Nat3p and Mdm20p are required for function of yeast NatB Nalpha-terminal acetyltransferase and of actin and tropomyosin.

NatB Nalpha-terminal acetyltransferase of Saccharomyces cerevisiae acts cotranslationally on proteins with Met-Glu- or Met-Asp- termini and subclasses of proteins with Met-Asn- and Met-Met- termini. NatB is composed of the interacting Nat3p and Mdm20p subunits, both of which are required for acetyltransferase activity. The phenotypes of nat3-Delta and mdm20-Delta mutants are identical or nearly the same and include the following: diminished growth at elevated temperatures and on hyperosmotic and nonfermentable media; diminished mating; defective actin cables formation; abnormal mitochondrial and vacuolar inheritance; inhibition of growth by DNA-damaging agents such as methyl methanesulfonate, bleomycin, camptothecin, and hydroxyurea; and inhibition of growth by the antimitotic drugs benomyl and thiabendazole. The similarity of these phenotypes to the phenotypes of certain act1 and tpm1 mutants suggests that such multiple defects are caused by the lack of acetylation of actin and tropomyosins. However, the lack of acetylation of other unidentified proteins conceivably could cause the same phenotypes. Furthermore, unacetylated actin and certain N-terminally altered actins have comparable defective properties in vitro, particularly actin-activated ATPase activity and sliding velocity.

as bait in a genome-wide screen by tandem affinity purification (TAP) protocol (12) (yeast.cellzone.com); however the requirement of Asc1p for NatA activity has not been tested. Furthermore, Asc1p, together with Eno1p, Mis1p, Myo1p, and YGR090w, was detected in the complex when Ard1p was tagged (yeast.cellzone.com).
Tercero and Wickner (13) and Tercero et al. (14) described the MAK3 gene that encodes a catalytic subunit of NatC, which is required for the N-terminal acetylation of the viral major coat protein, gag, with an Ac-Met-Leu-Arg-Phe-terminus. The copurification of Mak3p, Mak10p, and Mak31p suggests that these three subunits form a complex (15). Moreover, proteinprotein interactions between Mak3p and Mak10p as well as between Mak31p and Mak10 were detected in a two-hybrid screen (16). We demonstrated that each of the Mak3p, Mak10p, and Mak31p subunits are required for acetylation of the NatCtype N-terminal sequences in vivo (17). By using matrix-assisted laser desorption/ionization time-of-flight (MALDI-TOF) mass spectrometry, we found that deletion of any of the corresponding genes leads to the lack of acetylation of NatC-type altered iso-1-cytochromes c but not to the lack of acetylation of NatA, NatB, or NatAЈ substrates. In addition, all three deletion strains showed similar phenotypes, including slower growth on nonfermentable carbon sources at elevated temperature. Although biological functions of Mak31p and Mak10p are unknown, Séraphin (18) pointed out the sequence similarity of Mak31p to Sm proteins.
We identified previously (7) the catalytic subunit of NatB acetyltransferase, Nat3p, and suggested that, similarly to NatA and NatC, it may require other proteins for activity. In the present study, we have used the TAP protocol with Nat3p as the bait to identify Mdm20p as the other subunit of NatB, and to demonstrate that both proteins, Nat3p and Mdm20p, are required for acetylation of the NatB substrates. The corresponding deletion mutants showed similar if not identical phe-notypes, including the following: slow growth; temperature and salt sensitivity; osmotic sensitivity; calcium and caffeine sensitivity; deficiency in utilization of nonfermentable carbon sources; reduced mating efficiency; sensitivity to the mitotic drugs thiabendazole and benomyl; and susceptibility to many DNA-damaging agents, including methyl methansulfonate, bleomycin, camptothecin, and hydroxyurea. Furthermore, the phenotypes of act1, tpm1, and rnr4 mutants altered in the N-terminal region indicate that the nat3-⌬ and mdm20-⌬ phenotypes are due primarily, if not entirely, to the lack of Act1p (actin), Tpm1p, and Tpm2p (tropomyosins) acetylation. We have also demonstrated that unacetylated actin and actin altered in the N-terminal region have similar defective properties in vitro.
The ϩ 3 Ϫ Mutation Rate-The method for estimating ϩ 3 Ϫ mutation rates was based on the method of Ogur et al. (20), in which the relative mutant frequency equals the mutation rate when the culture is grown in a medium that totally prevents growth of the mutant cells. The strains were grown in YPG medium, and the frequencies of Ϫ cells were estimated from the number of small colonies on YPDG medium (19).
Construction of Deletion Mutants-Standard molecular biological procedures were performed as described (7). The NAT3 and MDM20 genes were disrupted by replacing portions of the genes with the kanMX4 gene produced by PCR and then using the appropriate fragment for yeast transformation as described earlier (7). For the nat3-⌬::kanMX4 disruption, primers Oligo1 and Oligo2 (Table II) were used to prepare the PCR fragment for transformation, and the correct disruption was identified by PCR, using the set of primers Oligo3 and Oligo4. Similarly, the fragment required for producing the mdm20-⌬::kanMX4 disruption was prepared with Oligo5 and Oligo6 and yeast genomic DNA made from a mdm20-⌬::kanMX4 deletion strain (Invitrogen) as template.
Gene Cloning and Mutagenesis-Cloning of the wild type ACT1 and TPM1 genes as well as cloning of the ACT1-203, TPM1-4 and TPM1-5 alleles was performed using PCR with oligonucleotides Oligo12 and Oligo13 for ACT1, Oligo14 and Oligo15 for both TPM1 alleles, and genomic DNAs prepared from the corresponding strains B-11679, B-14443, B-14444, and B-14445, respectively. The resulting PCR products were inserted in a TA TOPO pCR2.1 vector (Invitrogen), and the appropriate fragments were subsequently inserted in the yeast CEN URA3 vector pAA625 (pRS316). The yeast strain B-11852 (Table I) was transformed with the resulting plasmids to produce strains B-14464, B-14465, and B-14466 as well as control strain B-15073 (vector only) which were used to test the suppressor phenotypes.
Oligo16, Oligo17, Oligo18, and the ExSite TM PCR-based Site-directed Mutagenesis kit (Stratagene, La Jolla, CA) were used for making sitedirected mutations in TPM1 gene. The resulting TPM1 alleles were sequenced with Oligo14 to confirm the correct nucleotide replacements.
Subsequently, strain B-14821 was transformed with the TPM1-101 (D2H), TPM1-101 (D2S), and TPM1 plasmids and vector alone, and the phenotypes of the resulting strains (Table I) were analyzed by 1/10 serial dilutions on different media.
Similarly, wild type RNR4 was cloned by PCR using normal yeast genomic DNA, Oligo19, Oligo20, and the TA TOPO vector; the appropriate fragment was subsequently transferred to the pAA625 yeast vector. The RNR4-101 (D2H) and RNR4-101 (D2S) alleles were produced with the QuickChange® Site-directed Mutagenesis kit (Stratagene), using Oligo21 and Oligo22 for the D2H replacement and Oligo23 and Oligo24 for the D2S replacement. The rnr4-⌬ deletion strain, B-15086 (Table I), was transformed with the resulting plasmids.
Purification of Iso-1 and the Analysis of Its Acetylation Status by Mass Spectrometry-Iso-1 proteins were purified as described previously (7) by using two subsequent rounds of weak cation-exchange BioRex70 column chromatography, 100 -200 mesh and 200 -400 mesh (Bio-Rad), respectively, in potassium phosphate buffer, pH 7.0, with a 0 -1.0 M potassium chloride linear gradient. MS analysis was carried out with iso-1 proteins dialyzed against H 2 O. MALDI-TOF MS samples were prepared and analyzed in Voyager-DE STR linear time-of-flight mass spectrometer (PE Biosystems, Framingham, MA) at the Micro-Chemical Protein/Peptide Core Facility, University of Rochester, as described (7).
Purification and Characterization of the Yeast Nat3p Complex-The Nat3p protein complex was purified by TAP method (15). NAT3 was fused to the TAP tag by PCR-based tagging, using plasmid pAB2629 (originally designated pBS1479 and obtained from B. Séraphin, EMBL, Heidelberg, Germany) as template with Oligo7 and Oligo8 (Table II). Neither of the tags, ProtA and CBP, impaired protein function. The eluted complex was concentrated with the Centricon-3 (Millipore, Bradford, MA), and the subunits were separated by electrophoresis in a 4 -20% gradient Tris-glycine SDS-PAGE gel, and the protein bands were visualized by silver staining (Bio-Rad).
In Situ Gel Tryptic Digests and Peptide MS Analysis-The protein bands after staining were excised from the gel, washed, reduced, alkylated, and then incubated with trypsin essentially as described (15). The peptides were purified by Zip-tip (Millipore) before MALDI-TOF MS analysis in Voyager-DE STR mass spectrometer at the MicroChemical Protein/Peptide Core Facility, University of Rochester. Proteins were identified by searching a comprehensive non-redundant yeast TABLE II Oligonucleotides used in the construction and testing of the disrupted genes The position of the first nucleotide is presented, where A of the ORF ATG initiation codon is assigned position 1. The underlined sequences are homologous to the following regions: Oligo1 and 2, to kanMX4 cassette; Oligo7 and 8, to template plasmid pBS1479 (pAB2629); Oligo17 and Oligo18, positions of the mutated D2H and D2S in the TPM1 gene, respectively; and Oligo23 and Oligo24, positions of the mutated D2H and D2S in the RNR4 gene, respectively. Oligo11 that was used for sequencing the NAT3-TAP gene fusion is complementary to the TAP portion of the plasmid pBS1479 (original plasmid coordinate is presented). P-, in Oligo16 corresponds to the 5Ј-phosphorylated end used for mutagenesis by ExSite kit (Stratagene). ORF Oligo.
protein data base using the program MS-Fit/Prospector (University of California, San Francisco, ucsf.edu/ucsfhtml4.0/msfit.htm). Actin-activated S1 ATPase Assay-Yeast actin was isolated from each strain by DNase I affinity chromatography as described before (21). Myosin was prepared from rabbit skeletal muscle according to the method of Godfrey and Harrington (22). Myosin sub-fragment S1 was prepared according to Weeds and Pope (23). The actin-activated ATPase assays were performed according to Miller et al. (24).
In Vitro Actin Motility Assays-The actin motility assays were performed as described by Doyle et al. (25). Heavy meromyosin was prepared according to Kron et al. (26). Quantitation of the sliding velocities was performed with an ExpertVision System (Motion Analysis, Santa Rosa, CA).
Indirect Immunofluorescence Microscopy-Live or fixed cells were examined with a Eclipse E-600 fluorescence microscope (Nikon). Cells were grown to early log phase and fixed in 3.7% formaldehyde, then washed, resuspended at a concentration of 2 ϫ 10 8 cells/ml, and incubated with 20 units/ml rhodamine-phalloidin (this and other reagents for microscopy were obtained from Molecular Probes, Eugene, OR) to allow visualization of the F-actin. Vacuoles were visualized by staining live cells with N-(triethylammonium propyl)-4-(p-diethylaminophenylhexatrienyl) pyridium dibromide. Mitochondrial staining of live cells was performed with 3,3Ј-dihexyloxacarbocyanine.

RESULTS
Purification of the NatB Complex-The yeast NatB complex was purified by the tandem affinity purification (TAP) method (15), which has the advantage of using strains with normal expression of proteins. A NAT3 fusion gene was constructed by the PCR-based tagging procedure using the pBS1479 plasmid as a template and oligonucleotides complementary to the regions corresponding to the C-terminal end of the NAT3 coding region and to the 3Ј end of TAP tag-marker cassette (Table I). Both primers contained the appropriate regions of homology with the yeast genome to allow in-frame fusion of the TAP tag downstream of the NAT3 gene. Neither ProtA nor CBP tags impaired protein function as was judged by phenotype analysis of the transformants (data not presented). The resulting yeast strain, B-13396 (Table I), had normal growth rates on YPD and YPG at both 30 and 37°C, in contrast to nat3-⌬ mutants, which had slow growth phenotypes and were ts-sensitive (7). The expression of the Nat3p fusion protein in strain B-13396 was verified by Western blotting; yeast cell extract was probed with peroxidase anti-peroxidase antibody (Sigma) and a positive signal corresponding to 33-kDa band, and the molecular mass of Nat3p-ProtA-CBP was detected (data not shown).
The Nat3p fusion protein and associated components were recovered from yeast cell extracts by TAP affinity chromatography on IgG-agarose beads and by subsequent tobacco etch virus protease cleavage and purification with calmodulincoated beads. The eluted complex was precipitated; the subunits were separated by SDS-gel electrophoresis, and the protein bands were visualized by silver or Coomassie staining ( Fig.  1). Three proteins were revealed with molecular masses of 28 kDa, corresponding to the Nat3p-CBP fusion protein, and 95 and 47 kDa. In addition, the purified NatB complex was analyzed by MALDI-TOF mass spectrometry (data not presented; see also below).
The Correct Translation Start Site of Nat3p-The NAT3 open reading frame was reported previously to correspond to a 251-amino acid protein by the S. cerevisiae systematic sequencing project, as annotated in SGD (Saccharomyces genome data base, genome-www.stanford.edu/Saccharomyces/) and other data bases. However, SDS-PAGE analysis ( Fig. 1) of the Nat3p-TAP protein complex revealed a lower than expected molecular mass of the fusion protein. Tobacco etch virus protease cleavage of the fusion protein should consist of Nat3p (29,621 Da) and an additional calmodulin-binding peptide (5,070 Da), corresponding to a total molecular mass of 35 kDa rather than the observed 28 kDa. Furthermore, amino acid sequence alignment of possible Nat3p orthologous proteins from other species, including mammals (Fig. 2), indicated that yeast Nat3p has an N-terminal extension of 56 amino acid residues, if the protein initiated at the ATG site as annotated in SGD. The exact molecular mass of the yeast Nat3p fusion was determined by MALDI-TOF mass spectrometry analysis of the intact purified Nat3p-CBP complex after tobacco etch virus cleavage (data not presented), which showed that the molecular mass value of Nat3p-CBP is 27,932 Da, or Nat3p itself is 22,862 Da, subtracting the CBP domain of 5,070 Da. These results indicate that NAT3 utilizes the downstream ATG codon for initiation of protein translation.
Mdm20p Is a NatB Subunit-The protein associated with Nat3p was identified by standard peptide matching techniques. The 95-kDa band from NatB complex was excised from an SDS gel ( Fig. 1), washed, reduced, alkylated, and digested with trypsin. The resulting peptides were analyzed by MALDI-TOF MS, and the masses were compared with the yeast proteins using MS-Fit (ProteinProspector, prospector.ucsf.edu/ ucsfhtml4.0/msfit.htm). As shown in Fig. 3, 7 of 12 peptides with molecular masses in the range of 800 and 1,700 Da matched the known yeast protein Mdm20p (also designated Dec1p or YOL076w). Thus, 41% of the Mdm20p amino acid sequence was covered by the peptides detected by MS analysis after tryptic cleavage of the 95-kDa protein.
The same approach was used to establish that the 47-kDa protein band corresponded to the contaminant, rabbit immunoglobulin.
Mdm20p Is Required for N-terminal Acetylation-Crucial evidence that Mdm20p is required for N-terminal acetylation was obtained from the lack of acetylation of the NatB-type iso-1 in a CYC1-853 mdm20-⌬ strain. CYC1-853 encodes an altered iso-1 having an N terminus of Met-Glu-Phe-Leu-Ala-, which is normally acetylated but not in nat3-⌬ strains (7). Iso-1 were purified from both normal and mdm20-⌬ strains, and the protein molecular masses were compared by MALDI-TOF mass spectrometry. As shown in Fig. 4, Met-Glu-Phe-Leu-Ala-iso-1 was not acetylated in mdm20-⌬ deletion strain, similar to the results with the CYC1-853 nat3-⌬ strain (7). This clearly confirmed that Mdm20p is required for NatB activity.
Phenotypes of nat3-⌬ and mdm20-⌬-Many diverse phenotypes of nat3-⌬ (7, 28, 29) and mdm20-⌬ (30 -32) strains were previously reported in unrelated studies before the realization that Nat3p and Mdm20p have a common function. Most importantly, the identical phenotypes of the nat3-⌬ and mdm20-⌬ suggest that Nat3p and Mdm20p are part of the same NatB complex. In this study, several isogenic series (Table I) were used to compare numerous phenotypes of normal, nat3-⌬, and mdm20-⌬ strains. Both nat3-⌬ and mdm20-⌬ are viable but exhibit reduced growth at 30°C and markedly reduced growth at 37°C (Fig. 5). Furthermore, both nat3-⌬ and mdm20-⌬ strains exhibit reduced growth on the following media (see Fig.  5 for examples): nonfermentable carbon sources, such as glycerol or ethanol, as the sole source of energy; poorly utilizable carbon sources, such as galactose or raffinose; 1 M NaCl and 1 M KCl, as well as other osmotic stress agents including ethylene glycol, diethylene glycol, glycerol, and to a lesser degree sorbitol; chelating agents such as EGTA; calcium chloride; the antimitotic microtubule-destabilizing drugs thiabendazole and benomyl; DNA-damaging agents, including caffeine, bleomycin, camptothecin, HU, MMS, and EMS. It should be noted that the growth of serial dilution in different experiments, presented in Figs. 5-7, varied slightly because of different genetic background of the following normal strains: B-7687, B-11679, and B-13233 (see Table I for the genotypes).
Both nat3-⌬ and mdm20-⌬ strains exhibit abnormal mitochondrial and vacuolar inheritance. Vital staining with N-(triethylammoniumpropyl)-4-(p-diethylaminophenylhexatrienyl) pyridinium dibromide revealed that ϳ65% of the large daughter cells of the nat3-⌬ strain did not contain vacuoles or contained only fragmented portions, similar to the results reported previously (31) with the mdm20-⌬ mutant. Rhodamine-phalloidin staining revealed that nat3-⌬ cells were defective in actin cable formation (data not presented), as seen previously (30) with mdm20-⌬ cells, although actin patches in nat3-⌬ cells appeared normal. As dis- cussed below, most but not all of these phenotypes can be attributed to the lack of acetylation of actin and tropomyosins, and consequently by the defects in cytoskeletal organization.
Suppression of nat3-⌬ and mdm20-⌬ by ACT1 and TPM1 Mutants-Singer et al. (33) reported that certain phenotypes of mdm20-⌬ were partially suppressed by TPM1 and ACT1 mu-tations, including TPM1-5, TPM1-4, and ACT1-203. Interestingly, TPM1-5 encoded Tpm1p with an N terminus extended by seven amino acid residues because of initiation of translation at an upstream ATG codon and resulting in Met-His-instead of the native Ac-Met-Asp-at the N terminus. We have determined that these suppressors also act on nat3-⌬. The suppression of nat3-⌬ phenotypes by TPM1-5, TPM1-4, and ACT1-203 was determined by preparing the appropriate plasmids with PCR segments of genomic DNA corresponding to the original suppressor strains and transforming the nat3-⌬ strain B-11852 with the resulting plasmids. The phenotypes of the suppressed nat3-⌬ strains were subsequently determined by 10-fold serial dilution analysis (Fig. 6). TPM1-5 suppressed the ts phenotype of nat3-⌬, similar to the suppression of mdm20-⌬. TPM1-4 and ACT1-203 also suppressed the ts phenotype of nat3-⌬ strains but to a lesser degree. In addition, ACT1-203, TPM1-4, and TPM1-5 also partially suppressed the diminished growth of nat3-⌬ strains on a variety of media including the following: YPG; EGTA containing medium; hyperosmotic media containing 1 M NaCl and KCl salts, ethylene glycol, glycerol; calcium chloride; and caffeine-containing medium. In addition, TPM1-4 and TPM1-5, but not ACT1-203, also partially suppressed the diminished growth of nat3-⌬ strains on a medium containing bleomycin; ACT1-203 caused hypersensitivity to bleomycin. However, ACT1-203, TPM1-4, and TPM1-5 did not suppress the diminished growth on media containing the DNA-damaging agents camptothecin or hydroxyurea (Fig. 6).
Comparisons of nat3-⌬ and mdm20-⌬ Mutants with act1, tpm1, and rnr4 Mutants-The phenotypes associated with NAT mutants are due to diminished or complete loss of function caused by the lack of N-terminal acetylation of certain proteins. Because NAT mutants are viable, the lack of acetylation of essential proteins can be inferred to produce either no effect or only a partial deficiency of function. As discussed above, nat3-⌬ and mdm20-⌬ mutants have a variety of phenotypes, including slow growth, temperature, and salt sensitivity, reduced mating, diminished growth on nonfermentable carbon sources, etc. We suggested previously (7) that at least some of these phenotypes could be attributed primarily to the lack of acetylation of actin.
In order to determine which unacetylated proteins are responsible for the phenotypes associated with the nat3-⌬ and mdm20-⌬ mutants, we have examined act1, tpm1, and rnr4 mutants with amino acid changes that may be expected to mimic the corresponding unacetylated forms. The simplest of such mutants would have single amino acid replacement in the N-terminal region, and the protein could be either acetylated or unacetylated.
Wertman et al. (34) systematically generated and analyzed actin mutants containing alanine replacements throughout the molecule. The critical mutant act1-136 (D2A) encoded an altered actin that lacked one negative charge from N terminus, and in which the N-terminal alanine residue is probably acetylated by NatA as follows: normal, Ac-Met-Asp-Ser-Glu-Val; unacetylated, Met-Asp-Ser-Glu-Val; act1-136, Ac-Ala-Ser-Glu-Val. Thus, the physical properties of the act1-136 actin are similar to those of unacetylated actin. The act1-136 mutant grew slowly and was ts (34), reminiscent of nat3-⌬ and mdm20-⌬ strains.
Therefore, we performed a more comprehensive analysis of the three isogenic strains, nat3-⌬ (B-13382), mdm20-⌬ (B-14274), and act1-136 (B-13285) derived from the normal strain B-13233 (Table I). All three mutant strains exhibited very similar defects, including slow growth, temperature sensitivity, diminished growth on YPG, and media containing salts and osmotic stress agents, calcium, caffeine, hydroxyurea, thia- The diminished 42-Da mass of iso-1 from B-14069 indicates the lack of N-terminal acetylation of the Met-Glu-Phe-Leu-Ala-sequence due to the mdm20-⌬ deletion. The masses were determined with the Voyager-DE STR linear time-of-flight mass spectrometer as described previously (17). Minor peaks are due to contaminating proteins. bendazole, or camptothecin (Fig. 7). The act1-136 mutant was more sensitive to salts and other osmotic stresses, antimitotic drugs, hydroxyurea, and bleomycin, whereas the nat3-⌬ and mdm20-⌬ mutants were slightly more sensitive to camptothecin. In addition to the act1-136 mutant, we tested seven other actin mutant strains (34). These act1 mutants exhibited various degrees of sensitivity to the various agents, most likely due to various degrees of actin malfunction, with the act1-132, act1-112, and act1-122 mutants being the most sensitive. Also the mating efficiency of the act1-136 mutant was 1.7%, which is comparable with 2% mating, the value with the isogenic nat3-⌬ and mdm20-⌬ strains (see above).
Most, if not all of the nat3-⌬ and mdm20-⌬ phenotypes, can be reasonably explained by the lack of acetylation of actin. However, the results with the act1 mutants do not exclude the possibility that the lack of acetylation of other proteins is contributing to the nat3-⌬ and mdm20-⌬ phenotypes. Some of these phenotypes also may be caused by the lack of acetylation of different proteins that are epistatic to each other in regard to one or more of the phenotypes. We have investigated this possibility by examining mutant forms of Tpm1p and Rnr4p.
Urbancikova and Hitchcock-DeGregori (35) previously established the importance of N-terminal acetylation and the Nterminal region for function of tropomyosin in vitro. Unacetylated striated muscle tropomyosin expressed in Escherichia coli polymerizes poorly and binds weakly to F-actin (36). The Nterminal tail of yeast tropomyosin and its state of acetylation were also shown to be critical for protein function (37), and unacetylated Tpm1p expressed in E. coli has low affinity for actin compared with native protein (38).  6. Suppression of nat3-⌬ by the actin (ACT1-203) and tropomyosin (TPM1-5 and TPM1-4)  In this regard, proteins with penultimate histidine residues are not normally acetylated, whereas ϳ95% of the proteins starting with serine are acetylated (5). However, the rather basic nature of the N-terminal region of the D2S sequence suggests that it is also unlikely to be acetylated or only partially acetylated (5).
The TPM1-101 (D2H) and TPM1-102 (D2S) alleles were constructed by site-directed mutagenesis. The appropriate fragments were inserted in a CEN vector, and strain B-14821 (tpm1-2 tpm2-⌬) was transformed with the resulting plasmids, pAB2902 and pAB2915, respectively. B-14821 is unable to grow above 36°C because of the ts mutation, tpm1-2, and the lack of the second isoform due to the tpm2-⌬ deletion. The growth of B-15077 (TPM1-101, D2H) and B-15080 (TPM1-102, D2S) strains were compared with control strains B-15078, containing vector alone, and B-15074, containing the wild type TPM1 allele. Growth on different media (Fig. 8) revealed that the normal TPM1 and the altered TPM1-101 (D2H) tropomyosins were functional but that the TPM1-102 (D2S) tropomyosin was partially defective and that TPM1-102 strains grew poorly at 37°C. These results reveal the importance of the N-terminal region of tropomyosin and suggest that the residue at position 2 may be critical.
We have also explored the possibility that the lack of acetylation of proteins other than Act1p or Tpm1p can cause sensitivity to camptothecin and hydroxyurea (Fig. 6). One of the possible candidates whose acetylation could be important for its function in DNA damage repair is Rnr4p, which was directly shown to be a NatB substrate (7). Rnr4p is a subunit of ribonucleotide reductases, which catalyze the formation of deoxyribonucleotides by the reduction of the corresponding ribonucle-otides (39). RNR4 is not an essential gene, but rnr4-⌬ strains grow slowly and are sensitive to the DNA damage reagent, hydroxyurea, a phenotype that is characteristic of nat3-⌬ and mdm20-⌬ mutants; also rnr4-⌬ strains display an S-phase arrest at restrictive temperature (40).
The importance of the N-terminal region of Rnr4p was investigated by examining the phenotypes of plasmid-borne RNR4 mutants in an rnr4-⌬ strain: normal, Ac-Met-Glu-Ala-His-; RNR4-101 (E2H), Met-His-Ala-His-; and RNR4-102 (E2S), Ac-Ser-Ala-His-. Growth on SD and SD ϩ HU media for 3 days of serial dilutions revealed that both N-terminally altered proteins complemented the hydroxyurea sensitivity phenotype to the same degree as the wild type Rnr4p (data FIG. 8. Function of the tropomyosin alleles TPM1-101 and  TPM1-102 in the tpm1-2 tpm2-⌬  Activities of Unacetylated and DNEQ Actins in Vitro-Because in vivo function of unacetylated actin could not be directly addressed, in vitro analysis of three forms, normal, unacetylated, and DNEQ, described in Table III, was performed. DNEQ actin, containing the replacements D2N and E4Q, has been the subject of several studies (21,24,42,43). Unacetylated actin of Dictyostelium discoideum prepared in vitro was shown previously (45) to have a weaker interaction with actinomyosin. Functions of the actins were assessed by actin-activated S1 (skeletal-muscle myosin subfragment 1) ATPase assays, in vitro by motility assays, and by interaction with other proteins. DNEQ actin as well as other actins with replacements of charged residues have been reported (24,43,46) to result in an almost complete loss of activation of S1 ATPase activity, except at high actin concentrations. A decrease in activation was also observed for the unacetylated actin filaments ( Fig. 9 and Table III), although at the concentrations used in this assay, which were also the upper limits available, an accurate determination of the V max and K m values for each of the two mutant actins was not possible. However, the results shown in Fig. 9 demonstrate that S1 ATPase activity with unacetylated actin has a K m value approximately twice that of normal actin, whereas V max is unaffected. From the plot, a K m of ϳ73 M for the unacetylated actin might be interpolated, but because the highest concentration of actin in the assay is only 40 M, this value is only approximate. Normal actin shows an activation of S1 ATPase with K m ϭ 19.8 Ϯ 7 M and V max ϭ 3.0 Ϯ 0.5 s Ϫ1 . The unacetylated actin appears to have an activation 40 -60% that of the wild type protein, whereas that of the DNEQ mutant, as well as other charge mutants reported previously, was only ϳ10% that of the normal protein (Table III). Analysis of other mutants that have no charge at the N terminus are consistent with this result.
The ability of the unacetylated mutant actin filaments to slide over skeletal muscle heavy meromyosin was compared with the DNEQ actin and normal actin filaments by the in vitro motility assay. Because it was known that actin filaments of the DNEQ mutant fail to bind in this assay under standard conditions, even at low ionic strength of 50 mM (24), assays were performed in the presence of the viscosity agent methylcellulose. Under these conditions, velocities of both the unacetylated actin (4.3 Ϯ 0.8 m/s) and DNEQ actin (4.1 Ϯ 0.8 m/s) were similar to that observed for the wild type actin filaments (3.9 Ϯ 0.9 m/s; see Table III). In the absence of methylcellulose, however, the number of mutant filaments binding to the myosin heads is significantly decreased, as the filaments dissociated from the assay surface following the addition of MgATP. As a result, significantly less of the unacetylated and DNEQ actin filaments moved in the assay compared with the normal filaments (Table III). This behavior was previously observed with the DNEQ actin as well as other actin mutants that have replacements or deletions of charged residues (24). Thus, it is reasonable to conclude that the abnormal properties of the unacetylated actin in vitro are reflected in the phenotypes of the nat3-⌬ and mdm20-⌬ mutants.

DISCUSSION
All Major NATs Have Heteromeric Composition-In this study, we have established that Mdm20p interacts with the catalytic subunit Nat3p and is required for NatB N␣-terminal acetyltransferase activity. Thus, NatA, NatB, and NatC all have subunits that interact with their respective catalytic subunit and that are required for N-terminal acetyltransferase activity (Table IV). NatA, NatB, and NatC are required to acetylate different subclasses of proteins, and mutants of each N-terminal acetyltransferase have different phenotypes. On the other hand, mutations of the catalytic subunit and the associated subunits for each of the NATs cause identical phenotypes. As shown in this study, nat3-⌬ and mdm20-⌬ strains have identical or nearly the same phenotypes. Similarly, ard1-⌬ and nat1-⌬ strains have the same phenotypes (10) as do mak3-⌬, mak10-⌬, and mak31-⌬ strains (17). Although the sequences of the catalytic subunits Ard1p, Nat3p, and Mak3p are similar to each other (7), there is no significant sequence similarity among the auxiliary subunits, Nat1p, Mdm20p, Mak10p, and Mak31p, or the auxiliary subunits with the catalytic subunits. The function of the auxiliary subunits is unknown, but they may play a role in recognition of the proper NAT substrate. It is also unknown if there are still other auxiliary subunits required for function of NatA, NatB, and NatC. In this regard, it would be of interest to determine whether Asc1p and other proteins associated with Ard1p and Nat1p (see above) are required for NatA function. The increased activity of NatA in strains overexpressing just ARD1 FIG. 9. Actin-activated ATPase activities of S1 in the presence of wild type and mutant actins. S1 ATPase activation by the following actins (Table III) was measured in the presence of between 5 and 40 M phalloidin-stabilized F-actin, 0.4 M S1 and 1 mM MgATP: normal (q); unacetylated (); and DNEQ (E).

TABLE III
Description, relative S1-activating ATPase activity, and motility assays of the three forms of actin Velocities were measured in the presence of methylcellulose. DNEQ was previously shown to be N-terminally acetylated (41) and is most likely a NatB substrate (5 and NAT1 suggests that Asc1p is not required at least at a stoichiometric level for NatA function. NAT Phenotypes and Substrates-Although they have pronounced phenotypes, it is somewhat surprising that the NatA, NatB, and NatC deletion strains are viable. Approximately one-half of yeast proteins are N-terminally acetylated, with most being NatA substrates (3)(4)(5). We can consider that the lack of N-terminal acetylation diminishes activity or stability to various degrees, from no detectable level to the complete loss of function. Clearly, the complete loss of function of essential proteins does not occur by the lack of N-terminal acetylation in yeast. In fact, the common use of N-terminal fusions of peptides and proteins for convenient identification and purification procedures indicates that the state of the N terminus is by-andlarge innocuous for most proteins. On the other hand, the lack of acetylation of the viral major coat protein, gag, leads to the loss of maintenance of the double-stranded RNA virus or to the Mak Ϫ phenotype, and this deficiency can be considered as complete (13,14). In contrast, as emphasized in the present study, the lack of N-terminal acetylation of the essential protein, actin, leads to only a partial loss of activity.
The phenotypes of NAT mutants can be used to assess the importance of N-terminal acetylation of specific proteins. In addition to the Mak Ϫ phenotype, the mak3-⌬, mak10-⌬, and mak31-⌬ mutants also have reduced growth on media containing nonfermentable carbon sources as the sole source of energy, such as glycerol or ethanol (Nif Ϫ ), especially at 37°C. Tercero et al. (14) suggested that the diminished growth of mak3-⌬ strains on glycerol medium was due to lack of N-terminal acetylation of one or more of the mitochondrial proteins Kgd1p (␣-ketoglutarate dehydrogenase), Fum1p (fumarate dehydratase), and Mrp1p (a mitochondrial ribosomal protein), which all contained Met-Leu-Arg-Phe-termini and all may be NatC substrates. The corresponding deletion mutants all have mitochondrial defects; the kdg1-⌬ mutant was characterized as respiration-deficient and Nif Ϫ , the fum1-⌬ mutant, is respiration-defective; and the mrp1-⌬ mutant was found to produce high frequencies of Ϫ cells (SGD Basic Information, genomewww.stanford.edu/Saccharomyces/). On the other hand, the unacetylated proteins responsible for the main nat1-⌬ and ard1-⌬ phenotypes (slower growth, derepression of the silent mating type gene HML␣, and failure to enter G o (10), have not been identified; these phenotypes can be attributed to the lack of N-terminal acetylation of one or more of over 2,500 NatA substrates.
As discussed above, the various phenotypes of nat3-⌬ and mdm20-⌬ mutants have been reported not only in this study but also in previous studies directed toward either NAT3 (7,28,29) or MDM20 (30,32,33,47). We have attributed the nat3-⌬ and mdm20-⌬ phenotypes to the lack of acetylation of actin and tropomyosins, because of the similar phenotypes of certain act1 and tpm1 mutants. Most importantly, these act1 and tpm1 mutants contained amino acid replacements in the N-terminal regions of Act1p and Tpm1p, respectively, possibly mimicking the unacetylated forms.
The following phenotypes can be rationalized as due to diminished actin activity: defective actin cables formation; osmotic sensitivity; diminished mating; random budding pattern; abnormal mitochondrial and vacuolar inheritance; high Ϫ production; diminished growth at elevated temperatures and on nonfermentable media. The diminished growth on media containing antimitotic drugs, such as thiabendazole and benomyl, also could be attributed to the abnormal actin in light of a recently proposed potential link between actin cables and microtubules (48). However, the diminished growth on media containing DNA-damaging agents, such as bleomycin, camptothecin, hydroxyurea, MMS, and EMS, could be indirect and caused by partial deficiency of unidentified proteins due to the actin defect. We have not excluded the possibility that some of these phenotypes can also be caused by the lack of acetylation of other proteins. Although the analysis of RNR4-101 and RNR4-102 mutants appeared to exclude the possibility that unacetylated Rnr4p produces a phenotype, there are numerous proteins that function in DNA replication and DNA damage repair and that may be substrates for NatB because of their Met-Glu-or Met-Asp-termini, including, for example, those encoded by ERG3, SFP1, CWH36, and YLR320w.
Unacetylated Actin-The phenotypes in vivo of nat3-⌬ and mdm20-⌬ strains, and the defects in vitro that were caused by unacetylated actin, can be related to numerous studies of actin having amino acid changes, especially those causing a diminished charge. Many actin mutants are temperature-and saltsensitive and defective in other cytoskeleton functions, including act1-136, which has a D2A replacement (34).
Acidic residues on the surface of subdomain 1 of actin have been implicated in the weak actomyosin interaction and in its function (24,43,46,49). Furthermore, N-terminal acidic residues play an important role in the activation of myosin ATPase, and much of the diminished activity of the mutant forms can be attributed to a decreased weak binding interaction (24). At low ionic strength conditions and in the presence of methylcellulose, the DNEQ and ⌬DSE (lacking residues 2-4) actins, as well as unacetylated actin, moved in the in vitro motility assays at a mean velocity similar to that of wild type actin. Yet the sliding velocity of the N-terminal mutant actins decreased relative to that of the wild type at all levels. At higher ionic strength conditions, the sliding of the DNEQ and ⌬DSE actins ceased even in the presence of methylcellulose (44). Thus, the N-terminal actin residues contribute to acto-S1 binding in both the weakly and strongly bound states.
The addition of two extra acidic charges to the N terminus of the actins in the D24A and D25A mutants or the E99A and E100A mutants results in actins with the same overall charge density of sub-domain 1 as the wild type yeast actin. These mutants have identical sliding and force generation in the in vitro motility assays and indistinguishable ATPase activities from the wild type actin protein (50). These results suggested that the locations of the charged residues on the surface of sub-domain 1 of actin are not important for the weak actomyosin interactions and actomyosin function, but rather the overall charge density of this region is important for these functions. The kinetics of interactions of normal actin does not differ from the form having the presence of extra charged residues at the N terminus, as found in mammalian skeletal actin, which has four acidic residues at the N terminus (21,24). This suggests that the two acidic residues found in the yeast protein are the minimum needed for correct function. The results presented in this study indicate that incorrect process- a Acetylation occurs only on subclasses of proteins containing the indicated termini, except for Met-Glu-and Met-Asp-termini, which are apparently always acetylated.
ing of the N terminus of the actin molecule, whereby the acetyl group is not transferred to the positively charged terminus, results in a protein with actomyosin interactions falling between the wild type protein and mutants lacking two acidic residues. The net change of charge density expected from the unacetylated actin is ϩ1 compared with the wild type protein, whereas that for the removal of two acidic residues is ϩ2. As such, the lack of acetylation of the actin molecule only moderately affects the weak actomyosin interactions and actomyosin function. The wild type yeast actin therefore appears to have a charge density on sub-domain 1 that is optimal for actomyosin function, and that an increase in this charge density will affect interactions with other proteins. Thus, the correct N-terminal processing is essential for a fully functional protein, and the absence of processing observed in the nat3-⌬ mutant will result in altered interactions with other proteins of the cytoskeleton, including myosin.
Unacetylated Actin Fails to Form Completely Functional Cables-Previous work by Hermann et al. (30) and Singer et al. (33) revealed that mdm20-⌬ strains were defective in mitochondrial inheritance and actin cables and that extra copies of TPM1, a gene encoding the actin filament-binding protein tropomyosin, suppress mitochondrial inheritance defects and partially restore actin cables in mdm20-⌬ cells. Synthetic lethality was also observed between mdm20 and tpm1 mutant strains and certain dominant alleles of ACT1-and TPM1-suppressed mdm20-⌬ phenotypes. However, these authors did not identify the primary function of Mdm20p. In light of our finding, we can attribute the phenotypes described by Hermann et al. (30) and Singer et al. (33) to the lack of acetylation of actin, tropomyosin, and possibly other essential cellular proteins. Unacetylated actin fails to form completely active cables and to provide the normal cytoskeleton function.
Unacetylated Tropomyosin-Tropomyosin, an archetype of ␣-helical coiled coils, has a sequence with a 7-residue (so-called "heptad") repeat of the form a-b-c-d-e-f-g, with a and d residues being apolar (51). The first and last 9 residues of tropomyosin form an overlap region of the C terminus of one molecule with N terminus of the next molecule, a finding that was confirmed recently by crystal structure of N-terminal fragment (51). This overlapping region is required for polymerization in vitro and cooperative actin binding that results in high actin affinity and head-to-tail alignment of tropomyosin on the thin filament (52). Tropomyosin coiled coils are aligned head-to-tail in the grooves of the helical actin filament. In yeast, tropomyosin was thought previously to function in assembly and stabilizing of actin cables (53), playing only a structural role, or additionally regulating actomyosin ATPase.
N-terminal acetylation is important for cooperative binding to actin and head-to-tail polymerization of tropomyosin (35,38). Cho et al. (54) postulated that the N terminus might be an actin-binding site that contributes more to overall affinity than the internal periodic sites in tropomyosin. Recently the molecular requirements for the intact N terminus of tropomyosin were explained by the formation of coiled-coil structure that is necessary for head-to-tail binding (51). Thus, it is expected that the lack of acetylation of tropomyosin or removal of one negative charge, as in D2S Tpm1p mutant, should produce the phenotypes in nat3-⌬ and mdm20-⌬ strains. We also suggest that the D2S mutant had a diminished function in vivo because of not only the lack of acetylation, but disruption in "heptad" repeats as well. On the other hand, the D2H mutant apparently can compensate for lack of acetylation. This is not overly surprising because Hitchcock-DeGregori (52) reported that Met-Arg-tropomyosin binds with high affinity to actin. Al-though studies in vitro are necessary to confirm these conclusions with the yeast D2H and D2S tropomyosins, it is reasonable to suggest that unacetylated tropomyosin may cause growth defects in vivo.
It is also interesting to note that unacetylated yeast tropomyosins did not restore the polymerization of a defective mutant actin, GG (V266G and L267G) at restrictive temperature, whereas cardiac tropomyosin and yeast Tpm1p and Tpm2p did rescue such a defect (55). Therefore, the N-acetylation is crucial for interaction of tropomyosin with actin monomers at the end of the filament. Apparently, the addition of the new actin monomer to the filament end becomes favorable by the conformational change of tropomyosin due to N-acetylation.