Analysis of the compartmentalization of myristoyl-CoA:protein N-myristoyltransferase in Saccharomyces cerevisiae.

Myristoyl-CoA:protein N-myristoyltransferase (NMT) catalyzes the cotranslational, covalent attachment of a rare fatty acid, myristic acid (C14:0), to the amino-terminal glycine residue of a number of eukaryotic proteins involved in cellular growth and signal transduction as well as several viral proteins necessary for assembly-replication. NMT has become a target for both anti-viral and anti-fungal therapy. Analysis of purified Saccharomyces cerevisiae NMT plus yeast strains with conditional lethal nmt1 mutations have provided insights about how this process is regulated in vivo. We have now defined the location of NMT in two strains of S. cerevisiae to better understand the functional and spatial relationships between this enzyme and cellular systems that generate its acyl-CoA and peptide ligands. Western blot studies using an affinity purified antibody raised in rabbits against purified S. cerevisiae NMT indicate that the acyltransferase represents 0.06% of total cellular proteins in an exponentially growing haploid strain with a wild type NMT1 allele. Another strain containing a single, integrated copy of a GAL1/NMT1 fusion gene and a nmt1 null allele had 12-fold higher levels of NMT when grown on galactose-containing media. This increase in NMT production had no detectable effects on growth or cellular morphology. Cell fractionation studies, confocal fluorescence immunocytochemical analysis, and immunogold electron microscopic surveys of fixed, gelatin-embedded cryosections of both strains revealed that NMT is a cytosolic protein that is not associated with cellular membranes (including the endoplasmic reticulum and plasma membrane), the nucleus, mitochondria, Golgi apparatus, or vacuoles. This finding is discussed in light of what is known about the location and activities of enzymes involved in de novo fatty acid biosynthesis and in amino-terminal processing of nascent proteins.

Myristoyl-CoA:protein N-myristoyltransferase (NMT) catalyzes the cotranslational, covalent attachment of a rare fatty acid, myristic acid (C14:0), to the amino-terminal glycine residue of a number of eukaryotic proteins involved in cellular growth and signal transduction as well as several viral proteins necessary for assembly-replication. NMT has become a target for both anti-viral and anti-fungal therapy. Analysis of purified Saccharomyces cerevisiae NMT plus yeast strains with conditional lethal nmtl mutations have provided insights about how this process is regulated in vivo. We have now defined the location of NMT in two strains of S. cerevisiae to better understand the functional and spatial relationships between this enzyme and cellular systems that generate its acyl-CoA and peptide ligands. Western blot studies using an affinity purified antibody raised in rabbits against purified s. cerevisiae NMT indicate that the acyltransferase represents 0.06% of total cellular proteins in an exponentially growing haploid strain with a wild type NMTl allele. Another strain containing a single, integrated copy of a GALlINMTl fusion gene and a nmtl null allele had 12-fold higher levels of NMT when grown on galactose-containing media. This increase in NMT production had no detectable effects on growth or cellular morphology. Cell fractionation studies, confocal fluorescence immunocytochemical analysis, and immunogold electron microscopic surveys of fixed, gelatin-embedded cryosections of both strains revealed that NMT is a cytosolic protein that is not associated with cellular membranes (including the endoplasmic reticulum and plasma membrane), the nucleus, mitochondria, Golgi apparatus, or vacuoles. This finding is discussed in light of what is known about the location and activities of enzymes involved in de novo fatty acid biosynthesis and in amino-terminal processing of nascent proteins.
Cotranslational (Wilcox et al., 1987;Deichaite et al., 1988), covalent linkage of myristate (tetradecanoate, C140) t o the amino-terminal glycine residue of a variety of mammalian * This work was supported in part by National Institutes of Health Grants A130188 and A127179 and the Monsanto Company. The Washington University Confocal Microscopy Facility is supported by a grant from the Lucille P. Markey Foundation. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
n To whom correspondence should be addressed Dept. of Molecular Biology and Pharmacology, Box 8103, Washington University School of Medicine, 660 S. Euclid Ave., St. Louis, and viral proteins can have an important effect on their biological function (reviewed in Towler et al., 1988a;James and Olson, 1990;. For example, the myristoyl moiety has been shown (i) to be necessary for the binding of p60v-src to a 32-kDa plasma membrane receptor (Resh and Ling, 1990) and for this tyrosine kinase to transform cells (Cross et al., 1984;Kamps et al., 1985); (ii) to increase the affinity of a G protein a subunit for the @/r subunits (Linder et al., 1991); and (iii) to be required for the assembly of a number of retroviruses by virtue of its linkage to their gag polyprotein precursors (these include the P r 6 P g of the Moloney murine leukemia virus, the Pr75g*g of the Mason Pfizer monkey virus, and the Pr55gag of human immunodeficiency virus I (Rein et al., 1986;Rhee and Hunter, 1987;Gottlinger et al., 1989;Bryant and Ratner, 1990)). The latter observation has made the enzyme responsible for this modification, myristoyl-CoAprotein N-myristoyltransferase (NMT),' a target for anti-viral therapy either through the use of inhibitors (Shoji et al., 1988) or alternative fatty acid substrates whose physical chemical properties are different from those of myristate (Bryant et al., 1989(Bryant et al., ,1991. Insertional mutagenesis or deletion of the S. cerevisiue NMTl gene causes recessive lethality (Duronio et al., , 1991b, suggesting that NMT may also be a useful target for anti-fungal therapy. The genetic manipulability of Saccharomyces cerevisiae makes it an attractive system for analyzing how protein Nmyristoylation is regulated in vivo. S. cereuisiae NMTl encodes a 455-residue monomeric protein (Towler et al., 198713;Duronio et al., 1989) that catalyzes the transfer of myristate from myristoyl-CoA to -12 endogenous proteins (Duronio et al., 1991b). Five of these substrates have been identified to date. They include (i) vacuolar sorting protein 15 (Vpsl5, Herman et al., 1991), a serine-threonine kinase which is essential for growth at 37 "C; (ii) Gpal, a 55-kDa, haploid essential gene product that depends upon its myristoyl moiety to serve as a G protein a subunit homolog in the mating pheromone signal transduction pathway (Blumer and Thorner, 1991;Stone et al., 1991); (iii) two ADP-ribosylation factors, Arfl and Arf2, that are highly conserved structurally and functionally, are required for proper glycosylation of proteins in the secretory pathway, and whose combined absence is associated with a lethal phenotype (Stearns et al., 1990a(Stearns et al., , 1990b; and (iv) a nonessential 16-kDa protein which is the homolog of the regulatory B subunit of the mammalian protein phosphatase, calcineurin (Cnbl, Cyert et al., 1991). Purified S. cerevisiue NMT has no intrinsic methionylaminopeptidase (MAP) activity (Towler et al., 1987b) indicating that the initiator Met residue must be removed from these proteins by cellular MAP to expose their Gly' residues prior The abbreviations used are: NMT, myristoyl-CoAprotein Nmyristoyltransferase; MAP, methionylaminopeptidase; PBS, phosphate-buffered saline; BSA, bovine serum albumin. to addition of myristate. Kinetic studies have shown that S. cereuisiae NMT has a very high degree of specificity for 14 carbon fatty acids that is achieved, in part, because of cooperative interactions between its acyl-CoA and peptide-binding sites (Heuckeroth et al., 1988(Heuckeroth et al., , 1990Kishore et al., 1991). Moreover, NMTs mechanism is ordered Bi Bi  with myristoyl-CoA binding occurring prior to peptide binding and CoA release taking place prior to release of the acylpeptide product. This mechanism predicts that perturbations in intracellular pools of myristoyl-CoA or NMT's ability to gain access to them will have important effects on the efficiency of protein acylation in uiuo. This prediction was subsequently shown to be true: a Gly-451 + Asp substitution in NMT results in a temperature-dependent, 10-fold increase in the K,,, for myristoyl-CoA (compared to the wild type enzyme) and a 600fold reduction in specific activity (Duronio et al., 1991b). S. cereuisiae strains containing this mutant nmtl allele (nmtl-181) exhibit growth arrest and myristic acid auxotrophy in rich media at the nonpermissive temperature (37 "C, Meyer and Schweizer, 1974;Duronio et al., 1991b). Rescue of these strains by exogenous myristate requires acyl-CoA synthetase (Faal, Kamiryo et al., 1977) to restore intracellular myristoyl-CoA pools to a size sufficient to permit acylation of critical myristoylproteins at an efficiency needed for normal vegetative growth. Blockade of de novo fatty acid synthesis by cerulenin, an inhibitor of the fatty acid synthetase (Fas) complex (Lynen, 1980;Vance, 1972), causes growth arrest of nmtl-181-containing strains which can be relieved by myristate but not by palmitate (Duronio et al., 1991b). The failure of palmitate to rescue growth (either in the presence or absence of cerulenin) suggests that it cannot adequately replete endogenous myristoyl-CoA pools via metabolic interconversion (Duronio et al., 1991b). It is interesting to note that the K,,, of palmitoyl-CoA for purified S. cereuisiae NMT is equivalent to that of myristoyl-CoA, but that the efficiency of transfer of C16:O to a variety of peptide substrates is less than 5% that of C14:O and palmitoyl-CoA is a competitive inhibitor of NMT (Towler et al., 1987a;Heuckeroth et al., 1988;Rudnick et al., 1990). This raises the question as to how NMT is able to avoid inhibition by palmitoyl-CoA in uiuo.
In the present report, we describe a series of studies designed to ascertain the intracellular location of S. cereuisiae NMT. These experiments were initiated to test the hypothesis (Duronio et al., 1990a(Duronio et al., , 1991a that protein N-myristoylation is modulated, at least in part, by the spatial and functional relationships that exist in vivo between NMT, its nascent polypeptide and acyl-CoA ligands, and the systems which produce these substrates (e.g. acyl-CoA generating enzymes, the translational apparatus, and other amino-terminal modifying enzymes such as MAP).
Preparation of Affinity Purified Rabbit Anti-S. cerevisiae NMT Antibodies-The specificity of a rabbit polyclonal antisera raised against purified Escherichia coli-derived S. cerevisiue NMT (Rudnick et al., 1990) has been described in previous publications (e.g. see Fig.  7 in Duronio et al., 1991b). NMT antibodies were purified from this sera by antigen-blot affinity chromatography using the protocol of Olmsted (1986). The final concentration of the antibody preparation eluted from the filter with 0.1 M glycine-HC1 was 100 pg/ml. Subcellular Fractionation-Both strains were grown at 30 "C to an ODsw of 1 prior to harvesting by centrifugation (Duronio et al., 1991b).
Lysates of S. cerevisiae were prepared according to Walworth et al. (1989) except that lyticase (an enzyme from Arthrobacter luteus that hydrolyzes yeast cell wall glucan at its 8-1,3 linkages, final concentration = 2 units/ml) was used to generate spheroplasts, and EDTA (1 mM) was included in the lysis buffer. The lysate was spun at 100,000 X g, , for 1 h at 4 "C. The supernatant was removed. The pellet fraction was washed with lysis buffer (Walworth et al., 1989) and dissolved in 0.5% Triton X-100. The protein concentration in the supernatant and Triton X-100-extracted pellet fraction was determined according to Lowry et al. (1951). An equal mass of protein from each fraction was heat denatured, reduced, and then subjected to sodium dodecyl sulfate-polyacrylamide gel electrophoresis (Laemmli, 1970). Note that purified, E. coli-derived S. cerevisiae NMT standards were included in each gel. Following electrophoretic transfer of the separated proteins to nitrocellulose membranes (Burnette, 1981), the resulting blots were probed with the affinity purified, rabbit anti-S. cerevisiae NMT antibody preparation diluted 1:lOO in Blotto (Burnette, 1981). Antigen-antibody complexes were visualized using lZ5I-protein A (Burnette, 1981). The concentration of NMT in each fraction was determined by scanning filter autoradiographs with a LKB UltroScan XL laser densitometer. Cytochrome c reductase (EC 1.6.99.3) activity was also measured in all fractions according to Sottocasa et al. (1967).
Confocal Immunofluorescence Microscopy-S. cerevisiue strains YM2061 and YB210 were grown and harvested as described above and subsequently fixed and processed according to Kilmartin and Adams (1984) but with the following modifications. Fixation was in 3.7% formaldehyde for 90 min at 23 "C. Cells were quenched in 10 mM ethanolamine for 15 min at 23 "C and then resuspended in buffer A (potassium phosphate (100 mM), sorbitol (1.2 M), pH 7.0); 1 ml of buffer, 100 pl of packed volume of cells). Lyticase was added to this solution (final concentration = 2 units/ml (Sigma)) together with 2 mercaptoethanol(20 mM), and the mixture was incubated for 60 min at 37 "C. The resulting spheroplasts were added to polylysine-coated multiwell slides and then permeabilized by immersion in methanol for 6 min followed by acetone for 30 s (both steps performed at -20 "C). Permeabilized cells were incubated for 2 h at room temperature with the affinity purified antibody preparation after it had been diluted 1:lO in buffer B (NaC1 (0.14 M), potassium phosphate (10 mM), sodium azide (1.5 mM), bovine serum albumin (BSA, Fisher Scientific, final concentration = 1 mg/ml), pH 7.0). Slides were washed in buffer B (minus the antibodies) for 5 min and subsequently incubated for 2 h at room temperature with rhodamine-conjugated anti-rabbit IgG (Jackson ImmunoResearch Laboratories, West Grove, P A diluted 1:500 with buffer B). Mounting solution was prepared following the procedure of Kilmartin and Adams (1984) except that 4',6-diamidino-2-phenylindole was omitted. Slides were viewed using a Bio-Rad MRC-500 Scanning Laser Confocal Microscope.
Electron Microscopic Immunocytochemical Studies-S. cerevisiae strains were harvested during log-phase growth and fixed overnight at 4 "C in 4% paraformaldehyde, 0.2% glutaraldehyde. They were embedded in a 10% (w/v) solution of porcine skin gelatin (Sigma), phosphate-buffered saline (PBS) as described by Geuze and Slot (1980), infused with 2.3 M sucrose, PBS for 2 days, and prepared for ultracryotomy using a modification of protocol of Tokuyasu (1980) detailed in Slot et al. (1988). Tissue was frozen in liquid nitrogen and thin (100 nm) sections cut at -90 "C with a Reichert-Jung UltraCut equipped with a FC4E Cryosystem (Leica, Inc., Deerfield, IL) Cryosections were placed on the surface of a 2.3 M sucrose drop to thaw and were then transferred to formvar carbon-coated nickel grids. Immunolabeling was accomplished by floating the grids on drops of the following solutions (in order): 1) 0.02% glycine-PBS (2 drops/5 min incubation/drop); 2) 0.1% BSA-PBS (2 X 5 min); 3) buffer B Subcellular Location of S. cerevisiae NMT (described above) containing affinity purified antibody (final concentration = 100 pg/ml, preadsorbed with intact YM2061 cells for 1 h on ice) or purified, preadsorbed rabbit IgG (Cappel, Durham, NC, final concentration in buffer B = 100 pg/ml) (1 X 30 min); 4) 0.1% BSA-PBS (5 X 2 min); and finally 5) protein A-gold (prepared in 0.1% fishgel (Sigma), 0.1% BSA-PBS). The protein A (Boehringer Mannheim) -gold conjugate was prepared using 10-nm gold particles and purified by glycerol gradient centrifugation according to Slot and Geuze (1985).
Immunolabeled cryosections were processed according to a modification of the Tokuyasu method introduced by Griffiths et al. (1984). They were stained for 10 min at 23 "C with a solution of 2% uranyl acetate, 0.15 M oxalic acid and then incubated for 10 min at the same temperature in 0.4% uranyl acetate, 2% methyl cellulose (25 centipoise). Specimens were viewed and photographed using a Zeiss model EM10 electron microscope.

RESULTS
How Much NMT Is Required to Sustain Normal Growth of S. cerevisiae?-Two strains of yeast were used for our analysis of the subcellular location of N M T YM2061, a MATa haploid strain with a single copy of NMTl, and YB210, a strain isogenic with YM2061 but containing a deletion of the NMTl locus and a single integrated copy (at the URA3 locus) of a fusion gene containing NMTl linked to the very strong, galactose-inducible GAL1 promoter (Johnston and Davis, 1984). YB210 grows on YPD medium at a rate similar to that of YM2061, indicating that glucose repression of GAL11 NMTl still permits production of sufficient NMT to acylate critical yeast N-myristoylproteins at a level than can overcome the inviable phenotype of nmtl null cells. Incubation of YB210 on YP-galactose produces no detectable alteration in growth rates at 24 or 36 "C when compared to incubation of YM2061 on YP-galactose at the same temperatures (data not shown).
Western blotting was used to determine the steady state levels of NMT in strains YB210 and YM2061 during exponential growth in YP-galactose and YPD (respectively) at 30 "C. Total cell lysates were probed with affinity purified rabbit antibodies directed against S. cerevisiae NMT expressed in, and purified from, E. coli (Rudnick et al., 1990).
The antibody preparation reacts with two polypeptides of 53 and 42 kDa present in YM2061 cells (Fig. 1, lane 2). The 53-kDa species predominates in galactose-induced YB210 cells (lane 3). Neither species can be detected when duplicate blots are probed with (i) rabbit IgG prepared from animals who had not been immunized with S. cerevisiae NMT (lane 4 of Fig.  1) or (ii) with affinity purified antibody that had been preincubated with a 10-fold molar excess of purified antigen (data not shown). The 42-kDa species may be derived from proteolytic processing of the 53-kDa species. Previous studies in E. coli strains containing an S. cerevisiae NMT expression vector indicated that the 455-residue product of NMTl  is rapidly cleaved at its Lys-39/Phe-40 bond within the bacterium without losing its biological activity (Duronio et al., 1990a). Using known amounts of purified S. cerevisiae NMT as standards, laser densitometric tracing of filter autoradiographs indicated that the intact 53-kDa NMT polypeptide in strain YM2061 represented 0.06 k 0.02% of total cellular proteins while the steady concentration of this acyltransferase in galactose-induced YB210 cells was 12-fold higher (0.71 k 0.1%, compare panels A and B in Fig. 2). The value obtained in strain YM2061 is consistent with the fact that an 11,000-fold purification is needed to obtain an apparently homogenous preparation of the 53-kDa monomeric enzyme from S. cerevisiae (Towler et al., 1987b).
The 53-kDa protein was not detectable in YB210 grown in YP-glucose (limits of detection = 0.01% of total cellular proteins, data not shown). This observation, coupled with the lethal phenotype of nmtl null mutants, indicates that haploid strains of S. cerevisiae with NMTl produce at least a 5-fold excess of NMT over that required to maintain vegetative growth.
Fractionation of S. cerevisiae Suggests That NMT Is Primarily a Cytosolic Protein-Total cell lysates were prepared and fractionated by ultracentrifugation using a well-characterized protocol (Walworth et al., 1989). The resulting 100,000 X gave supernatant and pellet fractions were assayed for NMT by Western blot hybridization and for NADPH cytochrome c reductase activity. The latter enzyme was selected because of its known location in the endoplasmic reticulum and Golgi fractions of S. cerevisiae (Walworth et al., 1989). The specific activity of cytochrome c reductase was 10-fold higher in the 100,000 X gave pellet (membrane) fraction compared to super- natant (cytosolic) fraction in strains YM2061 and YB210 (Fig. 2). In contrast, the concentration of NMT was 4-fold higher in the cytosolic compared to Triton X-100-extracted pellet fractions. This [NMT,,~,,a~,t]/[NMTpellet] ratio did not change even with the 12-fold increase in intracellular NMT concentration obtained by galactose induction of GAL11 NMTl (compare panels A and B in Fig. 2). Western blot analyses indicated that the ratio of the 53-42-kDa polypeptides was no different in the soluble and membrane fractions compared to that observed in whole cell lysates (in each strain, data not shown). RNA was isolated from both the 100,000 X gave supernatant and pellet fractions using guanidine isothiocyanate (Chirgwin et aL, 1979). Although some degradation was evident, denaturing formaldehyde-agarose gel electrophoresis (Thomas, 1980) indicated that -75% of ribosomal RNA was associated with the pellet (data not shown). The finding that most of the ribosomal RNA is located in the 100,000 x gave pellet while most of NMT in the supernatant suggested that the enzyme is not "tightly" associated with S. cereuisiae ribosomes.
Confocal Immunofluorescence Microscopy Reveals That NMT Is Diffusely Distributed in the Cytoplasm-The Western blot studies described above suggested that the affinity purified antibody could be used to detect NMT in our two S. cereoisiae strains. Incubation of permeabilized (see "Experimental Procedures") yeast spheroplasts with the antibody preparation revealed diffuse cytoplasmic staining without any discernible concentration of antigen at the nuclear or plasma membranes (Fig. 3). The overall intensity of cytoplasmic staining was greater in the NMT-overexpressing strain YB210 (panel B) than in YM2061 (panel A ) . The signal intensity was markedly diminished over nuclear-vacuolar structures creating a donut-like pattern of staining which is evident in Fig. 3, A and B. Incubation of spheroplasts prepared from these strains with preimmune rabbit serum or with NMT antibodies preadsorbed with a 10-fold molar excess of purified enzyme, produced no detectable signal ( Fig. 3C and data not shown). The data presented in Fig. 3 are not only consistent with the results of the cellular fractionation studies but also establish that a 12-fold increase in the steady state level of NMT has no discernible effect on cellular morphology as judged by confocal microscopy.
Electron Micrograph Immunocytochemical Analyses of Cryosections Obtained from S. cerevisiae Strain YM2061 and YB210 Show That NMT Is a Cytosolic Enzyme Which Is Not Associated with Cellular Membranes or Organelles-Fixation of yeast strains in 4% paraformaldehyde, 0.2% glutaraldehyde followed by embedding in 10% gelatin and cryosectioning allowed good visualization of membrane structure while still preserving the antigenicity of NMT. Immunogold labeling revealed that NMT was a cytoplasmic enzyme in both YM2061 and YB210 cells (panels A-C in Fig. 4). No signal above background was associated with nuclei, mitochondria, vacuoles, or cellular membranes including the plasma membrane, the Golgi apparatus, or the endoplasmic reticulum. Predictably, the density of gold particles was greater in the NMT-overexpressing strain (compare panels B and A , respectively). There was no obvious focal compartmentalization of NMT within the cytoplasm or around the periphery of organelles in either strain. We were not able to resolve ribosomes in our preparations and therefore could not determine whether NMT was affiliated with the non-endoplasmic reticulum fraction of cellular ribosomes (see "Discussion"). The specificity of the immunogold labeling was confirmed in control experiments using rabbit IgG and the GALlINMTl- containing strain grown in YP-galactose (compare panel D with panels A-C in Fig. 4). Finally, the data presented in Fig.  4 provide additional evidence that overexpression of NMT has no effect on cellular morphology.

DISCUSSION
A series of independent assays have been used to establish that myristoyl-CoAprotein N-myristoyltransferase is a cytoplasmic enzyme in S. cereuisiae. The cytoplasmic location of S. cerevisiae NMT is consistent with the fact that it does not have a definable (von Heijne, 1985) signal peptide for cotrans-

FIG. 4. Electron microscopic localization of NMT in YM2061 and YB210 strains. Growth conditions
were identical to those employed for the cellular fractionation and confocal immunofluorescence microscopic studies. Panel A, 100-nm cryostate sections of gelatin embedded, paraformaldehyde/glutaraldehyde-fixed YM2061 cells stained with NMT antibodies purified by antigen-blot affinity chromatography. Antigen-antibody complexes have been visualized using protein A-gold. Bar = 200 nm; panel B, YB210 cells stained with the same antibody lational translocation across the endoplasmic reticulum membrane, or an amino-terminal sequence with the features of a mitochondrial transit peptide (Roise and Schatz, 1988;Pfanner and Neupert, 1990), or multiple membrane-spanning domains (as predicted by the algorithm of Kyte and Doolittle, 1982). Like many cytosolic proteins, purified S. cerevisiae NMT has a blocked amino terminus (Towler et al., 1987b).' Our findings raise a number of questions about how NMT acquires its myristoyl-CoA and nascent polypeptide substrates.
Myristate is a rare fatty acid in S. cereuisiue comprising 1-2% of total cellular fatty acids (Orme et al., 1972;Awaya et al., 1975). The cellular concentrations of myristoyl-CoA and palmitoyl-CoA in this yeast have not been reported. All four enzyme systems that regulate the biosynthesis of long chain fatty acids in S. cereuisiue (acetyl-coA carboxylase, fatty acid synthetase, fatty acid desaturase, and the fatty acyl chain elongation system) are located in the cytoplasm. The two enzymes that catalyze the synthesis of long chain saturated fatty acids from acetate (acetyl-coA carboxylase (Accl, Mishina et al., 1980)) and the multisubunit, multifunctional fatty acid synthetase (Fas) complex (Schweizer et al., 1986(Schweizer et al., , 1987Chirala et al., 1987;Mohamed et al., 1988)) are soluble while the desaturase is associated with microsomes (reviewed in Schweizer et al., 1978). The exact pathways employed for maintaining myristoyl-CoA pools in S. cereuisiue have yet to be defined. The major products of Accl and Fas are palmitoyl-CoA and stearoyl-CoA (Lynen, 1980) but myristoyl-CoA has also been identified in their reaction products (Lynen, 1969). The malonyl-CoA-dependent, fatty acyl chain elongation system can reversibly add one to three Cz units to the acyl-CoA derivatives of medium chain saturated or unsaturated fatty acids (Erwin, 1973;Fulco, 1974;Orme et al., 1972;Schweizer et al., 1978), but its relative contribution to the myristoyl-CoA pools of S. cerevisiue strains grown on rich media remains unclear.
The availability of a mutant NMT that requires increased levels of myristoyl-CoA due to its reduced affinity for this ligand (i.e. nmt-181), have allowed us to conclude that NMT can access myristoyl-CoA derived from the endogenous pathway for de mvo fatty acid synthesis (see Introduction and Duronio et al., 1991b). The cytosolic locations of NMT, acetyl-CoA carboxylase, and the fatty acid synthetase complex are compatible with this notion. However, the ability of NMT to utilize what is presumed to be a minor product of these enzymes (myristoyl-CoA) while avoiding inhibition by its principal product (palmitoyl-CoA) would appear to require some form of functional segregation of NMT from these enzymes and/or from palmitoyl-CoA. Such segregation may be achieved, at least in part, by the different solubilities of myristoyl-CoA and palmitoyl-CoA (critical micellar concentrations = 210 and 42 PM, respectively; Smith and Powell, 1986) and differences in their ability to affiliate with membranes and/or intracellular binding proteins.
Analysis of S. cerevisiae strains containing nmtl-181 indicated that exogenous myristate could rescue their temperature-sensitive growth arrest in a process that required active acyl-CoA synthetase. This presumed mechanism for augmenting available myristoyl-CoA pools raises two additional Although the blocking group has not been defined, the enzyme lacks a GIy2 residue and therefore cannot be N-myristoylated.
issues concerning how myristoyl-Coh is delivered to NMT. First, the ability of exogenous myristate to rescue growth and to be incorporated into cellular N-myristoylproteins (Duronio et al., 1991b) indicates that it can be imported into S. cereuisiae and metabolically processed to myristoyl-CoA prior to interaction with NMT. The mechanism of import of long chain fatty acids into S. cereuisiue is unknown. It is tempting to speculate that this organism, which has a cell wall, may have a cellular apparatus similar to that present in Gramnegative bacteria such as E. coli for specific binding and transport of exogenous C12-Cl8 fatty acids. Uptake of these fatty acids in E. coli requires a 421-residue, multifunctional, outer-membrane transporter encoded by the f d L gene (Black et al., 1987;Black, 1991;Kumar and Black, 1991) and a functionally coupled acyl-CoA synthetase (encoded by fadD) which is loosely associated with the inner membrane (Klein et al., 1971;Kameda and Nunn, 1981). If a comparable apparatus exists in S. cereuisiae, the functional and spatial relationships of Faal and NMT should be further defined in strains that contain FAA1 as well as temperature-sensitive or nullfuul alleles. Second, the Gly --., Asp-451 mutation in nmt-181 not only produces a reduction in K,,, but also a marked change in its net charge: the PI of apo-NMT is 8.15 compared to a value of 6.9 for apo-nmtl81 (Duronio et al., 1991b). Addition of myristoyl-CoA to wild type apo-NMT produces a marked change in PI to 6.7. This is due to a high affinity interaction between this acyl-CoA and the acyltransferase (Rudnick et al., 1990). Addition of a severalfold molar excess of myristoyl-CoA to apo-nmtl81 produces no alteration in its PI (Duronio et al., 1991b). The net charge of NMT may influence its ability to interact with cellular proteins-enzymes that are involved in the production and/or delivery of myristoyl-CoA. It may also regulate NMTs ability to interact with the cellular apparatus that produces its nascent protein substrate.
The ordered Bi Bi reaction mechanism of NMT indicates that after binding myristoyl-CoA, NMT must gain access to its peptide substrate. To do so would seem to require that a series of events be carefully orchestrated in both space and time within S. cerevisiue. The initiator methionine residue of N-myristoylproteins must be removed by MAP to expose the GIy2 residue. The presence of a GlyZ in all N-myristoylproteins makes them an excellent substrate for yeast MAP, an enzyme which requires that the Taa2 residue of its substrates have a radius of gyration of 1.29 A (Moerschell et al., 1990). It appears that the activity of S. cerevisiae MAP substrates depends on the physical chemical properties of their first 2-3 residues. In contrast, the amino-terminal 6-8, and in some cases more COOH-terminal, residues influence the kinetic properties of S. cereuisiue NMT substrates (Towler et al., 1987a, 198713;1988a, 1988bDuronio et al., 1991a). Since MAP appears to recognize a more restricted region of the amino terminus of (nascent) proteins than NMT, it seems likely that all potential NMT substrates will have an exposed Gly'. Current evidence suggests that both MAP and NMT act on nascent proteins that are less than 100 residues long (Deichaite et al., 1988). It is not known whether efficient processing requires a direct interaction between these amino-terminal modifying enzymes and the translational apparatus (e.g. ribosomes). The fact that NMT and MAP substrates are not targeted to the secretory pathway implies that if such an association exists then it would involve free, rather than membrane bound (ER) ribosomes. Our EM immunogold localization and subcellular fractionation studies of yeast suggest that the latter ribosomal population does not harbor a significant fraction of cellular NMT. However, the resolution of the micrographs was not high enough for us to determine whether the cytosolic NMT is affiliated with free ribosomes.
NMT substrates must avoid cotranslational acetylation by S. cereuisiue amino-terminal, a-amino, acetyltransferase, an enzyme (encoded by the A A A l gene) that transfers an acetyl group from acetyl-coA to the a-amino group of yeast proteins and plays an important role in growth and mating (uual null strains have defective sporulation and defective a-type mating and are sensitive to heat shock; Lee et al., 1988Lee et al., ,1989aLee et al., , 1989b: Fifty % of soluble yeast proteins are Ne-acetylated. Approximately 95% of acetylated proteins have a Ser, Ala, Met, Gly, and Thr at Xaa' (cfi Persson et al., 1985;Arfin and Bradshaw, 1988) although its substrate specificity may be determined by residues situated COOH-terminal to this position (Lee et d., 1990b). The development of an in vitro assay for N"-acetyltransferase activity, together with a method for purifying Aaal to apparent homogeneity (Lee e t al., 1988), allows a survey to be conducted of our panel of >lo0 synthetic peptides previously used to assess the substrate specificity of NMT (Towler et al., 1987a, 198%;1988a, 1988bDuronio et al., 1991a). If substrates with comparable activity for both enzymes are identified, then it may be necessary to directly compare and contrast the compartmentalization of the acetyltransferase and the acyltransferase in S. cereuisiue to explain how N-myristoylproteins avoid being acetylated. Alternatively it may be informative to investigate the efficiency of protein N-myristoylation in strains with uual null mutations.
Finally, our observation that S. cereuisiue NMT is not associated with any organelles is compatible with an earlier finding that this eukaryotic protein modification can be recapitulated in E. coli (a bacterium with no endogenous NMT activity) using a dual plasmid expression system (Duronio et al., 1990a, 199Ob;1991a, 1991bLinder et al., 1991). Sequential production of S. cereuisiue NMT and a substrate protein results in efficient acylation in this prokaryote while retaining an absolute requirement for myristate and a Gly' residue. The absence of organelles plus the availability of mutations affecting genes involved in fatty acid transport and/or metabolic processing make E. coli an attractive, "simple," and genetically manipulatable system for identifying and characterizing S. cereuisiue (and E. coli) proteins (other than NMT) that can regulate protein N-myristoylation.