37°C). In contrast, Asn 426 3 Ile results in the greatest reduction in affinity, with K i increases of 12-and 20-fold at 24 and 37°C (compared with 6XHis-Nmt1p). Asn 404 3 Tyr produces no change in affinity at 24°C but a 5-fold reduction at 37°C (K i ϭ 21 nM). The other substitutions result in ϳ3-fold increases in K i at 24°C and 6 -8-fold increases at 37°C (Table I).
Effects of the Amino Acid Substitutions on the Functional Properties of Nmt's Peptide Binding Site-As noted in the Introduction, once a binary myristoyl-CoA⅐Nmt1p complex forms, an allosteric transition occurs that allows formation of a functional peptide binding site. Previous in vitro and in vivo studies using myristic acid analogs have emphasized that changes in interactions between apo-Nmt1p and its acyl-CoA substrates can produce changes in the functional characteristics of the enzyme's peptide binding site (16,43,44). Therefore, we surveyed myristoylpeptide production at 24 and 37°C using Nmt1p, 6XHis-Nmt1p, each of the five mutants, saturating amounts of [ 3 H]myristoyl-CoA, and a panel of five octapeptide substrates. Two of the peptides had been used for defining each enzyme's acyl chain length selectivity and its affinity for S-(2oxo)pentadecyl-CoA (GARASVLS-NH 2 from human immunodeficiency virus type I Pr55 gag and GNSGSKQH-NH 2 from Ppz2p, respectively).
The addition of a 6XHis tag to Nmt1p has no demonstrable effect on peptide specificity as judged by myristoylpeptide production at 24 and 37°C. When compared with 6XHis-Nmt1p, the specific activity of each 6XHis-nmt1p mutant is reduced at 24 and 37°C for all peptides surveyed. For example, in the case of the Ppz2p peptide, the decrease ranges from 10-to 50-fold at 24°C and from 15-to 300-fold at 37°C. The extent of the FIG. 4. Assessing the stability of mutant nmt1ps at permissive and nonpermissive temperatures. A, equal numbers of nmt1⌬ cells, transformed with low copy pRS315-based centromeric plasmids containing the indicated wild type and mutant NMT1 alleles, were replicaplated onto YPD and incubated for 3 days at the indicated temperatures. The results shown are representative of those obtained in six independent experiments, each done in duplicate. B, the strains shown in A were grown to mid-log phase in YPD at 24°C and shifted to 37°C for 2 h. Western blots of cell lysates (100 g of total protein/lane) were probed with rabbit anti-Nmt sera. Antigen-antibody complexes were visualized using ECL. The position of migration of the intact 52-kDa N-myristoyltransferase is indicated. Preincubation of Nmt antibodies with purified Nmt1p blocked their subsequent reaction with the 52-kDa polypeptide (data not shown). Each blot was subsequently stripped and reprobed with antibodies to yeast actin. The results (not shown) indicated that each lane contained equal amounts of actin. reduction in specific activity varies between the mutants and as a function of the peptide substrate. Two peptides were selected for more detailed kinetic analysis: GNSGSKQHP-NH 2 (Ppz2p) and GAAPSKIV-NH 2 (representing the NH 2 -terminal sequence of Cnb1p, the N-myristoylated regulatory subunit of yeast Ca 2ϩ /calmodulin-dependent phosphoprotein phosphatase (45,46)). They were chosen because all of the mutants were able to generate readily detectable amounts of these myristoylpeptides at 24 and 37°C and because Ppz2p had been used when computing K i values for S-(2-oxo)pentadecyl-CoA.
Asn 404 3 Tyr, which produces the greatest temperature-dependent change in K i for S-(2-oxo)pentadecyl-CoA among the mutants (Table I), is associated with perturbations in peptide K m that also worsen with increasing temperature (Table II).
Ala 202 3 Thr, which produces 3-6-fold decreases in S-(2oxo)pentadecyl-CoA affinity at 24 and 37°C, results in increases in the K m for both peptides. The increase is more pronounced for the Ppz2p peptide (30 -50-fold) than for the Cnb1p peptide (6 -13-fold; Table II). Ser 328 3 Pro, which results in a similar 3-6-fold decrease in affinity for the nonhydrolyzable analog as Ala 202 3 Thr, has more modest effects on peptide K m ; the K m of the Ppz2p octapeptide is only increased 2-4-fold, and there is no significant change for the Cnb1p peptide (Table II).
All of the amino acid substitutions result in decreases in peptide V max . Ser 328 3 Pro produces the greatest reductions: ϳ350and 500-fold for the Cnb1p and Ppz2p peptides, respectively (Table II).
We selected the two mutants with the greatest increases in peptide K m (nmt217Rp (Cys 3 Arg) and nmt202Tp (Ala 3 Thr)) for additional analysis. Increases in peptide K m cannot be used to conclude that there are decreases in the mutant enzymes' affinities for these substrates. For an inhibitor of the second substrate of an ordered reaction, the inhibition constant is a dissociation constant under conditions where the first substrate is at a saturating level. Therefore, the effects of these two amino acid substitutions on the functioning of Nmt's peptide binding site were explored by determining the K i for a peptidomimetic inhibitor. SC-58272 was derived from the NH 2terminal eight residues of a yeast N-myristoylprotein, ADPribosylation factor-2 (Arf2p; Ref 7). Amino acids 1-4 were replaced with a para-(2-methylimidazole-N-butyl) phenylacetic acid derivative, amino acids 7 and 8 (Leu-Ser-NH 2 ) were deleted, and the 2-cyclohexylethyl amide derivative of lysine was used as the C-terminal residue (see Fig. 5). SC-58272 is competitive for binding of peptide substrates but not for myristoyl-CoA (7). ITC has been used to determine that the K d of SC-58272 for myristoyl-CoA⅐Nmt1p binary complexes is 29 nM. 3 Table III presents K i values for SC-58272 defined in the presence of saturating amounts of [ 3 H]myristoyl-CoA and varying amounts of the Ppz2p peptide. With Nmt1p, the K i at 24°C was very similar to the K d defined calorimetrically (43 versus 29 nM). The addition of a 6XHis tag to Nmt1p produces Ͻ2-fold changes in the K i at 24 and 37°C. Ala 202 3 Thr results in K i increases of 6-and 9-fold at 24 and 37°C compared with 6XHis-Nmt1p. Cys 217 3 Arg produces 3-and 6-fold increases at these two temperatures.
Conclusions from the Kinetic Studies-The finding that Cys 217 3 Arg produces an increase in the K i for SC-58272 but no change in the K i for S-(2-oxo)pentadecyl-CoA indicates that this substitution of a conserved Cys has selective effects on the functional properties of the enzyme's peptide binding site. In contrast, substitution of the conserved Asn 426 with Ile appears to selectively affect the functional properties of the enzyme's myristoyl-CoA binding site. Ala 202 3 Thr, which produces the most severe ts phenotype, provides an example of a substitution that affects the functional properties of both sites, resulting in increases the K i for S-(2-oxo)pentadecyl-CoA and SC-58272. Such a substitution may perturb specific contacts between the enzyme and the myristoyl or CoA moieties, which, in turn, could disturb the subsequent allosteric transition required for creation of a fully functional peptide binding site. Alternatively, Ala 202 3 Thr may produce more global changes in the conformation of Nmt1p.

Defining Levels of Acylation of an N-Myristoylprotein in NMT1 and nmt1 Cells
At 37°C, adding 500 M myristate to YPD media rescues growth of all but one of the ts nmt1 mutants (data not shown). (nmt1-328P is the exception. Recall that at 37°C nmt328Pp has the lowest steady-state cellular concentration among the mutants (Fig. 4B), the greatest reduction in peptide V max in vitro (Table II), and a 6-fold decrease in affinity for the myristoyl-CoA analog.) At 39 -40°C, adding 500 M myristate to YPD only partially rescues nmt1⌬ cells with nmt1-426I (lowest affinity for S-(2oxo)pentadecyl-CoA among the mutants), nmt1-202T (lowest affinity for SC-58272 plus moderate reductions in affinity for S-(2-oxo)pentadecyl-CoA), or nmt1-404Y (most marked change in S-(2-oxo)pentadecyl-CoA affinity between 24 and 37°C). Levels of cellular protein N-myristoylation produced by the wild type and mutant nmt1 alleles were defined under these various growth conditions using an Arf gel mobility shift assay. Arf1p and Arf2p are two functionally interchangeable but essential N-myristoylproteins involved in vesicular trafficking (47,48). They depend upon N-myristoylation for expression of their biological functions (49). Arf1p represents ϳ90% of cellular Arfs (47). N-Myristoylation produces a change in the electrophoretic mobility of Arf1p during SDS-PAGE; the acylated species migrates more rapidly than the nonmyristoylated species (19). We therefore reasoned that Arf1p could be used to report the extent of cellular protein N-myristoylation by noting the ratio of its distinctively migrating N-myristoylated and nonmyristoylated forms.
nmt1⌬ cells containing plasmids with NMT1 or ts nmt1 inserts were grown in YPD at 24°C to mid-log phase. Aliquots of the cultures were then incubated for an additional 2 h at 24°C or 33-40°C in YPD alone or in YPD plus 500 M myristate. Western blots of total cellular proteins were prepared and probed with rabbit antibodies that react with Arf1p but not Arf2p, 2 and the ratio of N-myristoylated to nonmyristoylated Arf1p was determined. Fig. 6 shows representative results obtained with NMT1, nmt1-404Y, and nmt328P. Only myristoyl-Arf1p is detectable in NMT1 cells, whether they are cul-tured in YPD at 24, 37, or 40°C. Defects in protein N-myristoylation are evident in nmt1-404Y cells when cultured in YPD alone. Even at the permissive temperature of 24°C, only ϳ60% of Arf1p is acylated; at 37°C, the value falls below 50%. In nmt1-328P cells, Ͼ95% of Arf1p is N-myristoylated when grown at 24°C on YPD; however, between 33 and 37°C, myristoyl-Arf1p decreases to 40% of total Arf1p. The ability of myristate to increase levels of myristoyl-Arf1p to Ն50% of total cellular Arf1p can be directly correlated with its ability to rescue the growth of a given nmt1 strain. At 37°C, myristoyl-Arf1p levels increase in nmt1-404Y cells from Ͻ50% to Ͼ95% 2 h after exposure to myristate (Fig. 6). These cells are able to sustain growth in this media at this temperature. In contrast, myristate has no detectable beneficial effects on the efficiency of Arf1p acylation in nmt1-328P cells at 37°C (Fig. 6), a temperature where they are unable to survive on YPD/myristate. The gel shift assay was used to establish that a reduction in levels of myristoyl-Arf1p to Յ50% is also associated with a failure of cells containing the other ts nmt1 alleles to survive (data not shown).

Prospectus
A panel of mutant Nmts, with defined defects in the functioning of their myristoyl-CoA and/or peptide binding sites has been generated using an error-prone PCR strategy. A full understanding of the structural significance of their amino acid substitutions must await determination of the atomic structure of Nmt1p with and without bound substrates or substrate analogs. Nonetheless, the temperature-sensitive growth arrest produced by various members of this panel offers an opportunity to conduct genetic screens for factors that affect Nmt activity, to examine the adaptive responses of cells to undermyristoylation of cellular proteins, and to identify the functions of cellular N-myristoylproteins in dividing and nondividing yeast cells.   N-myristoylation in nmt1⌬ cells containing NMT1, nmt1-328P, and nmt1-404Y episomes. Strains were grown at 24°C in YPD until they reached an A 600 of 0.8. The culture was divided into aliquots, which were then incubated at the indicated temperatures for an additional 2 h in YPD or in YPD plus 500 M myristate. Total cellular proteins (100 g of protein/lane) were fractionated by SDS-PAGE. Western blots were prepared and probed with a rabbit anti-Arf1p sera. The results shown are representative of those obtained in three independent experiments.