A novel acetyltransferase found in Saccharomyces cerevisiae Sigma1278b that detoxifies a proline analogue, azetidine-2-carboxylic acid.

L-Azetidine-2-carboxylic acid (AZC), a toxic four-membered ring analogue of L-proline, is transported into the cells via proline transporters. It causes misfolding of the proteins into which it is incorporated competitively with L-proline and thereby inhibits the growth of the cells. We recently have discovered, on the chromosome of Saccharomyces cerevisiae Sigma1278b, a novel gene MPR1 required for the resistance of Sigma1278 background strains to toxic AZC. This gene was missing in the particular yeast strain used for the genomic sequence determination. Although the protein sequence was homologous to that of the S. cerevisiae transcriptional regulator, Mpr1p did not affect the expression of genes involved in proline uptake. However, gene expression in Escherichia coli and enzymatic analysis showed that the MPR1 gene encodes a novel AZC acetyltransferase, by which L-proline itself and other L-proline analogues are not acetylated. Mpr1p was considered to be a member of the N-acetyltransferase superfamily based on the results of an Ala-scan mutagenesis through the highly conserved region involved in binding acetyl-CoA in members of the superfamily. Our findings suggest that Mpr1p detoxifies AZC by acetylating it in the cytoplasm. This enzyme might be utilized as a selective marker in a wide variety of organisms, because the cells expressing the MPR1 gene acquire the AZC-resistant phenotype.

Some amino acid analogues are known to induce in cells a transient stress response comparable with that of heat shock stress (1)(2)(3). The addition of certain amino acid analogues or mild heat shock causes a rapid increase in the synthesis of heat shock proteins in cells. The elevated levels of synthesis for this set of proteins begins to decrease shortly after restoration of the normal amino acids or normal temperature. Therefore, amino acid analogues have proved to be valuable reagents for studying cellular metabolism and the regulation of synthesis of macromolecules in both prokaryotic and eukaryotic cells (1)(2)(3). A toxic four-membered ring analogue of L-proline, L-azetidine-2-carboxylic acid (AZC) 1 , is incorporated into proteins compet-itively with L-proline and causes the synthesis of abnormal misfolded proteins, thereby inhibiting cell growth in both bacterial and animal cells (4 -7). In bacteria such as Escherichia coli and Serratia marcescens, L-proline biosynthesis from Lglutamate is regulated mainly through feedback inhibition by L-proline of ␥-glutamyl kinase, a key enzyme in the pathway (8,9). Because this enzyme is also feedback-inhibited by L-proline analogues such as AZC, L-proline-overproducing strains have been successfully isolated from L-proline-analogue-resistant mutants, which have a mutation in the proB gene coding for ␥-glutamyl kinase that causes desensitization to feedback inhibition (10,11).
To investigate the cryoprotective effect of L-proline on the freezing stress of yeast, we previously isolated AZC-resistant mutants from an L-proline-nonutilizing strain of Saccharomyces cerevisiae (12). In the process of this study, it was noteworthy that S. cerevisiae strain ⌺1278b showed greater AZC resistance than that of other laboratory strains, including the genome project strain S288C. The strain ⌺1278b, mainly used in European laboratories, is known to have unique genetic features for nitrogen metabolism and morphological characteristics for diploid pseudohyphal development and haploid invasive growth (13)(14)(15). We recently isolated novel genes involved in AZC resistance from the genomic library of strain ⌺1278b (16). Intriguingly, the novel genes MPR1 and MPR2 (sigma 1278b gene for L-proline-analogue resistance) required for AZC resistance were present on chromosomes XIV and X of strain ⌺1278b, respectively, but were absent in other laboratory strains (S288C etc.). The primary structure of the predicted Mpr proteins was found to be a member of the N-acetyltransferase (NAT) superfamily (17), although one amino acid change at position 85 occurred between the MPR1 and MPR2 genes. Both MPR genes are expressed in other S. cerevisiae strains, where they play global roles in AZC resistance (16). Previous reports have described only a point mutation or a deletion of a few bases, based on comparisons of certain genes between ⌺1278b and S288C strains (18 -20). To our knowledge, ours is the first report of novel genes present in strain ⌺1278b and absent in other laboratory strains. However, the detailed function of the MPR genes in AZC resistance has been unclear.
Thus, in the present study, we analyzed the function of the MPR1 gene required for AZC resistance in S. cerevisiae and E. coli. We show here by Ala-scan mutagenesis that the MPR1 gene encodes a novel AZC acetyltransferase and that Mpr1p belongs to a superfamily of NAT. In addition, a possible mechanism for AZC resistance by Mpr1p in vivo will be discussed.
Culture Media-The media used for growth of S. cerevisiae were SD (2% glucose, 0.67% Bacto yeast nitrogen base without amino acids (Difco Laboratories)) and YPD (2% glucose, 1% Bacto yeast extract, 1% Bacto peptone). SD medium contains ammonium sulfate (0.1%) as the nitrogen source. To examine the MPR1 gene function, 0.1% proline was used instead of ammonium sulfate as the sole source of nitrogen. When appropriate, required amino acids were added to the media for auxotrophic strains. Yeast strains were also cultured on SD agar plates containing L-proline analogue, AZC, 3,4-dehydro-DL-proline (DHP), or L-thiazolidine-4-carboxylic acid (TAC). All the analogues were obtained from Sigma. The E. coli recombinant cells were grown in M9 medium (21) containing 2% casamino acids and 50 g/ml ampicillin (M9CA). If necessary, 2% agar was added to solidify the medium.
Subcellular Localization of Mpr1p-Plasmid pRSMPR was first constructed by subcloning of the 3.7-kb SacI-SacI fragment containing the MPR1 gene from pMH1 into the SacI site of pRS416. The 750-bp XhoI-AatII fragment containing a mutated Aequorea GFP gene from pGreenscript A (22) was then ligated into the NcoI site 3 bp upstream of the termination codon in the MPR1 gene of pRSMPR by blunt-end ligation. The plasmid in which the GFP gene was fused to the carboxyl terminus of Mpr1p in the sense orientation was designated as pM-PRGFP. The S. cerevisiae CKY2 was transformed by pMPRGFP, and the Ura ϩ and AZC-resistant transformants were visualized with a confocal laser scanning microscopy (INSIGHT PLUS; Meridian Instruments).
Expression of the MPR1 Gene in E. coli-The enzymes used for DNA manipulations were obtained from Takara Shuzo. To introduce the NcoI sites on the amino terminus and carboxyl termini, polymerase chain reaction (PCR) was carried out with pMH1 as a template and primers 5Ј-AGCCATGGATGCGGAATC-3Ј and 5Ј-GTCCATGGTTATTCCAT TGAGAGGA-3Ј (the underlined sequence is the position of a NcoI site). The amplified DNA was digested with NcoI, and the ends were filled with T4 DNA polymerase. The blunt-ended DNA was ligated into the SmaI site of pQE30. The plasmid in which the MPR1 gene was placed under the T5 promoter/lac operator in the sense orientation was designated as pQE-MPR. The pQE-MPR carries the entire MPR1 gene with the sequence coding six consecutive histidine residues and nine additional amino acid residues (Gly-Ser-Ala-Cys-Glu-Leu-Gly-Thr-Pro) at the amino terminus. The nucleotide sequence was confirmed with a Model 377 DNA sequencer (PE Biosystems). The E. coli strain JM109 was transformed with pQE-MPR, and the recombinant cells were grown at 37°C in M9CA medium. When absorbance at 600 nm reached 0.5, IPTG was added to the culture medium to a final concentration of 1 mM to induce gene expression. After cultivation for 4 h at 37°C, the cells were harvested by centrifugation, and cell-free extracts were prepared by sonic oscillation under cooling. The His-tagged fusion proteins in the soluble fraction were then purified using the nickel-nitrilotriacetic acidagarose supplied by Qiagen.
Site-directed Mutagenesis-The replacements of Arg 145 , Gly 146 , Gln 147 , Lys 148 , Val 149 , and Gly 150 by Ala and of Lys 148 by Gly in Mpr1p were performed by PCR with oligonucleotide primers R145Aϩ (5Ј-GT-GCCCATG*C*T*GGTCAGAA-3Ј), R145AϪ (5Ј-TTCTGACCA*G*C*AT- , and K148GϪ (5Ј-GTAGCCAACA*C*C*CTGACCT-3Ј), respectively, using a Gene Amp PCR system 2400 (PE Biosystems). The asterisks show the locations of mismatches. A plasmid pMPR-BM was constructed by ligating the 1.6-kb BglII-MluI fragment containing the MPR1 gene from pMH1 to the large fragment of pYES2 digested with BamHI and MluI and used as a template DNA for site-directed mutagenesis. In addition, primers 5Ј-CACAG TTTGACA-AAACAGGG-3Ј and 5Ј-CGACGCGTCGTTATTCGTTC-3Ј were synthesized to complement regions upstream and downstream of the EcoRI restriction sites in pMPR-BM. The unique amplified band of 630 bp was digested with EcoRI to recover the 340-bp fragment and ligated to a 6.9-kb fragment of plasmid pMPR-BM digested with EcoRI. The mutations were confirmed by DNA sequencing.
Northern Blot Analysis-Northern blot analysis was carried out using Gene Images random-prime labeling and detection system (Amersham Pharmacia Biotech). Total RNA from S. cerevisiae was isolated by the method of Köhrer and Domdey (23). RNA was separated in 1.0% agarose gel and transferred to nylon membrane. As a DNA probe, the DNA fragments of the GAP1, PUT4, MPR1, and ACT1 genes were prepared by PCR with oligonucleotide primers GAP1ϩ (5Ј-TTGAC- , and ACT1Ϫ (5Ј-CGG-GATCCGGGCTCTGAATCTTTCGT-3Ј), respectively. Each unique amplified band was purified from agarose gel, denatured, and labeled according to the protocol recommended by the supplier.
Western Blot Analysis-Total cellular proteins from extracts of S. cerevisiae and E. coli were separated on SDS-polyacrylamide gel electrophoresis and transferred to poly(vinylidene difluoride) membrane. To determine whether Mpr1p contains an asparagine-linked carbohydrate, the S. cerevisiae extracts expressing the MPR1 gene were also digested with endoglycosidase H (New England Biolabs) under conditions recommended by the supplier. The wild-type and mutant proteins were detected by Western blot analysis using the Phototope-Star detection kit (New England Biolabs) and anti-Mpr1p polyclonal antibody prepared from the recombinant His-tagged Mpr1p overproduced in E. coli.
Assay of Acetyltransferase Activity-To determine acetyltransferase activities, the whole-cell extracts of S. cerevisiae and E. coli strains expressing the MPR1 gene were prepared by vortexing the cells with glass beads and by sonic oscillation under cooling, respectively, and used as enzyme sources. The acetyltransferase activity was assayed at 30°C by monitoring the increase of 5-thio-2-nitrobenzoic acid (TNB) because of the reaction of acetyl-CoA with 5,5Ј-dithiobis(2-nitrobenzoic acid) (DTNB) described previously with some modifications (24). The initial rate of the increase in absorbance at 412 nm of the reaction mixture (final volume, 1 ml) containing 50 mM acetate ammonium (pH 7.5), 1 mM EDTA, 1 mM L-proline, AZC, DHP, or TAC, 1 mM DTNB, 0.1 mM acetyl-CoA (Wako Pure Chemical) and enzyme solution was measured, and that obtained for a solution containing all the materials except L-proline, AZC, DHP, or TAC (blank) was subtracted. The reaction rate was calculated using an extinction coefficient for TNB of 15,570 M Ϫ1 cm Ϫ1 . One unit is defined as the amount of enzyme catalyzing the formation of 1 mol TNB/min at 30°C. Protein concentrations were determined with a Bio-Rad protein assay kit. Bovine serum albumin was used as the standard protein. The amount of AZC in the reaction mixture was quantified with an amino acid analyzer (L-8500A; Hitachi).

RESULTS
Identification of the MPR1 Gene Product-The MPR1 mRNA was identified by RNA hybridization analysis of total yeast RNA as a single 1.0-kb transcript in the strains L5685 and CKY2 (pMH1) in proportion to the copy number of the MPR1 gene (data not shown). Bands corresponding to the molecular mass of Mpr1p (ϳ26 kDa) were also detected in the same strains expressing the MPR1 gene by Western blotting using The S. cerevisiae MPR1 Gene Encodes an AZC Acetyltransferase an anti-Mpr1p polyclonal antibody (data not shown). Mpr1p was shown to be a nonglycoprotein, based on the finding that the apparent molecular mass after digestion with endoglycosidase H was similar to that of the untreated protein (data not shown), consistent with the fact that the Mpr1p sequence lacks consensus sites for N-linked glycosylation.
MPR1 Does Not Affect the Expression of Genes Involved in Proline Uptake-The Mpr1p sequence of 229 amino acid residues was found to be homologous only to the amino-terminal sequence of the S. cerevisiae SPT10 (SUD1)-encoded protein with 640 amino acids, a negative transcriptional regulator (25)(26)(27). Both L-proline and AZC are transported into the yeast cells via two transporters, the general amino acid permease (encoded by GAP1) and the proline-specific permease (encoded by PUT4) (28). We therefore examined whether the MPR1 gene functions as an Spt10p-like transcriptional regulator, based on the hypothesis that Mpr1p represses GAP1 and PUT4 genes, leading to the AZC-resistant phenotype. Northern blot analysis showed quantitatively similar amounts of GAP1 and PUT4 transcripts in the strains L5685, LD1014, CKY2 (pMH1), and CKY2 (pYES2) in the presence or absence of AZC (data not shown). Also, when the above yeast strains were grown on a minimal agar medium containing L-proline as the sole nitrogen source, the growth did not vary from that observed with ammonium sulfate as the nitrogen source, suggesting that the proline permeases function normally even in the presence of the MPR1 gene (data not shown).
Further, Mpr1p did not include any amino acid sequences that are similar to the DNA binding motifs and the aminoterminal basic domains found in other yeast transcriptional factors (26,27). A preliminary experiment showed that Mpr1-GFP fusion protein stayed primarily in the cytoplasm and did not localize in the nucleus (data not shown). These findings indicate that Mpr1p involved in AZC resistance is not a transcriptional regulator, which does not repress the yeast proline permease genes GAP1 and PUT4.
Effect of the MPR1 Gene on Growth Inhibition by Other L-Proline Analogues-To further analyze the function of the MPR1 gene, various S. cerevisiae strains were cultured on SD agar plates containing each of three toxic L-proline analogues (AZC, DHP, and TAC) (Fig. 1). The MPR1 gene was shown to confer only a four-membered ring analogue-AZC resistance to the yeast cells, not a five-membered ring DHP and TAC. Therefore, Mpr1p is supposed to work on AZC directly, and bacterial cells such as E. coli, which are naturally AZC-sensitive, would therefore become resistant to AZC when the MPR1 gene is successfully expressed in the cells. In addition, it is worth noting that the strain with the ⌺1278b background, regardless of the MPR genes, was less sensitive to DHP and TAC relative to the strain with the S288C background.
Acetyltransferase Activity of Mpr1p-Based on the hypothesis described above, we constructed an expression plasmid pQE-MPR for the MPR1 gene under the IPTG-inducible promoter. Surprisingly, the E. coli JM109 cells expressing the MPR1 gene in the presence of IPTG were capable of growing on agar plates containing AZC, thereby acquiring the AZC-resistant phenotype, whereas the vector-harboring cells failed to grow on AZC-containing plates ( Fig. 2A). Based on the finding that Mpr1p sequence was a member of the NAT superfamily, we examined the acetyltransferase activities for L-proline and various analogues (AZC, DHP, and TAC) in total cell extracts from the S. cerevisiae L5685, CKY2, and E. coli JM109 cells expressing the MPR1 gene (Fig. 2B). Among the substrates used, only acetyltransferase activities of AZC were clearly detected in proportion to the level of gene expression. No measurable activities were observed from the cell extracts harboring the vector only and with the substrates of L-proline, DHP, or TAC having a five-membered ring structure (data not shown). These results showed that the MPR1 gene encodes an AZC acetyltransferase that does not acetylate the five-membered ring compound L-proline, DHP, or TAC.
Ala-scan Mutational Analysis of Mpr1p-The NAT superfamily, which includes over 50 members, has four conserved regional motifs, A-D (17). In particular, the most highly conserved motif A would be involved in the binding of acetyl-CoA (29,30) and has a short consensus sequence, (Q/R)XXGX(G/A), in common with members of the superfamily (Fig. 3A). Further, point mutations in any of these residues were found to impair the activities of NAT enzymes (31,32). In the case of the Mpr1p sequence, the corresponding region consists of Arg 145 -Gly 146 -Gln 147 -Lys 148 -Val 149 -Gly 150 , with high degrees of similarity to other members, excepting position 148. To examine the function of Mpr1p as a member of the NAT superfamily, the six consecutive residues were individually changed to Ala by sitedirected mutagenesis, and the resulting mutants were characterized in terms of AZC resistance and acetyltransferase activity. In addition to these mutations, Lys 148 was changed to consensus Gly. All mutant genes were expressed using the original promoter and terminator of MPR1 to obtain normal levels of gene expression. The S. cerevisiae strain CKY2 with a S288C background was transformed with these plasmids, and the Ura ϩ transformants were cultivated in SD medium supplemented with His. The levels of MPR1 gene expression were almost the same for all mutants, based on densitometric estimates of the amounts of accumulated Mpr1p in the soluble fraction by Western blot analysis (Fig. 3B). Fig. 4C shows the growth phenotype on AZC-containing medium and the AZC FIG. 1. Mpr1p confers only AZC resistance to the yeast cells. The structural formula of each L-proline analogue is shown. Approximately 10 6 cells of each strain and serial dilutions of 10 Ϫ1 to 10 Ϫ3 (from left to right) were spotted onto SD plates in the absence (Ϫanalogue) and presence (ϩAZC, ϩDHP, and ϩTAC) of each analogue. Plates were incubated at 30°C for 3 days. acetyltransferase activities of strains having various mutant Mpr1 proteins. The enzymes in the whole-cell extracts were assayed without further purification. The Ala substitution of the highly conserved residues Arg 145 , Val 149 , and Gly 150 led to a growth defect in AZC and simultaneously resulted in the complete loss of AZC acetyltransferase activity. These residues in Mpr1p that corresponded to the consensus sequence of the NAT superfamily were found to be necessary for activity. On the other hand, G146A and Q147A mutants displaying AZC resistance in the growth assay also retained a significant level of activity for AZC acetylation. These residues are unlikely to be crucial for the function of Mpr1p. It was also shown that Gly or Ala substitution at position 148 had no influence on this activity. This is probably explained by the fact that Lys or Arg naturally occupies this position in some NAT members (17). AZC resistance and acetylation activity in the strains producing the various Mpr1 proteins correlated with each other. These results strongly suggest that Mpr1p is a novel member of the NAT superfamily.
AZC Acetylation by Mpr1p-The His-tagged fusion Mpr1p was purified from the recombinant E. coli cells expressing the MPR1 gene as described under "Experimental Procedures." A 2-ml reaction mixture containing 100 mM ammonium acetate (pH 7.5), 2 mM EDTA, 50 mM AZC (1 mg), 50 mM acetyl-CoA, and 15 milliunits of the recombinant enzyme was incubated at 30°C. The gradual decrease in AZC in the reaction mixture was observed with the time of incubation, whereas the initial content of AZC still remained under this condition without the enzyme (Fig. 4A). However, no new peak corresponding to acetylated AZC appeared, possibly because the ninhydrin reagent used for amino group detection did not bind to acetylated AZC. Based on these findings and the amino acid sequence similarity with the NAT superfamily, we now believe that AZC is converted to N-acetyl AZC by Mpr1p in the cytoplasm (Fig.  4B), and consequently, N-acetyl AZC no longer replaced Lproline during the biosynthesis of protein.

DISCUSSION
In general, unusual imino acid AZC is incorporated into the nascent polypeptide chain by competition with L-proline in the formation of AZC-tRNA, given rise to by virtue of its structural similarity to L-proline. Stereochemical considerations suggest that AZC in the protein will change the normal ␣-helical structure of a polypeptide to one having an angle about 15°smaller than that by L-proline, leading to an altered tertiary structure of the protein (5). Therefore, AZC shows growth-inhibitory and toxic activities in a wide variety of cells (4 -7). It is noteworthy that AZC is widely distributed in many plants belonging to the Lilaceae family (33,34), which includes Convallaria majalis (lily of the valley) and Polygonatum multiflorum (a Solomon's seal), at high concentrations. Although the physiological role of AZC in Lilaceae plants is not clear, AZC is supposed to protect them from attack by vertebrates and/or microbes. Some plant species that produce AZC were found to have L-proline-activating enzymes that differentiate between L-proline and AZC (34). It is speculated that only L-prolyl-tRNA is then formed so that L-proline alone is incorporated into proteins. In recent studies on the biosynthesis of AZC, labeled L-methionine was incorporated into AZC in lily of the valley plants (35), the result of which was in agreement with the earlier hypothesis by Leete (33) that AZC is formed by the intramolecular displacement of thiomethyladenosine by the ␣-amino group of S-adenosylmethionine. In contrast, little is known about its metabolic pathway in plants.
Mpr1p was found be a novel acetyltransferase with high substrate specificity for AZC. It is unlikely that this enzyme is responsible for AZC degradation, because the yeast cells expressing the MPR1 gene were unable to grow in media containing AZC as the sole nitrogen source (data not shown). Our findings suggested that N-acetylation of AZC is catalyzed by Mpr1p in the cytoplasm, which presumably transfers acetyl groups from acetyl-CoA to ␣-imino groups of AZC. It can be assumed that N-acetyl AZC is no longer recognized by L-prolyl-tRNA synthetase by which both L-proline and AZC bind to ATP, thereby only L-proline is incorporated into proteins, leading to cell growth even in the presence of AZC. The structural deter- FIG. 3. Mpr1p belongs to the NAT superfamily. A, amino acid sequence alignment of a motif A involved in acetyl-CoA binding in the NAT superfamily members with yeast Hat1p (yHAT1), yeast Spt10p (ySPT10), yeast Mak3p (yMAK3), E. coli spermidine NAT (eSNAT), and Streptomyces alboniger puromycin NAT (sPNAT). The numbering above the amino acid sequences refers to Mpr1p. B, Western blot analysis of the extracts of S. cerevisiae CKY2 transformants. Total cellular proteins (ϳ30 g) were subjected to a 15% polyacrylamide gel and were detected by using an anti-Mpr1p polyclonal antibody. C, growth phenotype on AZC-containing medium and AZC acetyltransferase activities of S. cerevisiae CKY2 transformants. Approximately 10 6 cells of each strain and serial dilutions of 10 Ϫ1 to 10 Ϫ3 (from left to right) were spotted onto SD plates containing L-histidine and 300 g/ml AZC. Plates were incubated at 30°C for 3 days. AZC acetyltransferase activity was assayed by using the enzyme solution prepared from each strain grown in SD medium at 30°C. The data shown are means from three independent experiments. The variations in the values are less than 5%. mination of N-acetyl AZC is needed to prove this hypothesis and is currently in progress.
The question arises as to why the S. cerevisiae strain ⌺1278b possesses the MPR genes, although a hypothetical protein with significant sequence similarity to Mpr1p was found in fission yeast (16). We consider the following two possibilities for the physiological role of MPR genes. One is that the S. cerevisiae strain ⌺1278b in nature made its habitat on the Lilaceae or in soils around these plants. To avoid the toxicity of AZC, the strain may have successfully acquired the MPR1 gene involved in AZC resistance from the Lilaceae or some unknown organism during the evolutionary process. It would be of interest to seek the common ancestor of the MPR1 gene to further understanding of the mechanism of AZC resistance. Therefore, this work might contribute to our understanding of the yeast in the natural habitat. Another possibility is that Mpr1p has a bona fide substrate(s) (i.e. not AZC) in the S. cerevisiae strain ⌺1278b. To identify the natural substrate for Mpr1p, we are now analyzing the growth phenotype on media containing various carbon and nitrogen sources, and examining the intracellular contents of the metabolites and the genome-wide transcriptional response to AZC in the wild-type strain and the mpr1 mpr2 disruptant.
Neuwald and Landsman (17) reported that proteins belonging to the NAT superfamily have been grouped by function as histone acetyltransferases, transcriptional regulators, protein acetyltransferases, metabolic enzymes, participants in detoxification and drug resistance, or precise function unknown. In particular, the functional significance of the highly conserved residues in motifs A and B, involved in the binding of acetyl-CoA, have been confirmed by site-directed mutagenesis experiments with many enzymes (29,36,37). To further analyze the function of Mpr1p as a member of the NAT superfamily, we performed an Ala-scan mutagenesis through a six-amino acid sequence (residues 145-150) possibly involved in the binding of acetyl-CoA to Mpr1p. These data, shown in Fig. 4, demonstrated a clear correlation between residues within Mpr1p that are critical for the acetylation of AZC and residues that are absolutely required for AZC resistance. V149A and G150A mutations severely affected AZC acetyltransferase activity, suggesting that Ala substitutions at these positions may interfere with acetyl-CoA binding to Mpr1p. A more detailed sequence comparison among members of the superfamily revealed that branched-chain amino acids such as Ile, Leu, and Val mainly occupy position X, corresponding to Val 149 (17). It is also possible that the steric hindrance, occurring via the G150A mutation that introduces a side chain, leads to reduced AZC acetyltransferase activity in Mpr1p. In contrast, Ala could be replaced with Lys at position 148 in terms of AZC resistance and acetyltransferase activity, so the positive charge is not likely to be indispensable for these functions. Recently, the crystal or solution structures of several proteins belonging to the NAT superfamily were determined in a complex with acetyl-CoA (29,30,37). In addition to a short consensus of residues (Q/R)XXGX(G/A) in motif A, it was also shown that other hydrophobic residues in motifs A and B also contribute to form an acetyl-CoA binding pocket and to bind acetyl-CoA, respectively (38). Given that Mpr1p does not contain any carbohydrates (data not shown) and remains active in the E. coli cells, a crystallization of the recombinant protein overexpressed in E. coli is now attempted to determine the tertiary structure.
In terms of medical applications, naturally occurring AZC was identified as an antimutagen against the spontaneous mutation of bacteria (39) and was tested for its antitumor activity in tissue culture (40). Therefore, when the MPR1 gene is expressed in normal cells, the toxic AZC is expected to be an efficacious agent. The MPR1 gene may also be promising as a new positive selection marker that confers AZC resistance in a wide variety of organisms, because AZC is incorporated, in place of L-proline, into the newly synthesized proteins. Construction and availability of versatile vectors with the MPR1 gene for E. coli, yeast, and animal cells will be published elsewhere.