A Role for Saccharomyces cerevisiae Fatty Acid Activation Protein 4 in Regulating ProteinN-Myristoylation during Entry into Stationary Phase*

Saccharomyces cerevisiae contains four known acyl-CoA synthetases (fatty acid activation proteins, Faaps). Faa1p and Faa4p activate exogenously derived fatty acids. Acyl-CoA metabolism plays a critical role in regulating protein N-myristoylation by the essential enzyme, myristoyl-CoA:protein N-myristoyltransferase (Nmt1p). In this report, we have examined whether Faa1p and Faa4p have distinct roles in affecting protein N-myristoylation as cells transition from growth in rich media to a growth-arrested state during nutrient deprivation (stationary phase). The colony-forming potential of 10 isogenic strains was defined as a function of time spent in stationary phase. These strains contained either a wild type or mutant NMT1 allele, and wild type or null alleles of each FAA. Only the combination of the Nmt mutant (nmt451Dp; reduced affinity for myristoyl-CoA) and loss of Faa4p produced a dramatic loss of colony-forming units (CFU). The progressive millionfold reduction in CFU was associated with a deficiency in protein N-myristoylation that first appeared during logarithmic growth, worsened through the post-diauxic phase, and became extreme in stationary phase. Northern and Western blot analyses plusN-myristoyltransferase assays showed that Nmt is normally present only during the log and diauxic/post-diauxic periods, indicating that N-myristoylproteins present in stationary phase are “inherited” from these earlier phases. Moreover,FAA4 is the only FAA induced during the critical diauxic/early post-diauxic transition. Although substitution of nmt1–451D for NMT1 results in deficiencies in protein N-myristoylation, these deficiencies are modest and limited by compensatory responses that include augmented expression of nmt1–451D and precocious induction of FAA4in log phase. Loss of Faa4p from nmt1–451D cells severely compromises their capacity to adequately myristoylate Nmt substrates prior to entry into stationary phase since none of the other Faaps are able to functionally compensate for its absence. To identify Nmt1p substrates that may affect maintenance of proliferative potential during stationary phase, we searched the yeast genome for known and putative N-myristoylproteins. Of the 64 genes found, 48 were successfully deleted in NMT1 cells. Removal of any one of the following nine substrates produced a loss of CFU similar to that observed in nmt1–451Dfaa4Δ cells: Arf1p, Arf2p, Sip2p, Van1p, Ptc2p, YBL049W (homology to Snf7p), YJR114W, YKR007W, and YMR077C. These proteins provide opportunities to further define the molecular mechanisms that regulate survival during stationary phase.

deleted in NMT1 cells. Removal of any one of the following nine substrates produced a loss of CFU similar to that observed in nmt1-451Dfaa4⌬ cells: Arf1p, Arf2p, Sip2p, Van1p, Ptc2p, YBL049W (homology to Snf7p), YJR114W, YKR007W, and YMR077C. These proteins provide opportunities to further define the molecular mechanisms that regulate survival during stationary phase.
Myristoyl-CoA:protein N-myristoyltransferase (Nmt; EC 2.3.1.97) catalyzes the co-translational covalent attachment of myristate (C14:0) to the N-terminal glycine of a subset of important eukaryotic and viral proteins (1)(2)(3). Genetic studies have established that Saccharomyces cerevisiae Nmt1p 1 is essential for survival during vegetative growth (4 -6). At least two metabolic pathways produce myristoyl-CoA that can be used by Nmt1p during vegetative growth: de novo synthesis by the fatty acid synthetase (Fas) complex and activation of myristate by cellular acyl-CoA synthetases (fatty acid activation proteins or Faaps) (7). Four unlinked FAA genes have been identified to date. None are needed to sustain vegetative growth when Fas is active (7). Faa1p and Faa4p are both able to activate imported myristate, have similar acyl chain length specificities in vitro, and together account for 99% of the myristoyl-CoA synthetase activity in log phase cells treated with the Fas inhibitor cerulenin (8 -10). While both Faa1p and Faa4p can activate exogenous fatty acids, including myristate, in log phase cells, studies of isogenic strains containing various combinations of faa null alleles have demonstrated that Faa1p plays the predominant role (7).
The acyl chain specificity of purified Faa2p, a peroxisomeassociated enzyme (11), is similar to that of Faa1p and Faa4p. Faa3p accommodates longer chain fatty acids than the other Faaps but has markedly lower acyl-CoA synthetase activity for C8:0 to C18:0 and prefers monoenoic fatty acids as substrates (palmitoleic acid (C16:1 ⌬9, 10 ) and oleic acid (C18:1 ⌬9, 10 ); Ref. 8). Forced expression of Faa1p or Faa4p, but not Faa2p or Faa3p, can compensate for the combined loss of Faa1p and Faa4p in faa1⌬faa4⌬ cells when Fas is inhibited. This suggests that Faa2p and Faa3p are either unable to gain access to exogenously supplied fatty acids or to deliver them to required intracellular sites of metabolic processing (7).
The apparent functional redundancy of Faa1p and Faa4p in log phase cells suggests that they may have distinct functions under varying physiologic conditions. One such condition is the transition from exponential growth when nutrients are abundant to growth arrest during a nutrient-deprived state (stationary phase). Exhaustion of nutrient supplies, combined with the reported increased rate of degradation of the ␣ and ␤ subunits of Fas during entry into stationary phase (12,13), could require additional contributions from one or more Faaps to maintain cellular pools of acyl-CoAs, including myristoyl-CoA. Acyl-CoA metabolism has an important impact on protein N-myristoylation and is likely to be affected by a cell's nutritional status. The role of protein N-myristoylation in regulating entry into, survival during, or exit from stationary phase has not been reported. However, adaptations required for survival during periods of nutrient deprivation and stress depend upon the ability of cellular signal transduction pathways to monitor and respond to changes in the external environment. N-Myristoylproteins participate in a number of signal transduction pathways; Nmt substrates include serine/threonine and tyrosine kinases, kinase substrates, phosphoprotein phosphatases, and the ␣-subunits of heterotrimeric G proteins (1,3).
To assess the roles of the Faaps and their impact on the regulation of protein N-myristoylation, we generated a panel of isogenic yeast strains containing a wild type or a mutant NMT1 in combination with various faa null alleles. The nmt1 mutant contains a Gly 451 3 Asp substitution that reduces its affinity for myristoyl-CoA (5,6,14,15). Our results reveal that the Faaps have distinct growth phase-dependent patterns of expression, that Faa4p plays a unique role in regulating protein N-myristoylation during the transition between log and stationary phases, and that specific N-myristoylproteins contribute to stationary phase survival.

EXPERIMENTAL PROCEDURES
Strains-Yeast strains used in this study were constructed using standard methods (16).
Assessment of Viability during Stationary Phase-Unless otherwise noted, strains were grown in 5 ml of 1% yeast extract, 2% peptone, 2% dextrose (YPD) at 24°C for 7-10 days. Growth rates were monitored by measuring optical density (OD) at 600 nm. After reaching stationary phase (defined as no further net increase in OD 600 ), cells were pelleted by centrifugation, washed twice in 1 M sorbitol, once in deionized water, and then resuspended in 5 ml of deionized water. Incubation was continued at 24°C with gentle shaking. Aliquots were withdrawn at regular intervals prior to, and 1-45 days after, resuspension in water. These aliquots were plated on YPD plus 2% agar and incubated at 24°C to define the number of colony-forming units (CFU) that remained as a function of time in water. All experiments were repeated on at least three separate occasions, each time in duplicate.
Transformants were selected by direct plating on complete synthetic media minus uracil (Bio101) and incubation at 24°C. Transformed strains were grown in 5 ml of the synthetic media and YPD, and their viability during stationary phase assayed as described above. Duplicate aliquots were plated on selective and YPD media to establish that their episomes had been retained during incubation in water.
Preparation of Total Cellular RNA and Northern Blot Analysis-An overnight culture (in YPD) of each isogenic strain was used to inoculate flasks containing 50 ml of fresh YPD. Growth conditions and subsequent resuspension of stationary phase cultures in water were as described above with the exception that the volume of water used to resuspend stationary phase cells was increased proportionally. At various times during growth in YPD or after resuspension in water (see Fig. 1), 5-ml aliquots were removed and the cells were harvested by centrifugation at 1,600 ϫ g for 5 min at 4°C. Following two washes with phosphate-buffered saline, the cell pellet was resuspended in 0.6 ml of cell wall digestion buffer (1 M sorbitol, 0.1 M EDTA, pH 7.4, and 250 units of Zymolyase 100-T (ICN)). Zymolyase 100-T was then activated by adding 10 l of 2-mercaptoethanol/ml of the cell wall digestion reaction solution, and the mixture was incubated at 24°C with shaking for 15 min. The resulting spheroplasts were pelleted at 1,600 ϫ g for 3 min, and the supernatant was discarded. Lysis of spheroplasts and RNA extraction were performed exactly as described in the Purescript RNA isolation kit protocol (Gentra Systems, Inc.). The integrity of all RNA samples was confirmed by denaturing agarose gel electrophoresis.
Total cellular RNA (10 g from each strain at each time point surveyed) was fractionated by electrophoresis through 1.2% agarose gels containing 1.1% formaldehyde. RNAs were then transferred to GeneScreen Plus membranes (NEN Life Science Products). The RNA blots were probed with 32 P-labeled DNAs of equivalent specific activity. Labeling was performed using the Random Primed DNA labeling kit from Boehringer Mannheim. Hybridization and washing stringencies were identical to those recommended in the protocol supplied by the manufacturer of the GeneScreen Plus membranes. RNA-DNA hybrids were visualized using a storage phosphorimaging system (Molecular Dynamics), and signal intensities were quantitated with software provided with the imaging apparatus.
At least two independent RNA samples were prepared per strain per time point. Each RNA, in turn, was surveyed in duplicate or triplicate blots. Each blot contained an internal control RNA prepared from wild type cells that had been harvested at mid-log phase. Blots containing all RNAs from all strains from all time points were simultaneously probed with a given 32 P-labeled DNA. These blots were stripped and reprobed with up to three different DNAs. To remove bound probe after each cycle of hybridization, blots were incubated for 30 min at 90°C in 0.1ϫ SSC, 0.5% SDS. After the final probing, blots were stripped and reprobed with rRNA DNA (kindly supplied by Susan Wente, Washington University). The signal obtained from each RNA sample after probing with each DNA was then divided by the 18 S rRNA plus 25 S rRNA signals obtained from that RNA sample after hybridization with rRNA DNA. This normalized value for the steady state concentration of a given mRNA was then expressed as a percentage of the value obtained from the internal reference RNA control. Values obtained from the reference control for a given mRNA species varied by Ͻ5% between blots.
Assessment of Cellular Morphology and the Ability to Exclude Methylene Blue-Aliquots of cultures were removed at time points 1-7, shown in Fig. 1. Cells were viewed under a 100ϫ oil immersion lens using a Nikon Optiphot-2 microscope equipped with Nomarski optics, both before and after incubation in 0.1% methylene blue (Sigma) for 1-90 min at 24°C.
Measurement of Steady State Levels of Nmt1p and nmt451Dp during Vegetative and Stationary Phases-Overnight cultures of the various isogenic strains were diluted to an OD 600 of 0.005 in duplicate flasks that each contained 75 ml of YPD. At time points 1-6 in Fig. 1, 20-ml aliquots were harvested by centrifugation at 1600 ϫ g. Spheroplasts were prepared as described above and were disrupted by resuspension in 0.5 ml of lysis buffer (50 mM Tris, pH 6.8, 10 mM EDTA, 2% SDS, 2 M leupeptin (Sigma), 2 M pepstatin (Sigma), and 50 M 4-(2-aminoethyl)-benzolsulfonylfluoride (Boehringer Mannheim)). Cellular debris was removed by centrifugation at 10,000 ϫ g for 5 min. Protein concentrations in the resulting cleared lysates were determined according to Peterson (24). An equal mass of protein from each sample was reduced, denatured, and fractionated by electrophoresis through 10% polyacrylamide gels containing 0.1% SDS (25). Proteins were then transferred to polyvinylidene difluoride membranes (Amersham Pharmacia Biotech), and the protein blots were probed with affinity-purified rabbit antibodies to Candida albicans Nmt that cross-react with S. cerevisiae Nmt1p (26). Antigen-antibody complexes were visualized using the ECL Western blotting kit (Amersham Pharmacia Biotech).
Assessment of the in Vivo State of Arf1p N-Myristoylation Using a Gel Mobility Shift Assay-Details of this assay have been described by Lodge et al. (15). Strains were grown, cells were harvested, total protein was extracted, and Western blots were prepared exactly as described above. The protein blots were subsequently probed with a rabbit anti-S. cerevisiae Arf1p sera (R40, a generous gift of R. Kahn, Emory University; diluted 1:50,000 in phosphate-buffered saline containing 1% gelatin, 0.2% Tween 20, and 0.1% sodium azide; see Ref. 15). Antigen-antibody complexes were detected using the Western-Light kit from Tropix.
Assays of Nmt Activity in Cellular Lysates-Nmt activity was assayed in cell lysates prepared from isogenic strains harvested during log and stationary phases (time points 1 and 6). Cells were collected from 50-ml cultures by centrifugation, washed once in phosphate-buffered saline, and resuspended in 0.5 ml of lysis buffer (10 mM Tris, pH 7.4, 1 mM dithiothreitol, 0.1 mM EGTA, 10 g/ml aprotinin, 100 M 4-(2-aminoethyl)-benzolsulfonyl fluoride, 4 M pepstatin A, 8 M leupeptin). An equal volume of glass beads (425-600 m diameter, Sigma) was added, and the suspension vortexed for 1 min at room temperature alternating with incubation on ice for 1 min for a total of three cycles. Cellular debris was removed by centrifugation at 10,000 ϫ g for 10 min, and the supernatants were assayed for Nmt activity using 0.23 M Disruption of Genes Encoding Known or Putative N-Myristoylproteins-The translation products of all standard ORFs that were listed in the S. cerevisiae Genome Database (SGD) 2 in August, 1997, were scanned for the presence of M 1 G 2 X 3 X 4 X 5 X 6 X 7 at their N terminus, using the PatMatch sequence analysis tool contained in the SGD website. Based on previously published studies of the peptide substrate specificity of Nmt1p (28 -30), E, D, R, K, H, P, F, Y, and W were not allowed at position 3 (X 3 ); all possible amino acids were allowed at X 4 and X 5 ; only S, T, A, G, C or N were permitted at X 6 ; and all residues except P were allowed at X 7 . A total of 64 known and putative N-myristoylpro-  (Table I) and replaced with HIS3 using a previously described protocol (31). PCR studies confirmed that 48 ORFs had been deleted successfully. Of the remaining 16 ORFs (indicated above by italics), two are known to result in lethality when removed: YGL022W (STT3, Ref. 32) and YHR005C (GPA1, Ref. 33).
The colony-forming potential of each of the resulting 48 isogenic strains and their wild type parent was assessed during stationary phase using the protocols described above (n ϭ 2 independent experiments/strain, each done in duplicate; time points surveyed in stationary phase ϭ 10 and 25 days after transfer to deionized water).
In Vitro Assays of Potential Nmt1p Substrates-Octapeptides representing residues 2-9 of the protein products encoded by YBL049W, YDL192W, YDL137W, YDR079W, YDR364C, YER089C, YGL208W, YJR114W, YKR007W, YML115C, YMR077C, YMR239C, and YOR181W were generated using an Applied Biosystems Model 431A synthesizer. Following removal from the resin and deprotection, each peptide was purified by high performance liquid chromatography, characterized by mass spectrometry, and tested as a substrate for Nmt1p. Details of the in vitro Nmt1p assay have been presented in an earlier publication (34). The 110-l reaction mixture contained 1-10 ng of purified E. coli-derived S. cerevisiae Nmt1p (35), 0.23 M [ 3 H]myristoyl-CoA, and 100 M peptide. Following a 10-min incubation at 30°C, [ 3 H]myristoylpeptide was purified from the reaction mixture by reverse phase high performance liquid chromatography using a C18 10-m Bondapak column (dimensions ϭ 3.9 ϫ 300 mm; Waters Corp.) and a linear gradient from H 2 O, 0.1% trifluoroacetic acid, 0.05% triethylamine to 100% acetonitrile, 0.1% trifluoroacetic acid. The amount of labeled myristoylpeptide recovered was quantitated with an in-line scintillation counter (36). All assays were performed in triplicate.

A Comparison of NMT1 and nmt1-451D Cells with a faa4 Null Allele Suggests that Protein N-Myristoylation Affects Colony Forming Potential during Stationary Phase-S. cerevisiae
undergoes logarithmic growth in nutrient-rich media. As nutrients are depleted, cells undergo a diauxic transition and then begin a period of slow growth known as the post-diauxic phase. When nutrients are exhausted, cells enter a stationary phase where proliferation ceases (Fig. 1). Stationary phase is thought to represent a differentiated state in which yeast cells display an enhanced capacity to withstand various forms of induced stress (37).
Ten isogenic strains containing either NMT1 or nmt1-451D, with or without one faa null allele (Table I) were grown in rich medium (YPD) at 24°C, which is a permissive temperature for nmt1-451D (6). Once cells had reached stationary phase, they were washed and resuspended in deionized water. Cells were subsequently maintained at 24°C in deionized water for up to 45 days. Aliquots were withdrawn at regular intervals and plated on YPD/agar to determine the number of cells that could Representative growth curve of a wild type strain at 24°C. Growth phases are indicated as are the time points that cells were recovered for various experimental studies. Time points: 1, mid-log phase; 2, diauxic shift/early post-diauxic phase; 3, mid post-diauxic phase; 4, late post-diauxic phase; 5, stationary phase (2 days after resuspension in deionized water); 6, 15 days after resuspension in water; 7, 45 days after resuspension in water.
return to the cell cycle and form colonies.
NMT1 strains with or without faa1⌬, faa2⌬, faa3⌬, or faa4⌬ null alleles all reached a similar maximum density: 10 7 CFU/ml of YPD. In each case, colony-forming potential decreased only 10-fold during their 45-day incubation in water (e.g. Fig. 2A). Deletion of any one of the FAAs from nmt1-451D cells did not produce an appreciable (i.e. Ͼ5-fold) effect on final maximal density when compared with their isogenic NMT1 counterparts. For four of the five nmt1-451D strains (nmt1-451D, nmt1-451Dfaa1⌬, nmt1-451Dfaa2⌬, and nmt1-451Dfaa3⌬), the time course and magnitude of decline in CFU during the 45-day incubation in water duplicated what was observed with the corresponding NMT1 strains. However, the nmt1-451Dfaa4⌬ strain underwent a progressive millionfold loss of colony-forming units (Fig. 2B).
There was no evidence that nmt1-451Dfaa4⌬ cells underwent lysis during their 45-day incubation in water. The OD 600 of the culture remained constant (four experiments, each in duplicate; data not shown). Direct microscopic examination and counting with a hemocytometer confirmed that the number of intact cells remained unchanged. After 45 days, only ϳ60% of nmt1-451Dfaa4⌬ cells displayed morphologic abnormalities, a remarkable figure considering that only 0.0001% of the cells formed colonies when transferred to a nutrient-rich environment. These morphologic abnormalities included increased number of vacuoles, a more granular cytoplasm, plus an elongated shape with multiple buds, and were not apparent in mid-log phase nmt1-451Dfaa4⌬ cells or in any other strain at any of the time points surveyed (Fig. 3, A-F; data not shown).
Although nmt1-451Dfaa4⌬ cells lose their colony-forming potential during stationary phase, two assays indicated that they do not suffer a metabolic death. Viable yeast cells are able to exclude methylene blue from their cytoplasm and appear colorless after incubation with the dye while non-viable cells appear blue (e.g. Ref. 38). There were no detectable differences in the ability of isogenic NMT1, NMT1faa4⌬, nmt1-451D, and nmt1-451Dfaa4⌬ cells to exclude methylene blue after a 10-or 45-day incubation in water (Ն90% of cells appeared colorless in all cases; Fig. 3, G-I). The second assay involved metabolic labeling with [ 3 H]myristate. The results indicated that deleting Faa4p from NMT1 and nmt1-451D cells had no appreciable effects on the incorporation of label into total cellular lipids, on the pattern of tritiated phospholipid species produced during log or stationary phases, or on the accumulation of radiolabeled triacylglycerols (Fig. 4).
Thus, the marked difference in CFU between stationary phase nmt1-451DFAA4 and nmt1-451Dfaa4⌬ cells discloses an essential role for FAA4 that is not apparent during growth in a nutrient-rich environment (7). Moreover, the millionfold reduction in colony-forming potential cannot be ascribed to defects in the ability of nmt1-451Dfaa4⌬ cells to direct exogenous fatty acids to phospholipid or neutral lipid biosynthetic pathways.
Evidence That the Stationary Phase Phenotype of nmt1-451Dfaa4⌬ Cells Is Associated with Reduced Protein N-Myris-toylation-Based on the findings described above, we hypothesized that the loss of colony-forming potential in stationary phase nmt1-451Dfaa4⌬ cells reflected loss of Nmt-dependent functions (N-myristoylproteins) from a combination of factors: a mutant Nmt with reduced affinity for myristoyl-CoA and a deficiency of Faa4p that leads to reduced myristoyl-CoA availability.
This hypothesis was tested in several ways. First, the millionfold reduction in CFU could be prevented entirely by addition of centromeric plasmids containing NMT1 under the control of its own promoter, NMT1 under the control of the glyceraldehyde-3-phosphate dehydrogenase (GPD) promoter (GPD-NMT1), or GPD-FAA4 (Fig. 2C). Loss of colony-forming potential was not rescued with GPD-FAA1, GPD-FAA2, or GPD-FAA3 episomes (data not shown). NMT1 and GPD-NMT1 completely reversed the morphologic abnormalities noted in nmt1-451Dfaa4⌬ cells while nmt1-451Dfaa4⌬ cells containing GPD-FAA4 resembled nmt1-451D cells. Second, when nmt1-451Dfaa4⌬ cells were grown in YPD media supplemented with 500 M myristate and then resuspended in water, there was complete rescue of the loss of CFU (Fig. 2D). Third, GPDdirected expression of Plb1p, a lysophospholipase/phospholipase B (39) and Tgl1p, a putative triglyceride lipase (40), resulted in full rescue of colony-forming potential (Fig. 2D). This rescue is consistent with Plb1p-and Tgl1p-mediated liberation of acyl chains from more complex cellular lipids and their subsequent metabolic processing to myristoyl-CoA.
A final assay established that the combination of nmt1-451D and a faa4 null allele was uniquely associated with deficits in cellular protein N-myristoylation during stationary phase. ADP-ribosylation factor 1 (Arf1p) is a known yeast N-myristoylprotein (41,42). We showed previously that non-myristoylated Arf1p has a slower mobility than myristoyl-Arf1p during SDS-polyacrylamide gel electrophoresis (26). Decreases in the efficiency of protein N-myristoylation in vivo can be followed by noting increases in the fractional representation of non-myristoylated (slowly migrating) Arf1p in Western blots containing total cellular proteins (15). Since there is no evidence of a de-myristoylating activity in S. cerevisiae (43), an increase in the ratio of non-myristoylated to N-myristoylated Arf1p can be ascribed to reduced acylation by a wild type or mutant Nmt (14,15).
Total cellular proteins were isolated from each member of the panel of 10 isogenic strains at various points during their growth and stationary phases. When Western blots containing these protein preparations were probed with antibodies to Arf1p, only the N-myristoylated form was detected in log phase NMT1 and nmt1-451D cells with four wild type FAA alleles. None of the faa⌬ null alleles, including faa4⌬, affected the efficiency of Arf1p N-myristoylation in log phase cells with wild type NMT1 (Fig. 5A). Although deletion of Faa1p and Faa2p from nmt1-451D cells did not perturb Arf1p N-myristoylation, ϳ25% of Arf1p was not N-myristoylated in log phase nmt1-451Dfaa3⌬ and nmt1-451Dfaa4⌬ cells (Fig. 5A). During the diauxic/post-diauxic phase, the nonmyristoylated Arf1p species disappeared in nmt1-451Dfaa3⌬ cells but persisted in nmt1-451Dfaa4⌬ cells (Fig. 5B). All other strains contained only N-myristoylated Arf1p. During stationary phase, nmt1-451Dfaa4⌬ cells only contained the nonmyristoylated isoform of Arf1p, and at concentrations that were substantially lower than the myristoyl-Arf1p present in all other strains (Fig. 5C). Episomal expression of NMT1 in nmt1-451Dfaa4⌬ cells resulted in complete disappearance of nonmyristoylated Arf1p during log, diauxic/post-diauxic and stationary phases (Fig. 5,

A-C).
These findings established that the combination of the nmt1-451D allele and the absence of Faa4p results in a deficiency in protein N-myristoylation that first becomes evident during logarithmic growth, worsens through the post-diauxic phase, and becomes extreme in stationary phase. The results raised two questions. At what phase is Nmt activity required to ensure retention of colony-forming potential? What is the relationship, if any, between the presence or absence of Faa4p (or other Faaps) and cellular Nmt expression?
NMT Expression Is Restricted to Log and Diauxic/Early Post-diauxic Phases-We compared levels of the mRNA and protein products of NMT1 and nmt1-451D in each member of the panel of 10 isogenic strains, at time points 1-6 denoted in Fig. 1. In all cases, NMT1 or nmt1-451D mRNA could only be detected by Northern blot analysis during logarithmic growth and during the diauxic/early post-diauxic phase (time points 1 and 2 in Fig. 6A). Western blot analysis revealed an identical pattern of accumulation of the N-myristoyltransferase with one exception: in nmt1-451Dfaa4⌬ cells, nmt451Dp persisted at barely detectable levels to the middle of the post-diauxic phase (time point 3 in Fig. 6B). In vitro assays of cell lysates prepared from the various isogenic strains harvested at time points 1 and 6 also revealed that there was no detectable Nmt activity in stationary phase (data not shown).
This restriction of Nmt mRNA to log and diauxic/early postdiauxic phases was not altered when centromeric plasmids containing NMT1 under the control of its own promoter or the GPD promoter were introduced into NMT1, nmt1-451D, NMT1faa4⌬, and nmt1-451Dfaa4⌬ cells. When NMT1 was expressed under the control of its own promoter, cellular Nmt1p mRNA levels increased two-fold in log and diauxic/early post-diauxic phases but the mRNA remained undetectable in middle or late post-diauxic or stationary phases. When NMT1 expression was controlled by the GPD promoter, cellular Nmt mRNA levels increased by an average of 4.5-fold compared with cells with no vector or the empty GPD vector, and then fell to below the limits of detection by mid post-diauxic phase (data not shown).
In contrast to the results obtained with GPD-NMT1 episomes, Northern analyses of NMT1, nmt1-451D, and nmt1-451Dfaa4⌬ strains containing GPD-PLB1, the empty GPD vector, or no vector at all indicated that GPD-PLB1 produced increases in Plb1p mRNA levels from log through stationary phases (15-fold in mid-log phase, 20 -25-fold in diauxic/early post-diauxic phase, and 20-fold in stationary phase; data not shown).
Together, these results suggest that Nmt mRNA is unstable during the post-diauxic and stationary periods. The data also suggest that rescue of the colony-forming potential of nmt1-451faa4⌬ cells in stationary phase by NMT1 and GPD-NMT1 episomes is due to increases in cellular protein N-myristoylation during log and/or diauxic/early post-diauxic phases, and to the persistence of some or all these acylated proteins during stationary phase (see Fig. 5).
Loss of Faa4p Affects NMT Expression during the Diauxic/ Early Post-diauxic Phase-Northern blot analysis revealed that in cells containing the wild type NMT1 allele, Nmt1p mRNA levels increase during log phase when any one of the Faaps is removed: loss of Faa1p, Faa2p, or Faa3p produces a significant (p Ͻ 0.05) 2-fold increase, while loss of Faa4p produces a 6-fold increase (time point 1 in Fig. 6, A and C). In contrast, only removal of Faa4p results in a increase in Nmt1p mRNA levels during the diauxic/early post-diauxic phase (time point 2 in Fig. 6C).
During log and diauxic/early post-diauxic phases, cells containing nmt1-451D and wild type FAAs have 4-fold higher FIG. 2. Deletion of FAA4 but not other FAA genes from nmt1-451D cells has a pronounced effect on their colony-forming potential during stationary phase. A, NMT1 strains were grown in YPD to stationary phase and transferred to deionized water where they were maintained at 24°C. Duplicate aliquots were withdrawn at various times from duplicate cultures of each strain, and the number of CFU was defined. In this and subsequent panels, CFU values were averaged for the four aliquots at each time point sampled. These values varied by Ͻ20% per time point per strain. The results of a representative experiment are shown (n ϭ four independent experiments). B, analysis of isogenic nmt1-451D strains. C, centromeric plasmids containing GPD-NMT1 and GPD-FAA4 rescue the loss of colony-forming potential of nmt1-451Dfaa4⌬ cells (centromeric plasmids are retained in this strain even after a 60-day incubation in water). D, rescue of the proliferative potential of stationary phase nmt1-451Dfaa4⌬ cells by centromeric plasmids encoding a phospholipase (PLB1) and a triglyceride lipase (TGL1). ϩ myristate refers to the rescue obtained when cells were grown in YPD supplemented with 500 M myristate prior to reaching stationary phase. levels of Nmt mRNA compared with isogenic NMT1 cells (Fig.  6, A and C). This increase is augmented further by the faa3⌬ null allele during logarithmic growth and by faa4⌬ during the diauxic/early post-diauxic phase (Fig. 6C), i.e. when the reduction in active cellular protein N-myristoylation is greatest (as defined by the Arf1p reporter; see Fig. 5).
These observations indicate that Nmt mRNA levels change in response to growth phase, loss of Faaps, and the presence of nmt1-451D. The fact that, among all of the Faaps, loss of Faa4p produces the greatest rise in mRNA encoding the wild type myristoyltransferase emphasizes the importance of Faa4p in ensuring that there is sufficient N-myristoylation of cellular proteins.

In Wild Type Cells, FAA4 Expression Is Induced during the Diauxic/Early Post-diauxic Shift and Sustained during Postdiauxic and Stationary Phases-
The fact that loss of Faa4p had such a pronounced effect on Nmt expression in nmt1-451D cells during the critical diauxic/post-diauxic transition prompted us to determine when the FAAs, including FAA4, are expressed. Faap mRNA levels were first defined in wild type cells at each of the six time points shown in Fig. 1. The results are presented in Table II. Faa1p mRNA is present during all phases and is the most abundant Faap mRNA. Levels are similar during log phase, the diauxic transition, and the postdiauxic phase but increase ϳ4-fold in stationary phase. Faa2p mRNA is present at low levels during log and diauxic/early post-diauxic phases (0.1% of Faa1p mRNA), rise 40-fold during the post-diauxic phase, then promptly return to their previous values in late post-diauxic and stationary phases. FAA3 expression is confined to log and diauxic/early post-diauxic phases. Faa4p mRNA is barely detectable in log phase (0.02% of Faa1p mRNA). There is a dramatic 100-fold increase in concentration as cells enter diauxic/early post-diauxic phase. Expression is sustained throughout all subsequent phases, albeit at slightly reduced levels. Together, these results reveal distinct, growth phase-dependent expression patterns for each FAA.
FAA4 Expression Is Increased Markedly in nmt1-451D Cells during Log Phase-Faap mRNA levels were subsequently defined in each of the other nine isogenic strains at each of the six time points. Deletion of other FAAs or introduction of nmt1-451D had no effect on the time course of Faa1p, Faa3p, or Nmtp mRNA accumulation. Moreover, the concentration of Faa1p and Faa3p mRNAs varied less than 4-fold among the various strains at a given phase (data not shown).
In contrast, the steady state level of Faa4p mRNA during log phase is affected by loss of other Faaps and by introduction of nmt1-451D (Fig. 7A). (i) Introducing nmt1-451D produces a 55-fold increase in Faa4p mRNA levels in cells with four wild type FAA alleles, when compared with the wild type strain. (ii) Loss of Faa1p, Faa2p or Faa3p results in a precocious induction of FAA4 expression during log phase in NMT1 cells (5-10-fold increase). The effect is much more pronounced in the corresponding nmt1-451D strains (70 -210-fold; Fig. 7A). Loss of the other Faaps or substitution of nmt451Dp for Nmt1p had minimal (Յ3-fold) effects on Faa4p mRNA levels as cells passed through the diauxic shift and entered stationary phase.
Loss of Faa4p, Faa1p, or Faa3p in NMT1 or nmt1-451D cells results in elevations in Faa2p mRNA during log (Fig. 7B) and   FIG. 3. Morphologic analysis of isogenic NMT1, nmt1-451D, and nmt1-451Dfaa4⌬ strains during log and stationary phases. Cells from the indicated strains were recovered during the mid-log phase of growth in YPD and 45 days after suspension in deionized water. All photographs were taken under Nomarski optics and are shown at the same magnification. A and B, nmt1-451D produces a slight increase in cell size during logarithmic growth. C, deleting FAA4 does not produce appreciable changes in nmt1-451D cell morphology during log phase. D and E, in contrast to wild type cells, a fraction of stationary phase nmt1-451D cells with wild type FAA alleles contain buds (e.g. arrow). V, vacuole. F, when FAA4 is deleted from nmt1-451D cells, they become elongated, acquire numerous granules and vacuoles, and appear aggregated (closed arrows). The open arrow points to a stationary phase cell with "normal" morphology. G-I, exclusion of methylene blue by stationary phase NMT1, nmt1-451D, and nmt1-451Dfaa4⌬ strains. Most cells are able to exclude the dye. Those that cannot appear purple.
all subsequent phases (data not shown). Post-diauxic nmt1-451Dfaa4⌬ cells have the highest levels of this mRNA (8-fold greater than post-diauxic wild type cells).
In summary, among Faap mRNAs, Faa4p mRNA undergoes the greatest change in response to nmt1-451D and/or to loss of other Faaps. These changes are most noticeable during log phase, presumably because FAA4 expression is already maximally up-regulated during subsequent phases as nutrients are depleted.

FAS1 Expression Is Restricted to Log and Diauxic/Early
Post-diauxic Phase and Is Not Induced in nmt1-451Dfaa4⌬ Cells-Fatty acid synthetase (Fas) is a predominant contributor to acyl-CoA pools during logarithmic growth. Myristoyl-CoA represents ϳ5% of its acyl-CoA products (44,45). Fas1p mRNA encodes the ␤-subunit of the ␣ 6 ␤ 6 Fas complex. In wild type NMT1 cells, Fas1p mRNA undergoes a 10-fold reduction in its steady state level during the diauxic/early post-diauxic phase and is no longer detectable after this period (data not shown). Removal of Faa1p, Faa2p, Faa3p, or Faa4p from NMT1 strains has a less than 2-fold effect on Fas1p mRNA levels during log Total cellular proteins were isolated from isogenic strains after they had been harvested during mid-log, post-diauxic, and stationary phases (time points 1, 4, and 6 in Fig.  1). Proteins (200 g/sample) were fractionated by SDS-polyacrylamide gel electrophoresis, Western blots were prepared, and then probed with a previously characterized rabbit anti-Arf1p sera. Arrows point to the N-myristoylated and nonmyristoylated isoforms of Arf1p. phase and the diauxic transition.
Substitution of nmt1-451D for NMT1 increases Fas1p mRNA concentrations two-fold in log phase, but does not alter its subsequent disappearance at later stages. This two-fold increase is observed in all nmt1-451D strains except nmt1-451Dfaa4⌬ cells where levels remain unchanged (data not shown). The failure of nmt1-451Dfaa4⌬ cells to augment FAS1 expression during log phase may contribute to their observed deficits in protein N-myristoylation.
Identification of N-Myristoylproteins That Affect Colony Forming Potential of NMT1 Cells during Stationary Phase-We next sought to identify cellular N-myristoylproteins that might contribute to the colony-forming phenotype of nmt1-451Dfaa4⌬ cells. Since most N-myristoylproteins depend on their myristoyl moiety for biological activity (46), we reasoned that gene disruption would, in general, mimic a nonor hypo-myristoylated state. The requirements of Nmt1p for peptide substrate recognition in vitro have been studied extensively (28 -30). Using these requirements as a guide, we identified 64 known and candidate N-myristoylproteins in the S. cerevisiae genome (see "Experimental Procedures") and attempted to delete each (individually) in the wild type strain.
Forty-eight genes were successfully deleted and are listed under "Experimental Procedures." Each of the resulting 48 isogenic strains and the wild type parent were assayed for their ability to maintain colony-forming potential during stationary phase. All strains achieved a similar maximum density in YPD (10 7 CFU/ml; variation Յ 3-fold). Thirty-four strains maintained colony-forming potential at levels equivalent to that of the wild type parent: i.e. Ͻ10-fold reduction in CFU after a 10or 25-day incubation in water at 24°C. One NMT1 strain containing a ARF1 deletion had a modest 100-fold reduction in CFU over this time period. Twelve other NMT1 strains exhibited a loss of proliferative potential similar to that observed with nmt1-451Dfaa4⌬ cells: i.e. 1000 -10,000-fold reductions after 10 -25 days in water (Table III). The disrupted genes in these 12 strains encode (i) Arf2p, a known Nmt1p substrate (e.g. Refs. 15 and 47); (ii) Cdc40p/Prp17, a protein involved in pre-mRNA splicing (48,49); (iii) Las17p/Bee1p, a protein reported to be critical for assembly of the actin cytoskeleton in yeast that also has homology to the human Wiscott-Aldrich syndrome protein (50,51); (iv) Sip2p, a protein that interacts with the Snf1p kinase required for the regulatory response of yeast to glucose starvation (52,53); (v) Van1p, a protein required for proper Golgi-mediated protein glycosylation, that has been implicated in producing cell cycle arrest in stationary phase, and that when removed is associated with altered protein phosphorylation (54,55); (vi) Ptc2p, a protein with homology to a Schizosaccharomyces pombe serine/threonine phosphatase 2c (56,57); (vii) Pet100p, a protein involved in cytochromec-oxidase assembly (58); (viii) Rnt1p, a double-strand specific endoribonuclease required for ribosome synthesis and with homology to bacterial RNase III (59); (ix) YBL049W, a protein with homology to Snf7, the serine/threonine kinase involved in the derepression of SUC2 (invertase) during periods of glucose deprivation (60,61) plus several proteins with no obvious homologies.
As with nmt1-451Dfaa4⌬ cells, NMT1 strains lacking any of these genes suffer a loss of colony-forming potential without losing their viability, as judged by methylene blue exclusion (data not shown).   Fig. 1. b Northern blots were probed with 32 P-labeled DNAs having identical specific activities. mRNA levels are expressed as a fraction of the sum of the signals produced by 18 S rRNA and 25 S rRNA in cells at each of the time points surveyed (see "Experimental Procedures"). Mean values from two independent experiments, each done in duplicate, are shown. Values for a given mRNA varied Ͻ20% between experiments. c ND, not detectable.
Octapeptides derived from the N-terminal sequences of these genes were synthesized and tested as substrates for purified Nmt1p. The results indicate that nine are substrates: YBLO49W, Arf1p, Arf2p, Ptc2p, Sip2p, YJR114W, YKR007W, Van1p, and YMR077C (Table III). No in vitro activity could be detected with the other four octapeptides (Pet100p, Cdc40p, Rnt1p, Las17p). This does not prove that these latter proteins are not N-myristoylated in vivo: in some cases, elements required for recognition by the N-myristoyltransferase are located beyond the N-terminal 8 -9 residues (e.g. Ref. 30). DISCUSSION Our studies have yielded the following results. First, Faaps have distinct growth phase-dependent patterns of expression. FAA4 is the only FAA that undergoes an induction of expression during the diauxic/early post-diauxic phase. Second, Nmt mRNA and its protein product are only detectable in log phase and during the diauxic/early-post-diauxic period. Third, only nmt1-451faa4⌬ cells show defects in protein N-myristoylation during both log phase and the diauxic/early-post-diauxic transition. The combination of nmt1-451D with this faa null allele is also the only one that results in a marked reduction in colony-forming potential during stationary phase.
These results are compatible with the following model. Induction of Faa4p during the diauxic/early post-diauxic phase plays a role in assuring sufficient N-myristoylation of cellular proteins for subsequent "survival" during the stress of stationary phase. Such a role is revealed when cellular Nmt activity is compromised. Cells with nmt451Dp and wild type FAA alleles have a reduced capacity for protein N-myristoylation but compensate for the presence of the mutant N-myristoyltransferase by augmenting expression of FAA4 (and nmt1-451D itself). Loss of Faa4p from nmt1-451D cells is devastating; they do not support a further increase in nmt1-451D expression, and other Faaps do not compensate for loss of Faa4p.
Faa2p mRNA levels are augmented during the post-diaxic phase in nmt1-451Dfaa4⌬ cells. However, the augmentation is not sufficient to overcome the observed defects in protein Nmyristoylation, even though the myristoyl-CoA synthetase activities of purified Faa2p and Faa4p are similar (8,10). This is consistent with the notion that Faa2p and Faa4p supply acyl-CoAs to distinct metabolic pathways and/or occupy distinct subcellular locations. Although Faa2p is known to be associated with peroxisomes (11), the subcellular distribution of Faa4p remains to be determined.
Removal of Faa3p from nmt1-451D cells produces a transient defect in cellular N-myristoylation that is confined to log phase and does not affect maintenance of proliferative potential during stationary phase. This phenotype likely reflects two features of FAA3. Unlike FAA4, FAA3 is expressed in log phase but not in post-diauxic or stationary phase cells. In addition, loss of Faa3p is associated with a pronounced upregulation of nmt1-451D expression during log phase. This augmentation of expression presumably helps to restrict the extent of the cellular N-myristoylation defect during log phase so that nmt1-451Dfaa3⌬ cells are able to properly prepare themselves for entry into stationary phase. The enhancement of nmt1-451D expression associated with loss of Faa3p or Faa4p represents the first reported example of gene products that influence expression of this N-myristoyltransferase. The underlying molecular mechanism needs to be defined. For example, is the change in nmt1-451D expression mediated by cellular lipids and/or one or more cellular Nmyristoylproteins whose state of acylation is altered?
Several of our findings indicate that retention of N-myristoylated proteins, acylated prior to entry into stationary phase, affects subsequent survival during this period. As noted above, Nmt1p mRNA and its protein are limited to log and diauxic/ early post-diauxic phases. Moreover, Nmt activity is not detect-  able in cellular lysates prepared from wild type and mutant strains during stationary phase. The deficit in protein N-myristoylation and the loss of colony-forming potential in nmt1-451Dfaa4⌬ cells can be rescued by forced expression of NMT1 even though its mRNA and protein products are only detectable in log and diauxic/early post-diauxic phases.
Many studies have shown that the myristoyl moiety has different functional significance for different N-myristoylproteins (3,46,62,63). In some cases, reduced N-myristoylation of a protein may result in constitutive activation of a cellular signal transduction pathway, as is the case with Gpa1p, the ␣-subunit of the heterotrimeric G protein involved in the mating response (e.g. Ref. 43). In other cases, undermyristoylation of a protein does not produce a detectable change in function (Cnb1p; Refs. 64 and 65). Nonetheless, our genetic and biochemical analysis of nmt1-451Dfaa4⌬ cells indicates that undermyristoylation of a subset of yeast proteins abrogates the ability of cells to "survive" for prolonged periods during stationary phase. Of the 64 known and candidate N-myristoylproteins encoded by the ϳ6220 standard ORFs in the yeast genome, we were able to identify 13 that, when deleted from NMT1 cells, markedly reduce colony-forming potential during stationary phase. We do not know what role myristate plays in the specific function(s) of these identified proteins during stationary phase. However, given the observations that protein N-myristoylation is an irreversible co-translational modification, and that Nmt activity is not present in stationary phase cells, we can conclude that the function of these proteins is not actively regulated during stationary phase by attachment of C14:0. One role of the myristoyl moiety during this period may be to influence protein turnover. For example, reduced N-myristoylation of Arf1p in nmt1-451Dfaa4⌬ cells is associated with a progressive loss of the protein during stationary phase.
Understanding stationary phase is important. The natural environment of most microorganisms only allows for brief periods of rapid growth. Stasis induced by nutrient deprivation is the more common natural situation, and thus the ability to remain viable during times of starvation is crucial for survival (66). The switch from anaerobic growth to aerobic respiration as glucose is depleted from the environment (i.e. the diauxic shift) is accompanied by pervasive changes in the pattern of yeast gene expression (37,67). It is interesting that one of the N-myristoylproteins we identified, Sip2p, is involved in regulating the global cellular response of yeast to glucose starvation (61). Another N-myristoylprotein, Van1p, has been shown to affect proper cell cycle arrest in stationary phase (54). Hopefully, these and some of the other proteins identified during the course of our studies will provide new opportunities for defining the molecular mechanisms that allow cells to properly prepare for entry into stationary phase, to survive the stress of nutrient deprivation, and to respond appropriately to the reintroduction of nutrients into their environment.