A Novel Phosphopantetheine:Protein Transferase Activating Yeast Mitochondrial Acyl Carrier Protein*

In Saccharomyces cerevisiae, the low molecular weight acyl carrier protein (ACP) of mitochondrial type II fatty acid synthase (FAS) and the cytoplasmic type I FAS multienzyme contain 4′-phosphopantetheine as a prosthetic group. Sequence alignment studies with the recently isolated phosphopantetheine:protein transferase (PPTase), Ppt1p, from Brevibacterium ammoniagenes revealed the yeast open reading frame, YPL148C, as a potential PPTase gene (25% identical and 43% conserved amino acids). In accordance with this similarity, pantetheinylation of mitochondrial ACP was lost upon disruption of YPL148C. In contrast, biosynthesis of cytoplasmic holo-FAS remained unaffected by this mutation. According to these characteristics, the newly identified gene was designated asPPT2. Similar to ACP null mutants, cellular lipoic acid synthesis and, hence, respiration were abolished in PPT2deletants. ACP pantetheinylation, lipoic acid synthesis, and respiratory competence were restored upon transformation ofPPT2 mutants with cloned PPT2 DNA. In vitro, holo-ACP synthesis was achieved by incubating apo-ACP with coenzyme A in the presence of purified Ppt2p. The homologous yeast enzyme could be replaced, in this assay, by the ACP synthase (EC2.7.8.7) of Escherichia coli but not by the type I FAS-specific PPTase of B. ammoniagenes, Ppt1p. These results conform with the inability of Ppt2p to activate the cytoplasmic type I FAS complex of yeast.

In Saccharomyces cerevisiae, the low molecular weight acyl carrier protein (ACP) of mitochondrial type II fatty acid synthase (FAS) and the cytoplasmic type I FAS multienzyme contain 4-phosphopantetheine as a prosthetic group. Sequence alignment studies with the recently isolated phosphopantetheine:protein transferase (PPTase), Ppt1p, from Brevibacterium ammoniagenes revealed the yeast open reading frame, YPL148C, as a potential PPTase gene (25% identical and 43% conserved amino acids). In accordance with this similarity, pantetheinylation of mitochondrial ACP was lost upon disruption of YPL148C. In contrast, biosynthesis of cytoplasmic holo-FAS remained unaffected by this mutation. According to these characteristics, the newly identified gene was designated as PPT2. Similar to ACP null mutants, cellular lipoic acid synthesis and, hence, respiration were abolished in PPT2 deletants. ACP pantetheinylation, lipoic acid synthesis, and respiratory competence were restored upon transformation of PPT2 mutants with cloned PPT2 DNA. In vitro, holo-ACP synthesis was achieved by incubating apo-ACP with coenzyme A in the presence of purified Ppt2p. The homologous yeast enzyme could be replaced, in this assay, by the ACP synthase (EC 2.7.8.7) of Escherichia coli but not by the type I FAS-specific PPTase of B. ammoniagenes, Ppt1p. These results conform with the inability of Ppt2p to activate the cytoplasmic type I FAS complex of yeast.
4Ј-Phosphopantetheine serves as a prosthetic group in a variety of enzyme systems such as fatty acid synthases (1,2), most polyketide synthases (3) and several non-ribosomal polypeptide synthetases (4). Similar to the chemically related soluble cofactor, coenzyme A, protein-bound phosphopantetheine fulfills in these enzymes a dual function, i.e. activation of acyl groups by thioester linkage to the terminal sulfhydryl, and acting as a flexible arm allowing translocation of intermediates between different catalytic sites. Apart from complex polycondensation systems, it is suggested that, in yeast, enzyme-bound phosphopantetheine is also involved in a single metabolic reaction, the reduction of ␣-amino-adipate to the respective semialdehyde (5).
According to their molecular structures, two classes of phosphopantetheine-containing proteins may be discriminated. One class is represented by a family of low molecular weight acyl carrier proteins (ACP) 1 functioning as structurally independent components of non-aggregated multienzyme systems and exhibiting no catalytic activity, by themselves (type II system). The other class contains phosphopantetheine bound to an ACP-like acyl-or peptidyl-carrier domain, which is part of a multifunctional polypeptide chain (type I system).
Using the type II ACP of Escherichia coli, it had first been demonstrated by Vagelos and co-workers (6) that 4Ј-phosphopantetheine is transferred from coenzyme A to the hydroxyl group of a specific serine on apo-ACP. This reaction is catalyzed by the enzyme, phosphopantetheine:protein transferase (PPTase). To date, five different bacterial PPTases have been purified and characterized in some detail, i.e. EntD (5, 7), Sfp (5), 0195 (5), ACPS (8), and Ppt1p (9). The substrates of EntD and Sfp are, respectively, the enterobactin synthetase of E. coli and the surfactin synthetase of Bacillus subtilis, while Ppt1p activates the two type I FAS enzymes of Brevibacterium ammoniagenes and ACPS the corresponding type II FAS of E. coli.
No eucaryotic PPTase has been biochemically characterized, to date. Nevertheless, multiple PPTases are likely to occur in most eucaryotes, as is suggested by the occasional existence of several different phosphopantetheinylated proteins within the same cell. For instance, apart from the type I fatty acid synthase present in the cytoplasm of all non-plant eucaryotes, a FAS-like enzyme system of mitochondrial origin has been suggested to exist in fungi (10), plants (10), and mammals (11). This system exhibits a type II molecular structure and contains a low molecular weight ACP being comparable, in its size and sequence, to the respective E. coli protein. In the fungi Saccharomyces cerevisiae and Neurospora crassa, mutational loss of mitochondrial ACP had no effect on bulk cellular fatty acid synthesis (12)(13)(14). Instead, yeast ACP null mutants exhibit a respiratory-defective phenotype when grown on glycerol as a carbon source (12,14). This phenotype is strictly connected to the loss of endogenous lipoic acid synthesis and cannot be compensated by exogenous supply of this cofactor (14). Thus, mitochondrial FAS appears to be involved in the biosynthesis of the lipoic acid precursor, octanoic acid (14,15).
As both the mitochondrial and the cytoplasmic FAS system require protein-bound 4Ј-phosphopantetheine as a prosthetic group, it remained to be shown whether the respective apoproteins were activated by the same enzyme or, independently of each other, by two different PPTases. As will be reported in this study, inspection of the yeast genomic DNA sequence disclosed the open reading frame, YPL148C, as a potential PPTase coding sequence. It was demonstrated by both mutant analysis and in vitro activation of purified mitochondrial apo-ACP that the respective gene product represents in fact a novel phosphopantetheine:protein transferase, which specifically activates the low molecular weight mitochondrial ACP but not the type I FAS complex present in the yeast cytoplasm.

EXPERIMENTAL PROCEDURES
Bacterial Strains, Plasmids, and General DNA Manipulation Techniques-Bacterial strains and plasmids used in this study are listed in Table I. Genomic DNA from S. cerevisiae and E. coli was isolated as described by Hoffman et al. (20). PCR was performed using Vent DNA polymerase (New England Biolabs) according to the recommendations of the manufacturer. For gene replacement the PPT2 gene was amplified by PCR together with 430-bp N-terminal and 100-bp C-terminal flanking DNA. The forward primer GACGTAGAATTCGAGCTGT-TATATACGCAT and the reverse primer GACGTAGAATTCGCGT-TCTAAGACTTCCAG created new EcoRI restriction sites (underlined), which were used to insert the 1060-bp PCR product into the E. coli vector pUC19 (Boehringer Mannheim). Subsequently, the 173-bp NheI/ NsiI fragment of PPT2 was eliminated and replaced by the kanMX4 marker (21) (Fig. 1). The BamHI/PvuII fragment comprising the ⌬ppt2::kanMX4 construct was used to replace the genomic yeast PPT2 gene by integrative transformation of the Saccharomyces cerevisiae strain C13-ABY.S86 according to the method of Soni et al. (22).
Overproduction and Purification of the Phosphopantetheine:Protein Transferases Ppt2p, ACPS, and Ppt1p-The Ppt2p coding sequence was amplified by PCR using genomic DNA isolated from the S. cerevisiae strain, JS91.15-23. The forward primer TCTACATTGCATGCCTCC-AGTGATGAG and the reverse primer GCACTGCGGATCCCTCTCTT-TCTACCAAGTTTG introduced a SphI restriction site at the start codon and a BamHI site substituting the stop codon (underlined). The SphI/ BamHI-digested PCR product was cloned into the expression vector pQE70 (Qiagen), and the resulting plasmid was transformed into E. coli DH5␣ (pRP4). The Ppt2p protein thus produced contained a hexahistidine sequence at its C terminus. For protein purification, PPT2-transformed E. coli cells were pre-grown on Luria Broth solid medium containing 50 mg/liter kanamycin and 100 mg/liter ampicillin and subsequently inoculated into 500 ml of liquid medium (pH 7.5) containing, per liter, 10 g of NaCl, 15 g of yeast extract, 30 g of tryptone, 100 mg/liter ampicillin, and 1 mM isopropyl-1-thio-␤-D-galactopyranoside. The cells were grown at 30°C until late log-phase, collected by centrifugation, and stored at Ϫ70°C. The cell pellet was suspended in 40 ml of 50 mM sodium phosphate, pH 7.5, 300 mM NaCl, 30 mM imidazole, and 1 mg/ml phenylmethylsulfonyl fluoride (buffer A). Cell lysis was performed by sonication with a Branson B-12 sonifier, and insoluble components were removed by centrifugation (100,000 ϫ g, 20 min, 4°C). Subsequently, 2 ml of Ni 2ϩ -agarose (Qiagen) equilibrated with 50 mM sodium phosphate, pH 7.5, containing 300 mM NaCl was added. For adsorption of the hexahistidine-tagged protein, the suspension was stirred on ice for 1 h. The Ni 2ϩ -agarose was then washed twice with 40 ml of buffer A (without phenylmethylsulfonyl fluoride) and twice with 40 ml of 50 mM sodium phosphate, pH 7.5, 300 mM NaCl, 30 mM imidazole, and 15% glycerol (buffer B). Subsequently the agarose was loaded onto a 5-ml column. The protein was eluted by a stepwise gradient from 0.1 to 0.5 M imidazole dissolved in buffer B. Fractions (1 ml) were collected, and aliquots were analyzed on a 15% SDS-polyacrylamide gel. The Ppt2p-containing fractions were stored in buffer B at Ϫ70°C.
The ACP synthase (ACPS) gene was PCR-amplified from E. coli M15 DNA using the forward primer ATCAGTTAGGCATGCCAATATTAG-GTTTAGGCACG and the reverse primer TGCGTGAACAGATCTAC-TTTCAATAATTACCGTGGC. The PCR primers created novel SphI and BglII restriction sites (underlined). The ACPS reading frame was cloned into the expression vector pQE70 (Qiagen), and the resulting plasmid was transformed into E. coli M15 (pRP4) (Qiagen). For ACPS purification, a 50-ml overnight culture of E. coli M15 transformants grown in Luria Broth with 50 mg/liter kanamycin and 100 mg/liter ampicillin at 30°C was harvested, suspended in 200 ml of fresh medium without kanamycin, and induced with 4 mM isopropyl-1-thio-␤-D-galactopyranoside. The induced cells were grown for 6 h at 30°C. Purification of ACP synthase was performed as described for Ppt2p.
The construct used for Ppt1p expression has recently been described (9). Growth and purification conditions were as described above for the E. coli ACP synthase.
Purification of Yeast Mitochondrial ACP upon Overexpression in S. cerevisiae or E. coli-For overexpression of mitochondrial ACP in S. cerevisiae the complete ACP1 coding sequence was PCR-amplified from yeast DNA of the strain JS91.15-23. The forward primer ATCGG-GATCCATGTTTAGATCCGTTTGCCG created a BamHI restriction site (underlined) in front of the ATG-start codon. The reverse primer ATCGCTCGAGTTAGTGATGGTGATGGTGATGGTTTGCGTCGGGA-TTGGAAGCGATATAATCGACCG introduced six histidine codons as well as a XhoI restriction site (underlined) downstream of the translational stop codon. The resulting PCR product was inserted into the E. coli/S. cerevisiae shuttle vector p425Met25 (23) and transformed into the S. cerevisiae strain C13-ABY.S86. Transformants were selected according to their leucine prototrophy on leucine-free SCD medium containing 1 g/liter methionine. For ACP purification cells were grown at 30°C in 1 liter of SCD medium containing, per liter, 40 g of glucose, 2 g of yeast nitrogen base (Life Technologies, Inc.), 5 g of ammonium sulfate, and 1 g of methionine. The cells were harvested at late logphase, inoculated into 1 liter of fresh medium without methionine, and grown for additional 16 h. The cell pellet obtained after centrifugation was frozen at Ϫ70°C and subsequently suspended in buffer A. Cells were disrupted with glass beads at 4°C, and any mitochondria still present in the resulting homogenate were lysed by sonication. The extract was fractionated by addition of ammonium sulfate to 50% saturation. After centrifugation for 20 min at 10,000 ϫ g, the ACP-containing supernatant was titrated with 11 M HCl to pH 1.0. The precipitate thereby obtained was collected by centrifugation and subsequently re-dissolved in 150 ml of buffer A by stirring the suspension for 16 h at 4°C. After centrifugation (100,000 ϫ g, 20 min, 4°C), the clear protein solution was combined with 2 ml of Ni 2ϩ -agarose. Subsequent purification steps were performed as described above for Ppt2p isolation.
For expression in E. coli, the yeast mitochondrial ACP gene was PCR-amplified without its 36-amino acid N-terminal leader sequence. The forward primer GCACTGAACGCATGCCTGCAAACTTGAG-CAAAGAT introduced an ATG start codon together with a SphI restriction site (underlined) at codon 37 of the ACP1 coding sequence, while the reverse primer GCACTGAACGGATCCGTTTGCGTCGGGATTGGA created a BamHI site (underlined) substituting the TAA stop codon. The SphI/BamHI-digested PCR product was cloned into the expression vector pQE70 (Qiagen) and transformed into E. coli M15 (pRP4) (Qiagen). Expression and purification of the recombinant ACP were as described above for the E. coli ACP synthase.
In Vitro Phosphopantetheinylation of Mitochondrial Apo-ACP by Ppt2p or ACPS-The transfer of [ 3 H]phosphopantetheine from [ 3 H]coenzyme A to mitochondrial apo-ACP was monitored by liquid scintillation counting. Apo-ACP preparations were obtained by expressing the yeast ACP1 gene either in the heterologous host, E. coli, or in S. cerevisiae ⌬ppt2 disruptants. Purified apo-ACP preparations were dialyzed against 50 mM sodium phosphate buffer, pH 7.5, and subsequently used for in vitro activation. In vitro phosphopantetheine transfer was carried out at 30°C in a total volume of 1 ml containing 200 g of apo-ACP, 10 g of PPTase dissolved in buffer B, 50 M coenzyme A, 0.5 ϫ 10 6 cpm [ 3 H]coenzyme A (3.5 Ci/mmol) and 750 l of 20 mM Tris-HCl, pH 7.5, containing 25 mM MgCl 2 and 3 mM dithiothreitol. After 1 h of incubation, a small amount of Ni 2ϩ -agarose was added to the reaction mixture. This suspension was incubated for 10 min under gentle shaking to allow ACP adsorption. Ni 2ϩ -agarose together with the ACP bound to it was collected by centrifugation and washed five times with buffer A and, subsequently, five times with buffer B. Finally, the radioactivity adsorbed to the Ni 2ϩ -agarose matrix was quantified by liquid scintillation counting.
Lipoic Acid Determination and Protein Analytical Techniques-Lipoic acid analysis was performed as described recently (14). SDS-PAGE was performed according to the method of Laemmli (24). Mitochondrial ACP was analyzed, after reduction with 10 mM dithiothreitol at pH 7.2, by native PAGE as described by Bollag et al. (25). N-terminal protein sequence determinations were performed by Toplab (Mü nchen, Germany). Amino acid composition of ACP was determined according to Hannappel et al. (26). Cysteamine was determined as taurine after performic acid oxidation of ACP (27).

RESULTS
Identification and Mutational Inactivation of the Yeast Phosphopantetheine:Protein Transferase Gene, PPT2-BLAST P searches with the yeast genome data base and the B. ammoniagenes PPTase, Ppt1p (accession number Y15081), as a reference revealed similarities to the C-terminal region of FAS subunit ␣ (28% identical and 42% conserved positions) and to the hypothetical product of the reading frame, YPL148C (25% identical and 42% conserved position). The PPTase-like domain at the C terminus of one of the two fungal FAS subunits had already been noticed by Lambalot et al. (5) and by Stuible et al. (9) when probing with the bacterial PPTase sequences, ACPS and Ppt1p, respectively. As was suggested by these authors, activation of fungal type I fatty acid synthases is possibly catalyzed by this PPTase domain and, thus, represents a capacity of the FAS enzyme itself. On the other hand, activation of mitochondrial ACP is unlikely to depend on this FAS2encoded PPTase-like domain, as no ACP-defective phenotype is observed with FAS2 deletants (data not shown). Thus, YPL148C was considered as a potential candidate for the gene controlling phosphopantetheinylation of yeast mitochondrial ACP. The YPL148 reading frame encodes a hypothetical protein of 177 amino acids in length and a calculated molecular mass of 20.319 kDa. Its overall sequence similarity to Ppt1p from B. ammoniagenes and to ACPS from E. coli is demonstrated in Fig. 2.
In order to characterize the biochemical function of YPL148C, part of its sequence was replaced by the kanMX4 gene, providing resistance against the antibiotic, G418 (cf. Fig.  1). Correct chromosomal integration of the kanMX4 marker was verified by Southern blotting and by PCR analysis (data not shown). YPL148C null mutants were viable on glucose as a carbon sources, but failed to grow on non-fermentable substrates such as glycerol. As these mutants grew on glucosecontaining medium without fatty acid supplementation, activation of the cytoplasmic FAS-complex is clearly not affected by YPL148C inactivation. Furthermore, glycerol utilization of YPL148C mutants was not supported by supplementation with the complex mixture of fatty acids present in an alkaline butter hydrolysate nor by any of the nutrients contained in commercial batches of yeast extract and peptone. The growth characteristics of YPL148C disruptants were therefore identical to those of yeast ACP1 deletion mutants lacking mitochondrial ACP (12). Similar to ACP1 null mutants, YPL148C disruptants exhibit the typical characteristics of Rho-negative yeast mutants, i.e. their respiratory defect is not complemented by a Rho-negative reference strain such as MYY110. All of these findings support the idea that YPL148C encodes the phosphopantetheine transferase activating the type II ACP of yeast mitochondria. Therefore, the so far unassigned reading frame, YPL148C, was designated as PPT2.
Isolation and Characterization of Apo-and Holo-ACP from Wild Type and PPT2 Mutant Cells-The involvement of Ppt2p in phosphopantetheinylation of mitochondrial ACP was studied upon overexpression of recombinant mitochondrial ACP in appropriate yeast strains. Among several strains tested, only the protease-negative mutant C13-ABY.S86 as well as the ⌬ppt2 disruptant SC 1383 and the Rho-negative mutant SC 1382, which both are derived from C13-ABY.S86, allowed overproduction of ACP from pSM80 under the control of the inducible MET25 promoter (23). The recombinant ACP was isolated, as C-terminal hexahistidine fusion, by nickel chelate chromatography.
The ACP preparations thus obtained from the three yeast strains, C13-ABY.S86, SC 1383, and the PPT2-positive, Rhonegative mutant SC 1382, were analyzed by N-terminal sequencing and by native polyacrylamide gel electrophoresis. It was found that independent on whether the ACP was isolated from respiratory competent or respiratory-defective cells, always the leader-free protein was obtained starting with Ser-37 as the N-terminal amino acid (data not shown). Thus, neither ACP overproduction nor mitochondrial inactivation precluded efficient processing and, hence, organellar import of ACP. Native gel electrophoresis of ACP derived from C13-ABY.S86 and SC 1382 transformants revealed the presence of two ACP variants of different electrophoretic mobilities (Fig. 3). The presence of pre-ACP or of ACP dimers in these preparations is excluded according to the N-terminal sequencing data and as a result of the reduction of ACP with dithiothreitol, respectively. It is known that the electrophoretic mobility of E. coli holo-ACP is higher than that of the respective apo-form (28). Correspondingly, only the retarded ACP isoform was isolated from ⌬ppt2 disruptant cells (Fig. 3). This finding strongly supported the idea that both apo-and holo-ACP are included in the preparations derived from PPT2-positive cells no matter whether these cells were respiratory-competent or respiratory-defective while exclusively apo-ACP but no holo-ACP is present in the PPT2 mutant. To verify this interpretation, ACP preparations isolated from C13-ABY.S86 and from the PPT2 null mutant, SC 1383, were tested for the presence of cysteamine. After performic acid oxidation, essentially no taurine was found in the hydrolysate of the SC 1383-derived ACP preparation. On the other hand, this cysteamine-derivative was clearly detectable in the preparation isolated from C13-ABY.S86 cells (data not shown).Together with the electrophoretic data shown in Fig. 3, these results suggest that about 50% of the ACP overproduced in wild type cells, and about 30% of that isolated from the Rho-negative mutant, SC 1382, were represented by holo-ACP while there was no ACP pantetheinylation to be detected in the PPT2-defective mutant, SC 1383. As is also evident from Fig. 3, holo-ACP synthesis is restituted to wild type level upon transformation of the PPT2 null mutant with the wild type PPT2 gene on a single-copy plasmid.
In Vitro Pantetheinylation of Yeast Mitochondrial ACP-Conclusive proof for Ppt2p acting as phospopantetheine:proteine transferase was derived from in vitro activation studies. For these studies, mitochondrial apo-ACP, the putative transferase, Ppt2p, and the bacterial PPTases, ACPS and Ppt1p, were purified by Ni 2ϩ -agarose chromatography upon overproduction of their hexahistidine-tagged derivatives by an inducible E. coli expression system. For efficient production in this host, the 36-amino acid pre-sequence of mitochondrial ACP had to be absent probably because of its negative effect on native protein conformation. Although endogenous ACP is known to strongly inhibit bacterial growth when overexpressed in E. coli (28), no such effect was observed upon heterologous expression of yeast mitochondrial ACP. This finding conforms with the observation of Guerra et al. (29) that overproduction of heterologous bacterial ACPs in E. coli is not toxic, either. To obtain sufficient amounts of recombinant Ppt2p protein, the slow growing E. coli strain, DH5␣, had to be used together with reduced cultivation temperature. In contrast to the putative yeast transferase, the bacterial PPTases, ACPS and Ppt1p, were produced at high rates although induction of ACPS expression clearly inhibited growth of the E. coli transformants. The homogeneity of yeast mitochondrial ACP and of the enzyme preparations finally obtained was verified by SDS-PAGE (Fig. 4). The unusual electrophoretic migration characteristics of mitochondrial ACP indicating an apparent molecular mass of 21 kDa rather than the calculated value of 9 kDa compare to that of endogenous E. coli ACP (28,30). In addition, the identity of the mitochondrial ACP and Ppt2p preparations isolated from E. coli were verified by N-terminal sequence analysis.
According to its migration characteristics in native PAGE, essentially all of the mitochondrial ACP isolated from E. coli was present as apo-ACP (data not shown). This agrees with the finding that even endogenous ACP is predominantly in its apo-form when overexpressed in E. coli (8,28). Thus, the isolated mitochondrial ACP appeared as a suitable substrate for in vitro phosphopantetheinylation studies using the purified PPTases, Ppt2p, ACPS, and Ppt1p. Upon incubation with 3 Hlabeled coenzyme A, formation of [ 3 H]phosphopantetheinylated ACP was monitored by re-adsorption of the His-tagged ACP to Ni 2ϩ -agarose and subsequent determination of the specifically adsorbed radioactivity. As is evident from Table II, both yeast Ppt2p and E. coli ACPS efficiently incorporated the radioactive label into mitochondrial ACP. In contrast, essentially no PPTase activity was observed with Ppt1p, which activates, in vivo and in vitro, the type I fatty acid synthases, FAS-A and FAS-B, of B. ammoniagenes (9).
The PPTase function of Ppt2p was further confirmed by incubating mitochondrial apo-ACP isolated from yeast PPT2 null mutants with the recombinant enzyme in the presence of excess unlabeled coenzyme A. Comparative gel electrophoresis of the reaction products demonstrated that, under the conditions applied, apo-ACP was in fact quantitatively converted into holo-ACP (Fig. 3).
Lipoic Acid Deficiency of PPT2 Null Mutants-In yeast as well as in plant cells, mitochondrial ACP has been found to be involved in lipoic acid biosynthesis (14,15). As a result of this function, and of the inability of yeast to use externally supplied lipoate, ACP null mutants are essentially free of this cofactor. The same biochemical defect was expected upon PPT2 inactivation as biosynthesis of functionally active holo-ACP is prevented by this mutation. The hypothesis was verified by analyzing the lipoic acid content of various ⌬ppt2::kanMX4 disruptants. If detectable at all, cellular lipoic acid concentrations were, in all PPT2 mutants investigated, comparable to those of ACP1 null mutants (Table III). Thus, inactivation of both biosynthesis and posttranslational phosphopantetheinylation of apo-ACP elicit the same biochemical defect. Transformation of PPT2 mutants with wild type PPT2 DNA restored normal holo-ACP formation (Fig. 3) and, consequently, lipoic acid synthesis (Table III) while the respiratory defect was not complemented. This persistence of respiratory deficiency was to be expected as the loss of mitochondrial DNA cannot be complemented by transformation with PPT2. Reintroduction of intact mitochondrial DNA into the mutant was achieved by back-crossing the PPT2 transformed ⌬ppt2 mutant with a mitochondrially competent, Rho-positive yeast strain and by subsequent sporulation of the resulting diploid. Tetrad analysis of the meiotic segregants thereby obtained revealed that all spores were respiratory-competent, no matter whether the genomic PPT2 gene was intact or not (data not shown). Similar characteristics have previously been reported for other lipoatedefective yeast strains such as ACP1 and LIP5 mutants (14,31). They also tend to lose their mitochondrial DNA and, thereby, undergo an irreversible secondary mutation. The complementation characteristics of PPT2 mutants clearly demonstrate that their respiratory defect is the consequence rather than the cause of cellular lipoic acid deficiency. DISCUSSION The present paper describes the identification and characterization of the phosphopantetheine:protein transferase, Ppt2p, that activates mitochondrial ACP in Saccharomyces cerevisiae. The PPT2 reading frame was identified according to its sequence similarity to the type I FAS-specific PPTase, Ppt1p from B. ammoniagenes (9). Surprisingly, a similar BLAST search with E. coli ACP synthase (8) as a reference failed to uncover PPT2 as a homologue, although the sub-strates of ACPS and Ppt2p are more alike than those of Ppt2p and Ppt1p. Both ACPS and Ppt2p activate type II acyl carrier proteins while Ppt1p pantetheinylates the ACP-like domain of a type I fatty acid synthase. The different extent of ACPS, Ppt1p, and Ppt2p sequence similarity as was suggested from the data base searching is documented and underlined by the alignment of the three protein sequences as shown in Fig. 2. The similarity between ACPS and Ppt2p is particularly reduced by three extended insertions present in Ppt2p when compared with ACPS. Apart from their restricted overall sequence similarity, several sequence elements are nevertheless conserved, and are spread over the entire lengths of the three proteins. In particular, the consensus motifs (V/I/L)GXD and (W/F)XXKE(A/S/C)XXK at the Ppt2p-positions 19 -22 and 85-93, respectively, are strictly conserved among all known PPTases (5).
In addition to cytoplasmic FAS and mitochondrial ACP, a third yeast protein, aminoadipate-semialdehyde dehydrogenase, is postulated to contain phosphopantetheine as a prosthetic group (5). The three proteins appear to be phosphopantetheinylated by specific PPTases, which do not interact with other than their cognate substrates. This is evident from the finding that null mutants of the respective PPTase-encoding genes, PPT2, FAS2, and LYS5, are specifically defective in either one of the three functions, respiration, fatty acid synthesis (32), or lysine biosynthesis (33) with the remaining two functions being unimpaired. In particular, pantetheinylation of FAS subunit ␣ by the PPT2-or LYS5-encoded enzymes is excluded as pantetheine-free cytosolic apo-FAS is produced by a specific class of FAS2 mutants despite the presence of wild type alleles of PPT2 and LYS5 (9,34). These FAS2 mutants harbor distinct missense mutations within the PPTase-like domain of FAS2 (34). In the present work as well as in a recent study of Stuible et al. (9), the differential activities of the PPTases, ACPS, Ppt1p, and Ppt2p were demonstrated with the type I and type II fatty acid synthases of B. ammoniagenes and yeast mitochondria, respectively. Apart from its inactivity with type I fatty acid synthases, E. coli ACPS does not act on any of the polypeptide synthetases tested (5,7). Correspondingly, pantetheinylation of gramicidin S synthetase in E. coli depends on coexpression of the orthologous PPTase, gsp (35). The lack of pantetheinylation observed after heterologous expression of the gramicidin synthetase modules also indicates a comparably high substrate specificity of the E. coli PPTases O195 (5) and EntD (5,7). In contrast, the surfactin synthetase activating PPTase from B. subtilis, Sfp, exhibits a remarkably broad substrate specificity recognizing the apo-forms of many peptidyl, aryl, and acyl carrier protein domains (36). Regarding the phosphopantetheinylated yeast proteins, FAS subunit ␣, mitochondrial ACP, and aminoadipate-semialdehyde dehydrogenase, the need for specific partner transferases is most easily explained by the extensive sequence variations of the respec-  III Lipoic acid levels in PPT2-defective S. cerevisiae mutants Cell growth and lipoic acid determination were as described previously (14). The complete genotype of the strains investigated is indicated in Table I. SC 1383/pSM75 and SC 1383/pSM73 designates transformants of SC 1383 with the single-copy centromer plasmid pSM75 and with the multicopy plasmid pSM73, respectively. tive apo-proteins. Apart from three positions around the phosphopantetheinylated serine, (V/L)GXXSXXXXXX(V/L), no consensus elements are found which are common to all three proteins. The considerable sequence variations observed between the putative yeast PPTases correspond to the dissimilarity of the respective apo-proteins.
When compared with E. coli ACP, yeast mitochondrial ACP contains an N-terminal leader sequence of 36 amino acids in length. This sequence exhibits the characteristics of a typical mitochondrial import signal (37) and is considered to function as a transit peptide directing ACP into the organelle. Upon isolation from yeast, exclusively mature ACP starting at Ser-37 is isolated, no matter whether the protein was overexpressed in wild type or in respiratory defective cells. Thus, mitochondrial import and concomitant processing of ACP are obviously unimpaired even with the functionally defective pro-mitochondria of Rho-negative mutants. This finding agrees with the notion that a basal set of matrix enzymes continues to be imported from the cytoplasm into DNA-free pro-mitochondria (38). As the apo-ACP produced in PPT2 null mutants is proteolytically processed, too, import of the ACP pre-protein into the organelle is suggested to be independent of its modification to the holoform. Fernandez and Lamppa (39) came to similar conclusions regarding the import of holo-and apo-ACP into plant chloroplasts. Additional information regarding the intracellular site of ACP pantetheinylation in yeast may be expected from studies on the localization of the respective PPTase, Ppt2p. Experimental evidence on the possible mitochondrial import of Ppt2p is still missing. However, the N-terminal sequence of Ppt2p extending 13 amino acids beyond the N terminus of the respective E. coli enzyme, ACPS, is compatible with the formation of an amphipathic "helical wheel," which represents a characteristic feature of mitochondrial matrix targeting signals (37). The chemical structure of this sequence, however, suggests it to be a relatively weak signal (40), as should be expected for a weakly expressed protein. The corresponding E. coli enzyme, ACP synthase, is present in extremely low cellular concentrations as well (8). Unprocessed Ppt2p as is isolated upon heterologous expression in E. coli is enzymatically active. Thus, N-terminal processing is obviously not required for PPTase activity. Final proof for the intracellular location and, hence, for the possible posttranslational processing of Ppt2p must await further experimentation.