Characterization of a New Type of Phosphopantetheinyl Transferase for Fatty Acid and Siderophore Synthesis in Pseudomonas aeruginosa

Phosphopantetheinyl-dependent carrier proteins are part of fatty-acid synthases (primary metabolism), polyketide synthases, and non-ribosomal peptide synthetases (secondary metabolism). For these proteins to become functionally active, they need to be primed with the 4 (cid:1) -phosphopantetheine moiety of coenzyme A by a dedicated phosphopantetheine transferase (PPTase). Most organisms that employ more than one phospho-pantetheinyl-dependent pathway also have more than one PPTase. Typically, one of these PPTases is optimized for the modification of carrier proteins of primary metabolism and rejects those of secondary metabolism (AcpS-type PPTases), whereas the other, Sfp-type PPTase, efficiently modifies carrier proteins involved in secondary metabolism. We present here a new type of PPTase, the carrier protein synthase of Pseudomonas aeruginosa , an organism that harbors merely one PPTase, namely PcpS. Gene deletion experiments amplified PA1165 recombinant C-terminal pQE70-PA2966— PA2966 the acyl (ACP) by with oligonucleotides 5 (cid:1) and 5 (cid:1) -ATAAGATCTTT-GCTGGTGAGCAACGATG-3 chromosomal aeruginosa PA01. digested ligated into Sph II plasmid PA2966 a C-terminal of pQE70-pchEArCP— fragment the domain from the (pchE1) pyochelin synthetase E amplified by PCR with oligonucleotides 5 (cid:1) and 5 (cid:1) -ATAAGATCTGTCCGCGCAGGCCAG- (cid:1) from chromosomal of aeruginosa PA01. fragment was digested with Sph I and Bgl II and ligated into the Sph I and Bgl II sites of pQE70. The plasmid pQE70- pchEArCP encodes the excised recombinant (21) with minor modifications. mutated or the wild type PA1165 genes were identified by PCR. Oligonucleotides 5 (cid:1) -ATA GAATTC GGTTGAGCCCGATCTTGC-3 (cid:1) and 5 (cid:1) -ATA AAGCTT CTTTGCCTGGCCGAATGG-3 (cid:1) were used to amplify an (cid:6) 3-kb fragment in both cases. Digestion of this fragment with Age I yields two 1.5-kb fragments in case of the wild type gene, whereas digestion with Spe I yields a 1-kb and a 2-kb fragment confirming presence of the gentamycin cassette in this fragment.

4Ј-Phosphopantetheine (Ppant) 1 -dependent carrier proteins (CP) are the central entity in fatty-acid synthases (FAS), polyketide synthases (PKS), and non-ribosomal peptide synthetases (NRPSs) (1,2). The superfamily of CP includes the acidic acyl and aryl carrier proteins and the neutral peptidyl carrier proteins (PCPs). They can be part of a larger polypeptide chain or exist as distinct proteins but always fulfill the same job; during the multistep assembly of the product, the reaction intermediates of the growing acyl or polypeptide chain remain covalently tethered to the Ppant cofactor moiety of these proteins. This moiety is about 20 Å in length and enables the bound intermediates to move between the reaction centers of multifunctional proteins. The thioester linkage that is used to bind the intermediates and final product is energy-rich, which facilitates cleavage after the final step of the assembly. After ribosomal synthesis, however, the carrier protein exists only in the inactive apo form. The Ppant moiety is post-translationally transferred from coenzyme A to a conserved serine residue of the CP in a Mg 2ϩ -dependent reaction by a dedicated phosphopantetheine transferase (PPTase), thus converting it to the active holoCP (3) (Fig. 1).
Many organisms utilize more than one Ppant-dependent pathway. For instance, aside from fatty acids, Bacillus subtilis and Escherichia coli produce the cyclic lipopeptide surfactin and the catecholic siderophore enterobactin, respectively (4,5). Both products that are secondary metabolites of these organisms were name-giving when the associated PPTases, Sfp and EntD, were discovered. The conversion of ACPs of fatty acid synthesis (primary metabolism) is catalyzed by a second PPTase, namely the acyl carrier protein synthase (AcpS), that does not recognize the CPs of the NRPS systems (1,6). It has been shown that EntD does not cross-interact with primary metabolism, whereas B. subtilis can sustain fatty acid synthesis even if AcpS is not present (7).
The substrate spectrum of the PPTases has, apart from the difference in size, lead to a classification into three groups. The name-giving prototype of the first group is AcpS of E. coli. PPTases of this group have a narrow substrate specificity, are about 120 aa in size, and are shown to act as homotrimers (8,9). AcpS of E. coli modifies only ACPs of FAS and type II PKS (10). Recently, these results were further supported by the characterization of other AcpS-type PPTase, especially AcpS of B. subtilis (7). PPTases of the second, Sfp group, are about twice the size of AcpS, have an extraordinarily broad substrate spectrum, and are active as monomers. Sfp has proven to recognize every CP tested so far, including PCPs of NRPSs and ACPs of FASs and PKSs. PPTases of the third group act as integrated domains on their cognate ACP of type I FAS, as is the case in Saccharomyces cerevisiae, for example (11) (Fig. 2).
Because Sfp has been shown to accept ACPs of FAS as substrate, the question remained whether the presence of a second AcpS-type PPTase would be essential for the survival of B. subtilis. Although many organisms contain both a PPTase of the AcpS type and one of the Sfp-type, some of the recently sequenced organisms seem to have lost its AcpS in the course of evolution (Table I). Mootz et al. (7) simulated this loss by the deletion of AcpS in an sfp ϩ strain of B. subtilis. The phenotype of this mutant was that of the wild type, but further in vitro characterization of Sfp revealed that the catalytic efficiency was low with the ACP of primary metabolism. This PPTase is evidently optimized for the modification of CPs of secondary metabolism.
Pseudomonas aeruginosa, a Gram-negative human pathogen, is a major origin of infection in patients with burns but is also known to cause pneumonia and ocular diseases (12). Pseudomonas spp. secret toxic compounds such as elastase, proteases, or cyanide and are very resistant to treatment with antimicrobial agents (13,14). Two siderophores, namely pyoverdin (pvd) and pyochelin (pch), which are produced during iron-limiting growth conditions, are associated with the high virulence of P. aeruginosa (15)(16)(17). Disruption of the synthesis of these two non-ribosomally produced peptides has proven to reduce significantly the virulence of this organism (18). This would make the PPTase responsible for the modification of the pvd-and pch-NRPSs an excellent target for the development of new antibiotics.
Our blast searches using E. coli EntD and B. subtilis AcpS as probes revealed that P. aeruginosa harbors merely one putative ORF encoding a PPTase (PA1165, GenBank TM accession number AAG04554) of 242 aa (27 kDa) with a pI of 6.77. According to these characteristics this enzyme ought to belong to the Sfp-type PPTases but exhibits only minor similarity to Sfp (13.9%) (Fig. 2). In contrast to Sfp, this PPTase is not clustered with any NRPS. We report here on the genetic and biochemical characterization of the PPTase of P. aeruginosa in vivo and in vitro with its natural substrates. The PPTase was renamed according to its function as a P. aeruginosa carrier protein synthase (PcpS). As we will show, this enzyme is of special interest because it represents a new type of PPTase, essential for the modification of both the CPs of primary and secondary metabolism. Unlike Sfp, which exhibits poor catalytic efficiency with the ACP of FAS and high efficiency with CPs of secondary  (1)) are shaded in gray. Highly conserved residues in the partial sequence alignment below are boxed. Also shown is the % similarity of PcpS toward E. coli EntD as well as Sfp and AcpS from B. subtilis. metabolism, PcpS behaves like a PPTase of primary metabolism toward ACP substrates.

EXPERIMENTAL PROCEDURES
General Techniques E. coli was grown on LB medium. Antibiotics were used at the following concentrations, ampicillin 100 g/ml, kanamycin 25 g/ml, gentamycin 10 g/ml. For E. coli techniques, such as transformation and plasmid preparation, standard protocols were used (19). Vent polymerase (New England Biolabs, Schwalbach, Germany) or Pwo polymerase (Roche Molecular Biochemicals) was used to amplify gene fragments for cloning and expression purposes. Oligonucleotides were purchased from MWG Biotech (Ebersberg, Germany). [ 3 H]CoA was purchased from Hartmann Analytics (Braunschweig, Germany).
For complementation and interaction experiments (Matchmaker Two-hybrid System 3, Clontech, Heidelberg, Germany), S. cerevisiae was grown on synthetic dropout (SD) minimal medium. Yeast extracts were prepared as described in the manufacturer's manual for the Matchmaker Two-hybrid System 3 (PT3247-1, Clontech, Heidelberg, Germany).
Antibodies against GAL4 DNA BD (Clontech, Heidelberg, Germany) and c-Myc (Clontech, Heidelberg, Germany) were used for Western blot analysis that was carried out according to the guidelines for the ECL kit (Amersham Biosciences) with horseradish peroxidase-conjugated goat anti-mouse IgG (Sigma) as secondary antibody.
For mating experiments, P. aeruginosa PA01 was grown on LB medium. Antibiotics were used at the following concentrations: carbenicillin, 500 g/ml; gentamycin, 150 g/ml. Sucrose was added to solid medium at 50 mg/ml.

Construction of Plasmids
All plasmids and strains used in this study are summarized in Table II. Construction of pQE70-PA1165-The PA1165 gene encoding the P. aeruginosa carrier protein synthase (PcpS) was amplified by PCR with oligonucleotides 5Ј-ATAGCATGCGCGCCATGAACGACCGTCTC-3Ј and 5Ј-ATAAGATCTGGCGCCGACCGCCACCAG (restriction sites are underlined) from chromosomal DNA of P. aeruginosa PA01 and, after restriction digest of the amplified fragment, ligated into the SphI and BglII sites of pQE70. The resulting plasmid pQE70-PA1165 encodes the recombinant PcpS with a C-terminal tag RSHHHHHH.
Construction of pQE70-PA2966 -The PA2966 gene encoding the acyl carrier protein (ACP) was amplified by PCR with oligonucleotides 5Ј-ATAGCATGCGCACCATCGAAGAACGCG-3Ј and 5Ј-ATAAGATCTTT-GCTGGTGAGCAACGATG-3Ј from chromosomal DNA of P. aeruginosa PA01. The amplified fragment was digested with SphI and BglII and ligated into the SphI and BglII sites of pQE70. The resulting plasmid pQE70-PA2966 encodes the recombinant ACP with a C-terminal tag RSHHHHHH.
Construction of pQE70-pchEArCP-The fragment containing the ArCP domain from the first module (pchE1) of pyochelin synthetase E was amplified by PCR with oligonucleotides 5Ј-ATAGCATGCATCTG-CCCCCCGATTCCCG-3Ј and 5Ј-ATAAGATCTGTCCGCGCAGGCCAG-CA-3Ј from chromosomal DNA of P. aeruginosa PA01. The amplified fragment was digested with SphI and BglII and ligated into the SphI and BglII sites of pQE70. The resulting plasmid pQE70-pchEArCP encodes the excised recombinant ArCP (residues 1-87 of pchE, 10.5 kDa) with a C-terminal tag RSHHHHHH.
Construction of pQE70-pvdD1-PCP-The fragment containing the PCP domain from the first module (pvdD1) of pyoverdin synthetase D was amplified by PCR with oligonucleotides 5Ј-ATAGCATGCATCGA-GCGCCCGGTAGC-3Ј and 5Ј-ATAAGATCTCAATCCCTGGGCGAACG from chromosomal DNA of P. aeruginosa PA01. The amplified fragment was digested with SphI and BglII and ligated into the SphI and BglII sites of pQE70. The resulting plasmid pQE70-pvdD1-PCP encodes the excised recombinant pvdD1-PCP (residues 1067-1140 of pvdD, 9.3 kDa) with a C-terminal tag RSHHHHHH.
Construction of BS-LYS5::HIS3-By using oligonucleotides 5Ј-GAT-GCGGCCGCCTGAGTCGAACAATGCCTTACG-3Ј and 5Ј-CCGCTCGA-GTGATCAATCTGATGATGGCGG-3Ј, a 1445-bp fragment was amplified by PCR from chromosomal DNA of S. cerevisiae GSY155 that contained 229 bp upstream and 397 bp downstream of the LYS5 ORF. This fragment was digested with NotI and XhoI and ligated into the NotI and XhoI restriction sites of pBluescript KS (Stratagene, Heidelberg, Germany) generating plasmid BS-LYS5. The disruption plasmid BS-LYS5::HIS3 was constructed by replacing the internal NcoI/EcoRV of LYS5 with an NcoI/blunt-ended fragment of pJJ215 (20) containing the HIS3 gene.
Construction of pGBKT7-LYS5-The LYS5 gene encoding the S. cerevisiae Lys5 PPTase was amplified by PCR using oligonucleotides 5Ј-CCGCTCGAGTTATAAACCATCATTTTC-3Ј and 5Ј-GGGAATTCGT-TAAAACGACTGAAGTA-3Ј from chromosomal DNA of S. cerevisiae GSY155. Following restriction digest with EcoRI and XhoI, the resulting fragment was ligated into the EcoRI and SalI restriction sites of pGBKT7 (Clontech, Heidelberg, Germany).
Construction of pGBKT7-entD-The entD gene encoding the E. coli EntD PPTase was amplified by PCR using oligonucleotides 5Ј-CCGC-TCGAGTTAATCGTGTTGGCACAGCG-3Ј and 5Ј-GGGAATTCGTCGA-TATGAAAACTACG-3Ј from chromosomal DNA of E. coli Top10FЈ. The resulting fragment was digested with EcoRI and XhoI and ligated into the EcoRI and SalI restriction sites of pGBKT7 (Clontech, Heidelberg, Germany).
Construction of pGBKT7-PA1165-The PA1165 gene encoding PcpS was amplified by PCR using oligonucleotides 5Ј-CGGAATTCATGCGC-GCCATGAACGACCG-3Ј and 5Ј-CCGCTCGAGTCAGGCGCCGACCG-CCACC-3Ј from chromosomal DNA of P. aeruginosa PA01. The resulting fragment was digested with EcoRI and XhoI and ligated into the EcoRI and SalI restriction sites of pGBKT7.
Construction of p⌬pcpS-A 2904-bp fragment was amplified by PCR from chromosomal DNA of P. aeruginosa PA01 that contained 1119 bp upstream and 1057 bp downstream of the PA1165 ORF using primers 5Ј-ATAGAATTCGGTTGAGCCCGATCTTGC-3Ј and 5Ј-ATAAAGCT-TCTTTGCCTGGCCGAATGG-3Ј. This fragment was digested with EcoRI and HindIII and ligated into the EcoRI and HindIII restriction sites of pEX18Ap (21) generating plasmid pEX18Ap-5ЈpcpS3Ј. This plasmid served as template in an inverse PCR using primers 5Ј-ATA- ACTAGTTCAGGCGTTCCCCGGCGT-3Ј and 5Ј-ATACATATGCCCAC-CAGTCACGTGGCG-3Ј. The resulting fragment was digested with SpeI and NdeI and ligated with a fragment containing the aacC1 gene (conferring gentamycin resistance) that had been amplified by PCR from pX1918G (22) with primers 5Ј-ATACATATGCGGTTCGGC-CAGCGGCAA-3Ј and 5Ј-ATAACTAGTCCGAACAACTCCGCGGCC-3Ј, and treated with SpeI and NdeI, to give the disruption plasmid p⌬pcpS.

Overproduction and Purification of Recombinant Proteins
E. coli M15-pREP4 was transformed with pQE70-PA1165, pQE70-PA2966, and pQE70-pchEArCP to give strains RF6, RF7, and RF9, respectively, for the production of the His 6 fusion proteins PcpS, ACP, and ArCP. 5 ml of an overnight culture of RF6, RF7, and RF9 in LB were inoculated into 500 ml of the same medium. The culture was grown at 37°C and 300 rpm. Expression was induced by addition of 1 mM isopropyl-␤-D-thiogalactopyranoside (final concentration) at an A 600 of 0.7, and the culture was allowed to grow for an additional 3 h before being harvested by centrifugation at 4,500 ϫ g and 4°C. The cells were resuspended in buffer A (50 mM Hepes, 300 mM NaCl (pH 8.0)) and disrupted by three passages through a cooled French pressure cell. The resulting crude extract was centrifuged at 36,000 ϫ g at 4°C for 30 min. Protein purification using Ni 2ϩ -affinity chromatography was carried out as described previously (7). The presence of the respective protein in the fractions was detected using SDS-PAGE analysis (15% Laemmli gels). Fractions containing the desired protein were pooled and subsequently concentrated using Vivaspin (Vivascience AG, Hannover, Germany) with a molecular mass cut-off of 5 kDa in the case of all carrier proteins and a cut-off of 10 kDa in the case of PcpS. The concentrated fractions were applied to a Superdex TM G-75 26/60 gel filtration column (Amersham Biosciences) that had been equilibrated with 50 mM Tris/ HCl (pH 7.0); 4-ml fractions were collected. Those fractions containing the desired protein were pooled and concentrated as described above, brought to 10% glycerol (v/v), and stored at Ϫ80°C. In the case of PcpS, all buffers contained 30% glycerol. TycC3-PCP, AcpS from B. subtilis, Sfp from B. subtilis, and ACP from B. subtilis, hereafter referred to as PCP, AcpS, Sfp and B.s.-ACP, respectively, were produced and purified as described previously (6, 7). Protein concentrations were determined based on the calculated extinction coefficient at 280 nm: PcpS-His 6 38,220 M Ϫ1 cm Ϫ1 ; ACP-His 6 1280 M Ϫ1 cm Ϫ1 ; ArCP-His 6

Determination of the Oligomeric State of PcpS
A Superdex TM G-75 26/60 gel filtration column (Amersham Biosciences) was calibrated using the following proteins (numbers in parentheses indicate the size of the protein and the amount used): aprotinin (6.5 kDa, 3 mg), cytochrome c (12.4 kDa, 2 mg), carboanhydrase (29 kDa, 2 mg), and egg albumin (45 kDa, 5 mg). Determination of the void volume, V 0 , of the column was carried out using ferritin (450 kDa, 5 mg). Isocratic elution at 1 ml/min of the proteins was performed with buffer 50 mM Tris/HCl (pH 7.0) using an Ä kta purifier system (Amersham Biosciences); absorbance at 220 nm was monitored. The retention times, V, of the proteins are as follows: aprotinin, 247.14 ml; cytochrome c, 214.97 ml; carboanhydrase, 186.21 ml; egg albumin, 163.15 ml; ferritin, 108.96 ml (V 0 ). The V/V 0 values of the proteins were plotted against the log(kDa), and a linear fit was applied. 5 mg of the His 6tagged PcpS was injected onto the same column and eluted in the same manner. The retention time of the His 6 -tagged PcpS was 185.67 ml.

Radioassay for the Detection of Post-translational Modification Activity of PcpS
PcpS activity was assayed using a radioactive assay method essentially as described previously (7). This method measures the incorporation of the 3  Reactions were stopped by the addition of 0.8 ml of ice-cold 10% trichloroacetic acid (w/v) and 15 l of bovine serum albumin (25 mg/ml). Precipitated protein was collected by centrifugation at 13,000 rpm and 4°C for 15 min. The pellet was washed twice with 0.8 ml of ice-cold trichloroacetic acid (w/v) and resuspended in 180 l of formic acid. The resulting suspension was mixed with 3.5 ml of Rotiszint Eco Plus scintillation fluid (Roth, Karlsruhe, Germany) and counted using a 1900CA Tri-Carb liquid scintillation analyzer (Packard Instrument Co., Dreieich, Germany).

Kinetic Analysis of PcpS
For kinetic studies, the amount of holo-carrier protein formed was determined by an HPLC method essentially as described previously (7). 800-l reaction mixtures contained the apo-carrier protein (1.5-250 M Plasmid for yeast two-hybrid system (interaction PcpS-ArCP) This work pGBKT7-pvdD1-PCP Plasmid for yeast two-hybrid system (interaction PcpS-pvdD1-PCP) This work pGBKT7-lamC Control plasmid for yeast two-hybrid system encoding GAL4-BD-c-Myc-human lamin C Clontech pGBKT7-p53 Control plasmid for yeast two-hybrid system encoding GAL4-BD-c-Myc-murine p53 Clontech pGADT7-SV40TAg Control plasmid for yeast two-hybrid system encoding GAL4-BD-c-Myc-SV40 large Tantigen  nM; reaction mixtures were incubated at 37°C for 10 -30 min. The reaction was stopped and the protein precipitated by the addition of 10% trichloroacetic acid. Precipitated protein was collected by centrifugation at 13,000 rpm and 4°C for 30 min using a microcentrifuge. The pellet was resuspended in 120 l of 50 mM Tris/HCl (pH 8.8). In the case of PCP, a 3-100-l sample of this solution was injected onto an analytical reversed phase HPLC column (Nucleosil C18, 250 mm, 5 m, 300 Å, Macherey & Nagel, Germany) that had been equilibrated with 60% solvent A (0.1% trifluoroacetic acid). Absorbance at 220 nm was monitored. Apo-and holoPCP could be separated by applying a 13.5-ml linear gradient to 57.3% solvent B (acetonitrile in 0.1% trifluoroacetic acid) followed by a 0.9-ml linear gradient to 95% solvent B (flow rate of 0.9 ml/min and temperature of 45°C). Under these conditions, holoPCP migrates faster than apoPCP (10.52 and 12.15 min, respectively). The amount of holoPCP formed was determined by comparing the peak area of the holoPCP formed with those of both apoPCP and holoPCP and subtracting the amount of holoPCP that was already present after the heterologous expression of the protein in E. coli (see "Results").
In the case of all other carrier proteins, a 24.3-ml linear gradient 5-70% solvent B (to 60% in the case of ArCP) followed by a 2.7-ml linear gradient to 95% solvent B at 0.9 ml/min at 45°C was applied.

Deletion of the PA1165 Gene Encoding PcpS
The vector p⌬pcpS was used for gene deletion experiments, and the sacB-based strategy described by Hoang et al. (21) was employed. Mating procedure and isolation of potential mutants was done essentially as described by Hoang et al. (21) with minor modifications.
The mutated or the wild type PA1165 genes were identified by PCR. Oligonucleotides 5Ј-ATAGAATTCGGTTGAGCCCGATCTTGC-3Ј and 5Ј-ATAAAGCTTCTTTGCCTGGCCGAATGG-3Ј were used to amplify an ϳ3-kb fragment in both cases. Digestion of this fragment with AgeI yields two 1.5-kb fragments in case of the wild type gene, whereas digestion with SpeI yields a 1-kb and a 2-kb fragment confirming the presence of the gentamycin cassette in this fragment.

Complementation of a LYS5 Deletion by PcpS
The ability of PcpS to complement the S. cerevisiae PPTase Lys5, encoded by LYS5, in vivo was used as a test to determine whether ORF PA1165 codes for a PPTase. The LYS5 deletion strain was constructed essentially as described previously (23). The haploid strain GSY155 was transformed with a PCR fragment of BS-LYS5::HIS3 to give strain AY61. The integration of HIS3 in the chromosomal LYS5 gene was confirmed by PCR from genomic DNA (not shown) and by loss of growth on lysine dropout plates (Fig. 3A). Strain AY61 was transformed with pGBKT7-LYS5, pGBKT7-entD, pGBKT7-PA1165, pGBKT7-lamC and the empty vector pGBKT7. The resulting strains were streaked on plates lacking lysine/tryptophan as a test for complementation or, as a control, on plates lacking histidine/ tryptophan or tryptophan, respectively. The plates were incubated at 30°C for 2 days. As shown in Fig. 3A, neither the strain transformed with the empty plasmid nor strain AY61-pGBKT7-lamC that expresses the human lamC, which is known not to form complexes nor to interact with most other proteins (24), could grow on plates lacking lysine. The control strain AY61-pGBKT7-LYS5, however, shows that the PPTase Lys-5 can complement itself, as expected. The ability of E. coli EntD and PcpS to complement the ⌬LYS5 strain was verified by streaking strains AY61-pGBKT7-entD and AY61-pGBKT7-PA1165 on plates lacking lysine as shown in Fig. 3A.
Complementation of Lys-5 in vivo characterized PcpS as a functional PPTase, however, does not define its substrate specificity. We therefore applied an in vivo and an in vitro approach to test putative protein substrates of PcpS (see below).

Overproduction and Purification of P. aeruginosa PcpS, ACP, and ArCP
All proteins were produced as C-terminal His 6 tag fusion proteins and purified using Ni 2ϩ -nitrilotriacetic acid-affinity chromatography followed by gel filtration (see "Experimental Procedures"). SDS-PAGE analysis (not shown) revealed two bands in the case of ACP resulting from partial apo-to holoconversion of the protein by an E. coli PPTase during expression as described previously for other CPs (7). B. subtilis ACP and PCP were produced and purified as described previously (6, 7). 5 mg of PcpS, 44 mg of ACP, and 19 mg of ArCP with a purity of Ͼ99% were obtained per liter of cell culture. Interestingly, PcpS precipitated after Ni 2ϩ -nitrilotriacetic acid purification. We have since added 30% glycerol to all buffers during the purification process. Production of pvdD1-PCP in E. coli M15 failed as judged by SDS-PAGE with subsequent Western blot analysis using a penta-His antibody (not shown).

Determination of the Oligomeric State of PcpS
A Superdex TM G-75 26/60 gel filtration column (Amersham Biosciences) was calibrated as described under "Experimental Procedures." The V/V 0 of the proteins used to calibrate the column was plotted against their log(kDa), and a linear fit was applied (not shown). The V/V 0 of the His 6 -tagged PcpS eluted from the same column was 1.704, indicating an apparent size of 27 kDa. This is in agreement with the size of the monomeric, His 6 -tagged protein (calculated size, 27.8 kDa).

Post-translational Modification of Carrier Protein Substrates by PcpS
In Vivo Studies on PcpS/Carrier Protein Partnerships Using the Yeast Two-hybrid System-Judging from the size (242 aa, 27 kDa) and the pI, it was assumed that ORF PA1165 codes for an Sfp-type PPTase. It was therefore expected to be able to modify CPs of the secondary metabolism of P. aeruginosa. For this purpose, the matchmaker two-hybrid system 3 (Clontech, Heidelberg, Germany) was used to test putative protein substrates of PcpS in vivo. Both pchEArCP and pvdD1-PCP were PCR-amplified from the respective pQE70 constructs (Table II) and cloned into the pGBKT7 vector. S. cerevisiae AH109 was transformed with either pGADT7-PA1165 or, as a control, with the empty pGADT7 and with pGBKT7-pchEArCP or pGBKT7-PvdD1-PCP. Transformants were streaked on plates lacking leucine/tryptophan (control) or on plates lacking leucine/tryptophan/histidine/adenine (test for activation of the reporters) and incubated at 30°C until colonies were visible. As a positive control, AH109 cells were transformed with pGBKT7-p53 and pGADT7-SV40TAg. Murine p53 encoded by the former and SV40 large T-antigen encoded by the latter were chosen because they are known to interact in yeast two-hybrid systems (25,26). As a test for autoactivation of PcpS, S. cerevisiae AH109 was transformed with pGBKT7-p53 and pGADT7-PA1165. All strains created were able to grow on plates lacking leucine/tryptophan as would be expected of AH109 strains harboring both a pGADT7 and a pGBKT7 construct (not shown). On plates lacking, in addition to this, histidine and adenine, activation of the reporter was observed in the positive control (not shown) and in strains AH109-pGADT7-PA1165 that carried either pGBKT7-pchEArCP or pGBKT7-pvdD1-PCP showing a clear interaction between these two CPs and PcpS (not shown). The interaction between PcpS and the former was also shown in vitro (see below) but pvdD1-PCP failed to express in E. coli and was therefore not available for the in vitro characterization of PcpS. The negative control using the empty vector pGADT7 as the second plasmid did not grow on these plates, as expected. Also, autoactivation of PcpS was not observed (not shown). This experiment revealed that, as would be expected of an Sfp-type PPTase, PcpS was indeed able to modify ArCP and pvdD1-PCP, both of which are CPs of the siderophore metabolism of P. aeruginosa.
In Vitro Studies Using 3 H-Labeled CoA-Because PcpS seemed to be the only PPTase present in P. aeruginosa, we expected this enzyme to also recognize and modify other CPs, especially the ACP of FAS. To assess whether the recombinant CPs are substrates of PcpS, a radioassay was applied. We assumed that both PcpS and Sfp would be able to modify all CPs described above, whereas we expected AcpS, being a PPTase of primary metabolism, to be more selective in this respect. However, it is worth mentioning that, although ArCP is normally part of an NRPS, it is nevertheless an acidic protein much like an ACP. Thus, the outcome of this experiment was not quite as clear as Fig. 4 shows. As expected, PcpS and Sfp recognize and modify B.s.-ACP, ACP, and ArCP. AcpS, on the other hand, efficiently modifies its natural substrate B.s.-ACP but fails to convert ArCP to its holo form, proving once again that it does not cross-interact with secondary metabolism. This is the first time that AcpS from B. subtilis was tested with other recombinant ACPs of FAS. Although this is a mere qualitative assay, AcpS seems to modify ACP to almost the same degree as does PcpS.
These in vitro studies revealed that PcpS efficiently recognizes and modifies the ACP of primary metabolism in addition to CPs of secondary metabolism, so there would be no need for a supplementary AcpS-type PPTase in P. aeruginosa. Consequently, we proceeded to gather quantitative information on the modification of these CPs.

Biochemical Characterization of PcpS
An HPLC assay was applied to determine the catalytic efficiency of P. aeruginosa PcpS. The ratios of apo-to holoCP after heterologous production in E. coli were as follows: ACP, 66 to 34%; ArCP, 100 to 0%; B.s.-ACP, 92 to 8%; PCP, 95 to 5%. Before the analysis of PPTase activity toward these substrates, the pH profile of PcpS was determined. Fig. 5A shows that the enzyme has a clear pH optimum at pH 6.5 with ACP, exhibiting less than 27 and 62% activity at pH 5 and 7, respectively. This pH was therefore chosen for the determination of K m and k cat values.
Kinetic constants were determined through a Michaelis-Menten fit of the experimental data sets. Kinetic constants of PcpS, Sfp, and AcpS that were determined in this study or elsewhere are summarized in Table III. For the determination of the kinetic constants for CoA, we varied its concentration between 1 and 500 M, whereas the ACP concentration was kept at 200 M. Saturation, however, was reached very early at 5 M, and PcpS was in fact inhibited by higher concentrations of CoA (Fig. 5B). Normal Michaelis-Menten behavior was observed if only the velocity values for CoA concentrations ranging from 0.5 to 5 M were used for the determination of kinetic constants. In all further experiments, the CoA concentration was therefore kept at 5 M. The Michaelis-Menten constant for CoA concentrations between 0.5 and 5 M was 1.1 Ϯ 0.3 M with the k cat being 168 Ϯ 13.8 min Ϫ1 . This is the first time that a PPTase is shown to be inhibited by CoA, neither Sfp nor AcpS of B. subtilis or E. coli show this phenomenon (3,4,7). In comparison with the values determined for other PPTases, PcpS has K m and k cat values for CoA that are very close to the values determined for Sfp. As described for other ACPs (6, 7), two K m and k cat values could be determined for ACP. This was first seen with AcpS of Streptococcus pneumoniae (9) and was also confirmed for AcpS and Sfp of B. subtilis (7). The kinetic constants of PcpS and AcpS for their respective cognate apoACP are comparable (Table III) Table III) that are increased by a factor of about 4.2 and 1.8, respectively, for low and high apoACP concentrations, and k cat values that are diminished by a factor of at least 13 (1.7 min Ϫ1 for low ACP concentrations) and 10 (12.5 min Ϫ1 for high ACP concentrations) compared with the former two PPTases (Table III). We also determined two kinetic constants for apoArCP and apoB.s.-ACP. The outcome of this experiment for the latter was not surprising; however, ArCPs are part of NRPS systems like PCPs, and this incident has not been encountered with any other PPTase. In comparison with values determined for Sfp, the k cat values of PcpS for ArCP are low (0.9 and 6.5 min Ϫ1 , respectively, for low and high apoArCP concentrations). For instance, the k cat value of Sfp for apoEntB-ArCP (4) was determined to be 10 times higher at 65 min Ϫ1 . PcpS also modifies the non-cognate B.s.-ACP. Although the K m value for low B.s.-ACP concentrations (5.9 M) is increased by a factor 12, the K m value for high concentrations (37.8 M) is 1.8 times lower compared with its cognate ACP (Table III). The k cat values are about twice as high and 0.9 times as high (50.9 and 108.3 min Ϫ1 ), respectively, for low and high B.s.-ACP concentrations. So it seems that P. aeruginosa ACP (59% identity with B.s.-ACP) is not a special case but that PcpS exhibits only slightly lower catalytic efficiency with non-cognate ACPs. Interestingly, we could also determine two kinetic constants for apoPCP. First saturation was reached between 1 and 10 M apoPCP with a K m of 1.  (Table III).

Deletion of PA1165
To substantiate further the in vivo role of PcpS, especially concerning fatty acid and siderophore metabolism, we attempted a knockout of the corresponding gene, PA1165. For this purpose, we decided to use the reliable knockout system based on the pEX18Ap suicide vector, which previously has been used successfully (21). In the suicide vector p⌬pcpS, a gentamycin resistance cassette replaces the PA1165 gene. The cassette is flanked by two ϳ1-kb fragments of homologous DNA. In addition, the vector contains an ampicillin/carbenicillin resistance cassette and a sacB gene for counter-selection on a chromosomally integrated vector by a single crossover event (see "Experimental Procedures"). After the mating experiment, several hundred gentamycin-resistant mutants were isolated, 50 of which were chosen for further analysis and found to be carbenicillin-resistant, which indicates a single crossover event in all isolated mutants. The presence of the gentamycin cassette replacing the PA1165 gene and the wild type PA1165 gene was confirmed by PCR in 10 mutants. PCR and restriction analysis (see "Experimental Procedures") confirmed the presence of the wild type and the gentamycin-cassette disrupted gene in all strains tested (not shown). The complete suicide vector was evidently integrated into the genome by a single crossover event. As expected, these mutants were also sensitive to growth on sucrose. We selected for possible double crossover mutants on LB medium containing sucrose and gentamycin. We obtained spontaneous sucrose and gentamycin-resistant mutants of each of the 10 initially sucrose-sensitive mutants. The isolated spontaneous mutants also lost carbenicillin resistance, which is another indication for a possible double crossover event by deprivation of the vector. These findings suggest the replacement of the wild type gene by the resistance cassette. However, in this case, PCR and restriction analysis of the DNA of these mutants (not shown) revealed the presence of the wild type and the mutated PA1165 gene in all mutants. The outcome of these experiments utilizing this reliable knockout system strongly indicates that the PA1165 gene is essential for growth of P. aeruginosa and cannot be deleted. DISCUSSION Post-translational modification is absolutely essential for Ppant-dependent CPs to be functionally active, be it as distinct proteins like ACPs (primary metabolism) or as integrated domains of NRPSs and PKSs such as PCPs or ArCPs (secondary metabolism). The PPTase responsible for the modification of CPs of primary metabolism is usually AcpS, whereas CPs of secondary metabolism are modified by a second, Sfp-type, PPTase. For this reason, most organisms harbor two PPTases (Table I) if they produce a secondary metabolite via an NRPS or PKS. Some of the recently sequenced organisms, however, seem to have sorted out their PPTase of primary metabolism. We present here the characterization of PcpS of P. aeruginosa, the first example of a PPTase that has to serve primary as well as secondary metabolism, in vivo and in vitro with natural substrates.
A recent study (7) has shown that deletion of acpS in a B. subtilis strain that also possesses a functional Sfp gene has no negative effect on the organism. The growth curves of the mutant strain showed no difference compared with wild type. Moreover, the production of the secondary metabolite surfactin in this strain was not impaired, indicating that Sfp is catalytically competent enough for the modification of all CPs present. In the light of these results it seems plausible that organisms harboring an AcpS-type PPTase in addition to an Sfp-type PPTase would gradually expel this dispensable ballast. Our blast searches in the P. aeruginosa genome using AcpS of B. subtilis and E. coli, Sfp and EntD as probes, revealed that this organism seems to have merely one PPTase, termed PcpS. ORF PA1165 encodes PcpS, a protein of about 27 kDa (242 aa), with an isoelectric point of 6.77 that shows the typical core motifs of PPTases (Fig. 2). These data lead to the assumption that ORF PA1165 encodes a PPTases of the Sfp type. However, alignments with other PPTases revealed that the sequence similarity of PcpS to the well characterized "prototype" PPTases Sfp is merely 13.9% (Fig. 2). This led us to suspect that this PPTase may represent a special case in the pool of Sfp-type PPTases.
Mootz et al. (23) established a genetic system to test for PPTase activity by complementation of S. cerevisiae Lys-5. Essentially the same approach was used in this study to show that ORF PA1165 does in fact code for a PPTase. LYS5 encodes Lys-5, a PPTase that is essential for post-translational modification of the ␣-aminoadipate reductase (Lys-2) with the Ppant cofactor. PcpS as well as E. coli EntD and S. cerevisiae Lys-5 were expressed in a ⌬LYS5 strain of S. cerevisiae. The ability of these PPTases to complement Lys-5 in vivo was verified by streaking these strains on plates lacking lysine. As Fig. 3A shows, both PcpS and EntD can modify Lys-2 to its active holo form thereby conferring the ability to synthesize lysine on the LYS5 deletion strain which characterized them as PPTases.
Although for the best characterized Sfp-type PPTase, Sfp, no exemplary CP is known that is not accepted as substrate, the enzyme does differentiate between CPs of primary and secondary metabolism in vitro (7). As can be seen in Table IV, the catalytic efficiency of Sfp with PCP is 21.6 M Ϫ1 min Ϫ1 . It is worthwhile to stress that the tested PCP is not a natural substrate of Sfp, but the enzyme has been characterized with the excised PCPs of modules 1 and 2 of surfactin synthetase B (4). The catalytic efficiency with those PCPs was even higher at 80 and 31 M Ϫ1 min Ϫ1 , respectively. The situation is quite different with CPs of primary metabolism. Catalytic efficiencies of Sfp with B.s.-ACP are as low as 0.3 M Ϫ1 min Ϫ1 (Table  IV), which is also the case for other ACPs with which it has been tested (4). These facts show that Sfp is clearly an enzyme  Organisms that carry both types of PPTases seem to have selected one PPTase responsible for primary metabolism and have optimized the other one for the modification of CPs of secondary metabolism. In E. coli this idea extends even further if the two PPTases AcpS and EntD are taken into account. Whereas the kinetic efficiency pattern for AcpS of E. coli and B. subtilis is the same, EntD has been shown not to cross-interact with primary metabolism (1). It has been argued that the justification for the existence of two different PPTases could be a defined regulation of the CoA pool (27). The other explanation that AcpS-type PPTase cannot recognize type I CPs for steric or size reasons has recently been ruled out (6).
P. aeruginosa is a Gram-negative organism with PcpS as the sole PPTase that exhibits high similarity to E. coli EntD. If this PPTase behaved like EntD, the organism would not be able to survive without an additional AcpS-type PPTase. We first determined the protein partners of PcpS and found that it recognizes, in contrast to AcpS, both ACPs and ArCP (Fig. 4) as well as pvdD1-PCP as shown in the yeast two-hybrid system. This is in accord with the broad substrate specificity that would be expected of this PPTase. Also, the pH optimum determined is very close to that of Sfp. The interpretation up to this point was that we are faced with the same situation found for the acpS deletion mutant of B. subtilis described above. However, our first attempt to determine the Michaelis-Menten constant of PcpS for CoA revealed that this enzyme is inhibited by higher concentrations of this cofactor which has not been found for any other PPTase (Fig. 5B). Moreover, further characterization of the enzyme showed that it is evidently optimized for the activation of CPs from primary metabolism. The catalytic efficiency with its natural ACP is 32.5 and 2.6 M Ϫ1 min Ϫ1 , respectively, whereas the efficiencies with ArCP are 1.1 and 0.13 M Ϫ1 min Ϫ1 (Table IV). This corresponds to a 30-and 20-fold drop in catalytic efficiency, respectively, compared with ACP.
The enzyme also efficiently recognizes and modifies noncognate CPs. After several attempts failed to express the P. aeruginosa PCPs of the pyoverdin and pyochelin NRPSs in E. coli, we turned to the well characterized excised TycC3-PCP for the characterization of PcpS. Nevertheless, both the PCP of the first module of the pyoverdin synthetase D (pvdD1-PCP) and ArCP were successfully tested for interaction with PcpS in a yeast two-hybrid system (see "Results"). In vitro, the situation for the TycC3-PCP is the same as for ArCP, catalytic efficiency is low at 0.04 and 1.1 M Ϫ1 min Ϫ1 . B.s.-ACP, on the other hand, is modified at 8.6 and 2.9 M Ϫ1 min Ϫ1 ; especially the last value for high (25-206 M) concentrations of B.s.-ACP is the same (within experimental error) as for the natural ACP. The situation is almost the exact reverse found for Sfp.
Another very notable difference between PcpS and Sfp was that we could determine two K m and k cat values for all carrier proteins tested. This phenomenon has first been described for AcpS of S. pneumoniae (9) and also recently for AcpS of B. subtilis (7) with B.s.-ACP. By contrast, saturation at low PCP and ArCP concentrations with increasing velocity values at higher concentrations has never been encountered with any other PPTase. It has been proposed that this is due to the fact that AcpS acts as a homotrimer, holding three active sites, and may therefore exhibit allosteric regulation (9). This sort of regulation is not conceivable for the monomeric Sfp and PcpS because they only have a single active site (28). Another expla-nation for this behavior may be that CPs undergo a conformational change that is pH-dependent (29) and concentrationinduced (6). This may account for the values determined for Sfp with ACPs but points to a difference in the way substrate recognition is carried out by PcpS compared with Sfp.
The chromosome of B. subtilis MR168 carries the sequence information for the dhb NRPS cluster. This NRPS usually produces the catecholic siderophore bacillibactin (30). B. subtilis MR168, however, contains a defective sfp gene; thus all carrier proteins of the NRPSs remain in the inactive apo form and no siderophore is produced. P. aeruginosa produces two siderophores, namely pyoverdin and pyochelin, that contribute to the high virulence of this organism (18). Both are synthesized non-ribosomally, and thus production relies on posttranslational apo to holo conversion of the corresponding NRPS templates catalyzed by PcpS. In the light of the above stated facts, PcpS represents an excellent target for antimicrobial agents. The selective inhibition of this enzyme may be sufficient to get infections caused by P. aeruginosa under control. Two big advantages are connected with this approach. First, because catalytic efficiency is low with CPs of secondary metabolism but at least 20-fold higher with the ACP of FAS, it is possible that only siderophore synthesis is affected by partial inhibition of PcpS. This would leave Pseudomonas viable, albeit deprived of the pvd and pch siderophores that are associated with the high virulence of this organism (18). Because primary metabolism would not or be only slightly influenced, it is unlikely that the organism becomes resistant to the inhibitor. Second, if the agent is selective enough to inhibit Sfp-type PPTases only, other organisms that are beneficial to the host would not be killed off as is the case with many antibiotics.
Characterization of PcpS in vivo using a yeast two-hybrid system and in vitro with several different CPs has shown that PcpS is a PPTase capable of modifying CPs of primary and secondary metabolism. This, however, does not answer the question whether PcpS plays an essential functional role in both fatty acid and siderophore metabolism. To prove this we have attempted a disruption of the corresponding gene, PA1165. For this purpose, the p⌬pcpS vector was used in mating experiments where the PA1165 gene was replaced by a gentamycin resistance cassette and flanked by homologous DNA. However, subsequent PCR and restriction analysis of the conjugants revealed the presence of both the mutated and the wild type gene in all mutants that were gentamycin-sensitive and carbenicillin-and sucrose-resistant, indicating that the gene encoding PcpS is essential (see "Results"). Consequently, PcpS ought to possess a functional role in fatty acid synthesis. In addition, the blast searches did not unveil the presence of a second Sfp-type PPTase indicating that PcpS is also needed for siderophore metabolism.
Our data show that the sole PPTase of P. aeruginosa, PcpS, although related to E. coli EntD and Sfp of B. subtilis, has significantly different catalytic properties than other enzymes of this superfamily of PPTases. From these in vitro data we propose that PcpS is the first representative of a new sub-class of PPTases that putatively have a similar structure as Sfp-type PPTases but were evolutionary selected for high catalytic efficiency with CPs of primary metabolism. A crystal structure of PcpS would be an important contribution to make a decision in this case.