HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

J. Biol. Chem., Vol. 280, Issue 6, 4339-4349, February 11, 2005

This Article Abstract Full Text (PDF) All Versions of this Article: 280/6/4339    most recent M411026200v1 Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Schmoock, G. Articles by Keller, U. Search for Related Content PubMed PubMed Citation Articles by Schmoock, G. Articles by Keller, U. Social Bookmarking                      What's this?

Functional Cross-talk between Fatty Acid Synthesis and Nonribosomal Peptide Synthesis in Quinoxaline Antibiotic-producing Streptomycetes^*

Gernot Schmoock, Frank Pfennig, Julien Jewiarz, Wilhelm Schlumbohm, Werner Laubinger, Florian Schauwecker, and Ullrich Keller

From the Institut für Chemie, Arbeitsgruppe Biochemie und Molekularbiologie, Technische Universität Berlin, Franklinstrasse 29, D-10587 Berlin-Charlottenburg, Germany

Received for publication, September 24, 2004 , and in revised form, November 24, 2004.

	ABSTRACT

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
Quinoxaline antibiotics are chromopeptide lactones embracing the two families of triostins and quinomycins, each having characteristic sulfur-containing cross-bridges. Interest in these compounds stems from their antineoplastic activities and their specific binding to DNA via bifunctional intercalation of the twin chromophores represented by quinoxaline-2-carboxylic acid (QA). Enzymatic analysis of triostin A-producing Streptomyces triostinicus and quinomycin A-producing Streptomyces echinatus revealed four nonribosomal peptide synthetase modules for the assembly of the quinoxalinoyl tetrapeptide backbone of the quinoxaline antibiotics. The modules were contained in three protein fractions, referred to as triostin synthetases (TrsII, III, and IV). TrsII is a 245-kDa bimodular nonribosomal peptide synthetase activating as thioesters for both serine and alanine, the first two amino acids of the quinoxalinoyl tetrapeptide chain. TrsIII, represented by a protein of 250 kDa, activates cysteine as a thioester. TrsIV, an unstable protein of apparent M_r about 280,000, was identified by its ability to activate and N-methylate valine, the last amino acid. QA, the chromophore, was shown to be recruited by a free-standing adenylation domain, TrsI, in conjunction with a QA-binding protein, AcpPSE. Cloning of the gene for the QA-binding protein revealed that it is the fatty acyl carrier protein, AcpPSE, of the fatty acid synthase of S. echinatus and S. triostinicus. Analysis of the acylation reaction of AcpPSE by TrsI along with other A-domains and the aroyl carrier protein AcmACP from actinomycin biosynthesis revealed a specific requirement for AcpPSE in the activation and also in the condensation of QA with serine in the initiation step of QA tetrapeptide assembly on TrsII. These data show for the first time a functional interaction between nonribosomal peptide synthesis and fatty acid synthesis.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
The family of quinoxaline antibiotics embraces two structurally similar series of compounds, the triostins and quinomycins, produced by various Streptomyces strains (1). They consist of octadepsipeptide rings to which are attached two quinoxaline-2-carboxylic acid (QA)¹ residues in amide linkages (Fig. 1). The triostins and quinomycins differ in their cross-bridges, represented by either a disulfide or a thioacetal structure, respectively. These are contributed from the thiol groups of the two cysteines present in the octadepsipeptide rings (2) (Fig. 1). The various members of each group differ by amino acid substitutions in the peptide chains (3, 4). Other compounds with similarity to the quinoxaline antibiotics are the luzopeptins and sandramycin, which are cyclodecadepsipeptides carrying 3-hydroxyquinoline residues as side groups (5, 6).

View larger version (24K): [in this window] [in a new window]   FIG. 1. Structures of triostin A and quinomycin A (echinomycin). The amino acids in the peptide chains are: D-Ser (D-serine), Ala (alanine), MeCys (N-Methyl-L-cysteine), MeVal (N-Methyl-L-valine).

In contrast to the wealth of information on the mechanisms of action of the bis-intercalating chromopeptides, only little is known on their biosynthesis. In the case of the quinoxaline antibiotics, in vivo studies of quinomycin A (echinomycin) formation in Streptomyces echinatus have shown that QA is derived from tryptophan and that the amino acids of the peptide rings stem from the free cellular pool (12). Importantly, the demonstration of the conversion of triostins into quinomycins by reduction and rearrangement of the triostin cross-bridge with subsequent S-methylation showed that triostins are the immediate precursors of quinomycins. Accordingly, it was shown that triostin-producing streptomycete strains lack the converting enzyme activity, whereas quinomycin producers possess it (13).

Triostin synthetase I (TrsI), a 58-kDa protein present in triostin A-producing S. triostinicus and in echinomycin-producing S. echinatus, activates QA as the adenylate (14). It can also activate a number of structural analogues of QA such as quinoline-2-carboxylic acid, thieno[3,2b]pyridine-5-carboxylic acid (TPA), and 3-hydroxyquinoline-2-carboxylic acid. These compounds were readily incorporated into triostins or quinomycins when fed to cultures of S. echinatus instead of the endogenous QA to give new compounds (15). These findings suggested that TrsI is involved in the assembly of the quinoxalinoyl peptides, but this assembly remained enigmatic.

In several chromopeptide biosynthesis systems, adenylating enzymes (i.e. free-standing A-domains) have been described that activate aryl carboxylic acids as adenylates with subsequent transfer to specific binding proteins that contain ArCP (aroyl carrier protein) domains (16, 17). These ArCP domains are always fused to other proteins such as isochorismate lyase in the case of several aryl capped siderophore peptide syntheses, e.g. EntB in Escherichia coli, or to a nonribosomal peptide synthetase (NRPS) such as HMWP2 from Yersinia pestis (18, 19). The ArCP-thioester interacts with the next downstream module in the biosynthetic sequence by condensation with the tethered amino acid, resulting in an aroyl amino acid that can undergo elongation cycles until the peptide product is completed. The ArCP domains of siderophore synthetases resemble the peptidyl carrier protein (PCP) or acyl carrier protein (ACP) domains of the modular NRPS or polyketide synthases of type I (17). In the case of synthesis of the chromopeptide actinomycin, the binding protein for the 4-methyl-3-hydroxyanthranilic acid starter unit (4-MHA) was found to be a small free-standing ArCP domain, AcmACP, with significant similarity to fatty acyl carrier proteins of type II fatty acid synthase and polyketide synthases (20). AcmACP is loaded by the adenylating enzyme actinomycin synthetase I (ACMSI), and transfers the 4-MHA to the bimodular NRPS actinomycin synthetase II on which 4-MHA is condensed with threonine, the first amino acid of the actinomycin peptide chain. 4-MHA threonine is later elongated in four further cycles to the 4-MHA pentapeptide lactone end product (21) (Fig. 2).

View larger version (14K): [in this window] [in a new window]   FIG. 2. NRPS assembly line of the actinomycin half-molecule (4-MHA pentapeptide lactone). The assembly line comprises six modules including the "truncated" initiation module consisting of the stand-alone A-domain ACMSI and the stand-alone AcmACP. The symbols denote condensation domains (C), adenylation domains (A), peptidyl carrier protein domains (PCP), epimerization domain (E), methylation domains (M), thioesterase domain (Te). The amino acids in 4-MHA pentapeptide lactone chain are: Thr (L-threonine), D-Val (D-valine), Pro (L-proline), Sar (sarcosine, N-methyl-glycine), MeVal (N-methyl-L-valine).

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
Chemicals and Radiochemicals—L-[U-¹⁴C]Serine (155 Ci/mol), L-[U-¹⁴C]alanine (162 Ci/mol), L-[methyl-¹⁴C]methionine (56.7 Ci/mol), p-[ring-U-¹⁴C]toluic acid (56 Ci/mol), and L-[³⁵S]cysteine (>1000 Ci/mmol) were purchased from Amersham Biosciences. L-[U-¹⁴C]Valine (260 Ci/mol) was from ICN Pharmaceuticals (ICN Radiochemicals, Zoetermeer, The Netherlands). [Benzene ring-¹⁴C(U)]QA (28.8 Ci/mol) and tetrasodium [³²P]pyrophosphate (2.9 Ci/mmol) were from PerkinElmer Life Sciences. Quinoxalinoyl-2-chloride, pyrazine-2-carboxylic acid, quinoline-3-carboxylic acid, and echinomycin were obtained from Sigma. Quinoline-2-carboxylic acid was from Janssen (Beerse, Belgium). Thieno[3,2-b]-pyridine-5-carboxylic acid was kindly provided by Dr. S. Gronowitz (University of Lund, Sweden).

Strains and Cultures—S. echinatus A8331 was kindly provided by Drs. J. Nüesch and K. Scheibli (Basel). Streptomyces triostinicus ATCC21043 and Streptomyces chrysomallus ATCC11523 were obtained from the American Type Culture Collection (Manassus VA). Streptomyces griseoviridus DSM40229was from the Deutsche Sammlung für Mikroorganismen (Braunschweig, Germany). Maintenance, preparation of inocula for liquid culture of S. echinatus, and S. triostinicus, as well as growth in liquid cultures were carried out essentially as described previously, except that S. echinatus liquid cultures were incubated for up to 38 h depending on the maximum rate of [methyl-¹⁴C]methionine incorporation into echinomycin during short term incorporation experiments (14). Streptomyces lividans TK64 and Streptomyces coelicolor M145 were obtained from the John Innes strain collection (Norwich, United Kingdom). E. coli strains for cloning were DH5 and JM109, and the strain for expression was M15 (Qiagen, Hilden, Germany). They were maintained and grown according to standard protocols.

Plasmids, DNA Manipulations, Cloning, and Sequencing Procedures—Techniques for DNA isolation, manipulation, and transformation were as described (22, 23). The plasmids used for subcloning fragments were pTZ18U (Amersham Biosciences), pSP72 (Promega), or pSL1180 (Amersham Biosciences). Expression plasmids for E. coli were pQE30, pQE32, and pQE70 (Qiagen). Plasmid pIJ702 (24) was used as replicon for gene expressions in S. lividans TK64 as described in Ref. 25.

Analytical Methods—SDS-PAGE was performed according to Refs. 26 and 27. Protein determinations were carried out according to Ref. 28. Amino-terminal microsequencing of proteins was performed on an Applied Bioscience (ABI) Procise Microsequencer. DNA sequencing was performed on a ABI capillary electrophoresis sequencing system (3100 Genetic Analyzer). Amino acid and peptidyl thioester intermediates bound to protein were released either by performic acid treatment or saponification with 0.1 M NaOH and subsequent acidification and solvent extraction with ethyl acetate (EtOAc). Labeled compounds on thin layer chromatography (TLC) plates were detected by radioscanning with a Bertolt Radioscanning System or by autoradiography using Kodak Bio-Max x-ray film. Amino acid or peptidyl intermediates were analyzed by TLC. Reference material on plates was either visualized by spraying with ninhydrin (amino acids) or by UV detection (aroyl peptides). Amino acid hydrolysis was performed with 6 M HCl under reduced pressure for 22 h at 110 °C. Autofluorography or autoradiography of SDS-PAGE gels for visualization of labeled protein bands was as described (20). Radioactive determinations of filters or of liquid samples were achieved by liquid scintillation counting using a Wallac scintillation system. Echinomycin production in cultures was monitored by short term incorporation tests as described (14). Echinomycin or triostin extracted from cultures was identified by TLC along with authentic standards using solvent system I. Amino acids were chromatographed on silica gel plates using solvent system II. Quinoxalinoyl peptides were chromatographed on silica gel plates using solvent system III. Determination of optical configurations was done by TLC on chiral plates using solvent system IV.

Matrix-assisted Laser Desorption Ionization (MALDI)-TOF Spectrometry—Protein samples were analyzed with a Voyager-DE PRO (Applied Biosystems). The MALDI-TOF MS system was operated in the linear mode. Samples were further purified with ZipTipC18 and dissolved in a 50% solution of acetonitril, 0.1% trifluoroacetic acid containing sinapinic acid (5 mg ml^–1) and dried on a sample plate. A nitrogen laser operating at 337 nm and a 3-ns pulse rate was used. The accelerating voltage was set at 25 kV, and a delay of 350 ns was used to accelerate ions into the flight tube of the mass spectrometer. The mass scale (m/z 5000–20000) was calibrated with a mixture of peptides, and 80 laser shots were used to obtain each spectrum.

Buffers and Solvent Systems—Solvent systems were: I (EtOAc, MeOH, H₂O (100:5:5, by volume)), II (BuOH, AcOH, H₂O (4:1:1, by volume)), III (2-propanol, di-n-butylether, AcOH, H₂O (4:3:3:2, by volume)), and IV (acetone, MeOH, H₂O (10:2:2, by volume)). Buffer for nickel affinity chromatography was NTA (15% glycerol (w/v), 50 mM KPO₄ (pH 8.0), 300 mM NaCl, 1 mM benzamidine, and 1 mM phenylmethylsulfonyl fluoride) and buffer NTA-W (buffer NTA with 50 mM KPO₄ (pH 6.0)). Buffer A was 15% glycerol (w/v), 150 mM Tris-HCl (pH 8.0), 30 mM MgCl₂, 1 mM EDTA, 4 mM dithioerythritol, 1 mM benzamidine, 1 mM phenylmethylsulfonyl fluoride. Buffer B was 15% glycerol (w/v), 100 mM Tris-HCl (pH 8.0), 1 mM EDTA, 4 mM dithioerythritol, 1 mM benzamidine, 1 mM phenylmethylsulfonyl fluoride. Buffer C was buffer B with 2 mM EDTA and without phenylmethylsulfonyl fluoride and benzamidine. Buffer D was buffer C with 50 mM Tris-HCl (pH 8.0). Buffer E was buffer C with 100 mM Tris-HCl (pH 7.2). Buffer F was buffer E with 10% glycerol (w/v).

Cloning of trsA, trsB, and trsC from S. triostinicus—A probe suitable for screening the gene trsA encoding TrsI in a plasmid gene bank of S. triostinicus DNA in E. coli was obtained by PCR using forward primer FTRIO1 derived from a part of the known amino-terminal sequence of TrsI (5'-ATGCSGACGGCTTCGTCCCSTGGGTCGACCACCT-3') and reverse primer F2RC derived from the conserved A3 motif of the A-domains ACMSI (20), SnbA (29), and EntE (30) (5'-GGAATTCCCTTSGGCTTGCCGGTGGTGCC-3'). The template was chromosomal DNA from S. triostinicus. A 580-bp fragment was cloned in E. coli via SalI and EcoRI restriction sites using SalI/EcoRI-cleaved pTZ18U. Sequencing revealed that the fragment encoded an amino acid sequence that comprised the complete sequenced amino terminus of TrsI and showed high sequence similarity to ACMSI, SnbA, and EntE (31). The 580-bp gene fragment was ³²P-labeled by using a Random Primers DNA Labeling System (Invitrogen, Eggenstein, Germany) and used as a probe to localize the complete trsA gene by hybridization analysis of various restriction blots of chromosomal DNA of S. triostinicus.

A 6.1-kb BamHI fragment apparently comprising the complete trsA and its flanking regions was cloned by isolating BamHI-digested S. triostinicus total DNA size fractionated on agarose gels to about 6 kb, followed by fractionation and ligation of the mixture to BamHI-cleaved pTZ18U. After introduction by transformation into E. coli DH5, 8000 colonies were subjected to screening with the labeled probe. Two hybridizing colonies were found to contain the 6.1-kb BamHI band in the plasmid, which was named pTR1. To facilitate sequencing of the insert of pTR1 was removed by KpnI digestion with subsequent religation to yield pTR1sub5. Sequencing the 4-kb BamHI-KpnI insert contained in pTR1sub5 revealed the trsA gene flanked by the two genes trsB and trsC, to the 5'- and 3'-end, respectively (GenBank^TM accession number AY825941 [GenBank] ).

Expression of trsA in S. lividans TK64 —To express TrsI as a carboxyl-terminal His₆ fusion protein from the mel promoter of pIJ702, forward primer FTRS1 (5'-AAAGGAACGCCGCATGCTCGACGGTTTCGTCCCCCTG-3') and reverse primer RTRS1 (5'-CGTAGATCTTCAGTGGTGGTGGTGGTGGTGAGGCATCCGCGCCGTCACGGCACG-3') were used for PCR amplification with 200 ng of chromosomal DNA as template. The 1.6-kb fragment was digested with SphI and BglII and ligated to BglII/SphI-cleaved pSP72. From the plasmid obtained (pTRSA1) the fragment was excised with BglII and SphI and ligated to BglII/SphI-cleaved pIJ702 and introduced by transformation into S. coelicolor M145 or S. lividans TK64, to yield pTRSA2. For expression of TrsI from pTRSA2, S. lividans TK64 harboring pTRSA2 was grown for 3 days in 300-ml Erlenmeyer flasks containing 100 ml of YEME medium with 50 µg ml^–1 thiostrepton equipped with steel springs as baffles.

Expression of trsA in E. coli—Expression of trsA in E. coli was as amino- and carboxyl-terminal His₆ fusion proteins. The trsA gene was amplified by PCR using pTR1sub5 as template with forward primer FTRS1 (5'-AAAGGAACGCCGCATGCTCGACGGTTTCGTCCCCTG-3') and reverse primer GS230201 (5'-GGCAAGCTTTCAAGATCTAGGCATCCGCGCCGTCACGGCACGCTC-3'). The resulting 1.56-kb PCR fragment was digested with SphI and HindIII and ligated to SphI/HindIII-cleaved pQE32 to yield pTRSN1. Alternatively, the same PCR fragment was digested with SphI and BglII and ligated into SphI/BglII-cleaved pQE70, producing pTRSC1. pTRSN1 and pTRSC1 were introduced by transformation into E. coli strain M15, respectively. Cultures of M15/pTRSN1 and M15/pTRSC1 (1.6 liters of 2x YT medium, 100 µg ml^–1 ampicillin, 25 µgml^–1 kanamycin) were grown at 37 °C to an A₆₀₀ of 0.7 and then induced with 0.5 mM isopropyl 1-thio--D-galactopyranoside (final concentration). Cultivation was continued for 18 h at 29 °C and cells finally harvested.

Isolation of QA-binding Protein from S. echinatus—All operations were carried out at 2–4 °C. Mycelium of S. echinatus (12.5 g wet weight) was suspended in a 3.5-fold volume of buffer C (w/v). After passage through a French press (Aminco) at 10,000 p.s.i., MgCl₂ was added at 15 mM (final concentration) and solid DNase (Type I, grade 2, Sigma, 50 µg ml^–1 final concentration) were added. After stirring for 1 h on ice the suspension was centrifuged for 30 min in a Sorvall centrifuge at 25,000 x g. The supernatant (83 ml) was applied to a Q-Sepharose (fast flow) column (4 x 5 cm) and, after washing, a 200-ml 0–0.5 M NaCl gradient was applied. 5-ml fractions were collected. QA binding activity eluted between 0.3 and 0.4 M NaCl. Active fractions were pooled and diluted to 0.1 M NaCl concentration by adding 3 volumes of buffer D. The solution was then applied onto an anion exchange column (Resource Q, 6 ml) and proteins were eluted with 1 M NaCl. Fractions of 0.5 ml were collected and those with the highest protein concentrations were combined (final volume 3.1 ml). This pool was applied to a Superdex 75 column (Amersham Biosciences) previously equilibrated with buffer D, and 2-ml fractions were collected at a flow rate of 0.5 ml min^–1. Active fractions were combined and subjected to anion exchange chromatography on a Mono Q (HR/5R) column, eluting with a 0–0.5 M NaCl gradient in buffer E. Fractions of 0.75 ml were collected (flow rate 0.5 ml min^–1, gradient rate 5 mM min^–1 NaCl). Activity eluted between 0.2 and 0.3 M NaCl. Further purification of the QA-binding protein was achieved by chromatography on a high-resolution anion exchange chromatography system (Smart Chromatography System, Amersham Biosciences) using a Mono Q PC 1.6/5 column. For this purpose, the enzyme pool from the Mono Q (HR/5R) step was diluted with 100 mM Tris-HCl, 2 mM EDTA, 4 mM dithioerythritol to 10% glycerol final concentration, to obtain the same salt concentration as in buffer F. After application to the Mono Q PC 1.6/5 column, the enzyme was eluted with a 0–0.35 M NaCl gradient in buffer F and 100-µl fractions were collected. The activity resided in one single fraction at 185 mM NaCl. SDS-gel electrophoresis revealed the presence of two bands representing sizes of about 10 and 28 kDa.

Determination of the Amino-terminal Sequence of the Quinoxaline-2-Carboxylic Acid-binding Protein—The fraction containing the quinoxaline-2-carboxylic acid-binding protein (about 200 pmol) was incubated for 30 min at 28 °C with 2 picokatal TrsI and about 0.25 µCi of [¹⁴C]QA in the presence of 20 mM MgCl₂ and 10 mM ATP. After this, the sample was subjected to preparative SDS-gel electrophoresis and subsequently blotted to a polyvinylidene difluoride membrane. Autoradiography of the blot showed that the radioactivity stemming from QA exclusively resided in the 10-kDa band. The band was cut out and subjected to microsequencing.

Cloning of the Gene of Quinoxaline-2-Carboxylic Acid-binding Protein from S. echinatus and Expression in E. coli—Based on the aminoterminal sequence of the QA-binding protein (G/A)ATQEEIVAGLAEIVNEIAAPIVEDDQ(Q/L)(L/D), the oligonucleotide QANTERM (5'-GCCACCCAGGAGGAGATCGTCGCCGGC-3') was synthesized. After 5' labeling with [³²P]phosphate, the labeled oligo was used for Southern hybridizations with various restriction blots of S. echinatus DNA. In each restriction lane, two bands with different intensities were present. In the case of the BglII/PstI double digestion, these bands were of 3.6 and 1.8 kb. BglII/PstI-digested DNA was eluted from the respective size ranges of parallel gels and each DNA mixture was ligated to BamHI/PstI-cleaved pTZ18U (pM2PB). Transformation of E. coli DH5 and colony hybridization yielded clones for each fragment. In the case of the 1.8-kb fragment (with lower hybridization signal), sequence analysis did not reveal a sequence coding for a protein with similarity to an ACP or PCP domain, whereas sequence analysis of the 3.6-kb fragment (stronger hybridizing fragment contained in pM2PB) showed an open reading frame that comprised the previously determined amino-terminal sequence of the QA-binding protein. Furthermore, the sequence displayed nearly total identity with the amino termini of the acyl carrier proteins of fatty acid synthases from streptomycetes. The gene was named acpPSE. It was flanked by genes encoding putative 3-oxoacyl-ACP synthase III and 3-oxoacyl-ACP synthase II, respectively (deposited under GenBank^TM accession number AY825942 [GenBank] ).

The acpPSE gene encoding the QA-binding protein was engineered with suitable restriction ends for insertion into expression plasmid pQE32 to express it as an amino-terminal His₆ fusion protein. To this end the gene was amplified using forward primer FECH1 (5'-AGCGCCTGGCATGCTCGCCACTCA-3') and reverse primer RECH1 (5'-AGCCTGGATCCTACGGGATGGCTCA-3') with pM2PB as template. The resultant approximate 300-bp PCR product was digested with BamHI and SphI and ligated to SphI/BamHI-cleaved pTZ18U. Transformation of DH5 yielded pEACPN1. From pEACPN1, the engineered acpPSE was excised as a SphI/KpnI fragment and ligated to SphI/KpnI-cleaved pQE32. The resultant plasmid was designated pEACPN2. For expression of acpPSE, pEACPN2 was introduced by transformation into E. coli strain M15. A culture of M15/pEACPN2 (1.6 liters of 2x YT medium, 100 µg ml^–1 ampicillin, 25 µg ml^–1 kanamycin) was grown at 37 °C and expression was induced with 0.1 mM isopropyl 1-thio--D-galactopyranoside (final concentration) at an A₆₀₀ of 0.7 and the cells were allowed to grow for an additional 42 h at 26 °C.

Expression of acpPSE in S. lividans TK64 —To express acpPSE in S. lividans TK64 under the control of the mel promoter of pIJ702, acpPSE was engineered by PCR using forward primer GER240401A (5'-ACTCGCATGCACCACCACCACCACCACATGGCCGCCACTCAGGAAGAGATCGT-3') and reverse primer RECH1 (5'-AGCCTGGATCCTACGGGATGGCTCA-3'). The resultant 300-bp PCR band was eluted from the gel and, after digestion with SphI and BamHI, ligated to SphI/BamHI-cleaved pSP72. Transformation of E. coli DH5 gave pEACPN12. The acpPSE gene was excised from pEACPN12 by cleavage with SphI and BglII and ligated to SphI/BglII-cleaved pIJ702 (pEACPN13). pEACN13 was introduced into S. lividans TK64 and the transformants were grown for 3 days in YEME medium under the same conditions as above for pTRSA1.

Cloning and Heterologous Expression of sfp—The phosphopantetheine transferase gene sfp from Bacillus subtilis ATCC6633 was amplified and cloned into pUC19 (accession number X63158 [GenBank] ; Ref. 32). The resulting plasmid served as template for PCR with primers FSFP1 (5'-GGAGGATCTGGATCCAAGATTTACGGAA-3') and RSFP1 (5'-GCTCGGTACCCGTCGACCCCATTTATAA-3'). The 696-bp amplicon was digested with BamHI and KpnI and ligated into BamHI/KpnI-cleaved pTZ18 (pSFP1). The insert from this plasmid was removed as a BamHI/SalI fragment and cloned into BamHI/SalI-cleaved expression vector pQE30. The resulting plasmid was named pSFP30 and was used for expression of an NH₂-terminal His₆-tagged protein. Expression was achieved after transformation of E. coli M15 and a culture of M15/pSFP30 was grown at 37 °C and 200 rpm up to an A₆₀₀ of 0.6. Induction with 0.2 mM isopropyl 1-thio--D-galactopyranoside was followed by further incubation for 18 h at 30 °C (1.2 liters of 2x YT medium, 100 µg ml^–1 ampicillin, 25 µg ml^–1 kanamycin).

Purification of Recombinant TrsI from E. coli M15 Carrying pTRSN1—All operations were carried out at 0–4 °C. 16 g of cell paste were suspended in 48 ml of cold buffer NTA and passed through a French press (Aminco) at 10,000 p.s.i.. After adding MgCl₂ to give 20 mM final concentration and adding solid DNase (type I, Sigma) to give 20 µg ml^–1 final concentration, the homogenate was stirred for 45 min on ice. The suspension was centrifuged for 30 min at 25,000 x g. The supernatant was applied onto a Ni-NTA-agarose column (1.3-ml gel volume) equilibrated previously with NTA buffer. After washing with NTA-W buffer, the bound protein was eluted with increasing concentrations of imidazole in NTA buffer (containing 10, 50, and 250 mM imidazole). Fractions containing QA-activating activity were pooled (3 ml) and gel-filtered on a Superdex 75 (Amersham Biosciences) column previously equilibrated with buffer B. Fractions of 2 ml were collected and those containing the enzyme were pooled. Pooled enzyme was applied onto a -aminohexyl-Sepharose (Sigma) column (4.7 x 1 cm). After washing with buffer B, the bound protein was eluted with a 75-ml 0–0.4 M NaCl gradient (2 ml fractions). After concentration of the pooled fractions with Centricon30 microconcentrators to a protein concentration of about 1 mg ml^–1 protein concentration, the enzyme was frozen at –80 °C.

Purification of His₆AcpPSE from E. coli M15 Carrying pEACPN2— 11.3 g of cell paste of E. coli suspended in 45 ml of cold buffer NTA (with 4 mM dithioerythritol) was passed through a French press (Aminco) at 10,000 p.s.i. After addition of MgCl₂ to a 20 mM final concentration and DNase treatment with DNase I (Sigma) (20 µg ml^–1), the homogenate was stirred for 45 min on ice. The suspension was centrifuged for 30 min at 25,000 x g. To reduce the dithioerythritol concentration for the subsequent step, the supernatant was diluted with 4 volumes of buffer NTA and applied onto a Ni-NTA-agarose column (1.8 ml volume) equilibrated previously with buffer NTA. After washing with NTA-W buffer (see above), protein was eluted with increasing concentrations of imidazole in NTA buffer (containing 10, 50, and 250 mM imidazole). Fractions were analyzed by SDS-PAGE and those containing recombinant His₆AcpPSE were pooled and gel filtered on a Superdex 75 column (Amersham Biosciences) equilibrated with buffer B. Fractions of 2 ml were collected and the fractions with QA-binding activity were pooled.

Purification of Triostin Synthetases from S. echinatus—30 g of freshly harvested mycelium of S. echinatus was suspended in 75 ml of buffer A. The suspension was passed twice through a French press (Aminco) at a cell pressure of 8,000 p.s.i. MgCl₂ (20 mM final concentration) and DNase I (Sigma, 20 µg ml^–1 final concentration) were added. The suspension was stirred on ice for 30 min. The homogenate was centrifuged for 30 min at 25,000 x g. Saturated ammonium sulfate was added to the supernatant to give 60% saturation. After leaving on ice for 3 h, the precipitate was collected by centrifugation (30 min, 25,000 x g). The pellet was taken up in a minute volume of buffer B and applied onto a Ultrogel AcA34 gel filtration column (48 x 2 cm) that had been equilibrated previously with buffer B. Fractions of 6.0 ml were collected. Fractions catalyzing covalent binding to protein as thioester of either L-serine and L-alanine (triostin synthetase II) or cysteine (triostin synthetase III) or valine (triostin synthetase IV) were pooled. The various fractions were separately subjected to anion exchange chromatography, respectively, on a ResourceQ column (6 ml, Amersham Biosciences) using buffer B. After washing, enzymes were eluted with a 60-ml 0–400 mM NaCl gradient in buffer B (fraction size 2 ml). Fractions catalyzing serine/alanine thioester or valine thioester formation were pooled for analysis of thioester formation or (in the case of TrsII) for testing quinoxalinoyl peptide formation in conjunction with TrsI and AcpPSE.

Purification of Recombinant Sfp from E. coli M15 Carrying pSFP30— 6.3 g of cell paste was suspended in 19 ml of buffer NTA and passed through a French press (Aminco) at 10,000 p.s.i., followed by digestion with DNase I (Sigma) (20 µg ml^–1 final concentration and 15 mM MgCl₂) for 30 min on ice with gentle stirring. This suspension was centrifuged (30 min, 25,000 x g) and the supernatant (20 ml) was applied onto a Ni-NTA-agarose column (3 ml resin) equilibrated prior with buffer NTA. After standard wash steps as described above for the purification of recombinant TrsI, the His₆ fusion protein was eluted with 50 mM imidazole in NTA buffer. Finally a gel filtration step on the AcA54 column (Ultrogel) previously equilibrated with buffer B yielded a homogeneous 29-kDa protein.

Purification of Other Enzymes Used in this Work—ACMSI was obtained by expression of acmA in S. lividans as described previously (20). AcmACP was obtained by expression of acmD in E. coli (20). Because most of the protein was in the apo-form, conversion into the holo-form had to be performed. This was achieved by incubation of His₆AcmACP, which had been expressed in E. coli, with the 4'-phosphopantetheine transferase Sfp and CoA. After purification by gel filtration on Superdex 75, checking for intactness by MALDI-TOF mass spectrometry revealed the exclusive presence of the holo-form of protein (10641.5, theoretical value 10640.9). The hydroxypicolinic acid activating enzyme from S. griseoviridus (mikamycin synthetase I, MkmsI) and actinomycin synthetase II (ACMSII) from S. chrysomallus were prepared as described previously (33, 47).

Chemical Synthesis of Quinoxalinoyl-L-serine and Quinoxalinoyl-D-serine—To L-serine or D-serine (26.3 mg, 0.25 mmol) dissolved in 0.5 ml of 2 M NaOH was added quinoxalinoyl-2-chloride (48.2 mg, 0.25 mmol dissolved in 0.6 ml of diethyl ether) under vigorous mixing at 0 °C. After 2 min, solid KHCO₃ (60 mg, 0.6 mmol) was added gradually during a 10-min period. After stirring for an additional 2 h at room temperature, 0.5 ml of 2 M HCl was added and the mixture was applied on a reversed-phase column (SuperPac Pep-S, Amersham Biosciences). Chromatography was performed at a flow rate of 1.0 ml min^–1 with solvent A (0.1% trifluoric acid) and solvent B (acetonitril with 0.1% trifluoric acid) with the following gradient profiles: 0 min, 0% B; 25 min, 25% B; 40 min, 100% B. Fractions of 1.0 ml were collected and quinoxalinoyl L- or D-serine were identified by their different migration behavior compared with QA. Mass spectrometric analysis gave peaks for the [M–H]⁺ ions with m/z = 262 as expected for quinoxalinoyl serine (261), confirming the identity of the compounds (courtesy of Dr. Fred Rosche, Probiodrug AG, Halle, Germany).

Enzyme Assays—In the case of NRPS, 1 unit is defined as the amount of enzyme that binds 1 nmol of serine (TrsII) or threonine (ACMSII) during 30 min of incubation at 30 °C. Specific activity in the ATP-pyrophosphate exchange reaction is determined as nanokatal mg^–1 protein. 1 katal is the amount of enzyme catalyzing the incorporation of 1 mol of [³²P]pyrophosphate into ATP per second. The ATP-pyrophosphate exchange reaction dependent on the aromatic carboxylic acids was carried out as described previously (14, 20). For routine measurements of enzyme thioester formation, the standard filter binding assay was used as described previously (20). The latter assay was also used for characterization of the kinetic behavior of ArCPs such as AcpPSE and AcmACP as substrates toward adenylating enzymes TrsI, ACMSI, and MkmsI. The assays contained 150 µM [¹⁴C]QA, [¹⁴C]p-toluic acid, or [¹⁴C]picolinic acid, 10 mM ATP, 20 mM MgCl₂, purified recombinant TrsI (30 nM) or ACMSI (7.5 nM), and the respective ArCPs at concentrations varying from 1.5 to 12 µM for AcpPSE and 0.25 to 4 µM for AcmACP. To calculate K_cat/K_m, initial velocities were determined by recording time curves of ArCP loading for each concentration of ArCP. At different times, the reactions were stopped by the addition of trichloroacetic acid, proteins were collected by suction filtration on ME25 membrane filters (Schleicher & Schuell) and counted. Alternatively, precipitated protein was centrifuged, washed several times with trichloroacetic acid and ethanol, finally dissolved in 0.1 M NaOH, and counted. For testing aroyl peptide formations, TrsI (3–5 picokatal) and TrsII (0.05–0.1 units) were incubated with 0.5 µCi of [¹⁴C]L-serine, 2 mM QA or thieno[3,2-b]pyridine-5-carboxylic acid, 10 mM ATP, 20 mM MgCl₂, and AcpPSE (0.4–0.8 µM) in a total volume of 250 µl. In parallel experiments, TrsII was replaced by ACMSII (0.05–0.15 units). After incubation at 30 °C for 30 min the reaction was stopped by trichloroacetic acid precipitation. Isolation and identification of quinoxalinoyl peptide was as described above.

Data Processing—Multiple alignments and sequence analyses were performed using the Clustal W package (34).

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
Isolation and Identification of Triostin Synthetases (TRSs)— To isolate NRPS activities involved in quinoxalinoyl peptide assembly, protein fractions from S. echinatus or S. triostinicus were tested for protein-thioester formation with the amino acids serine, alanine, cysteine, and valine. These amino acids occur in the peptide chains of triostin A or echinomycin. Fractionated concentration by ammonium sulfate precipitation of broken cell extracts of the streptomycetes and subsequent gel filtration on Ultrogel AcA34 revealed covalent binding of alanine and serine in the fractionation range of M_r 280,000–230,000. Valine activation was found in a broad range of M_r 320,000 and 100,000 with two maxima at M_r 280,000 and 130,000 as determined from the calibration curve of the Ultrogel AcA34 columns (results not shown). The M_r 280,000 maximum did not reproducibly appear in preparations, whereas the smaller M_r 130,000 species did. This indicated that the smaller enzyme might have arisen from the larger by proteolysis. Anion exchange chromatography of protein taken from the fractionation range 300,000–200,000 of the Ultrogel AcA34 column is shown in Fig. 3 (panel A). There was a peak of serine activation that coincided with alanine activation (the latter not shown). Valine activation was represented by an enzyme peak eluting at higher salt strength. This showed that the serine/alanine activating enzyme is distinct from the one activating valine.

View larger version (45K): [in this window] [in a new window]   FIG. 3. Separation and characterization of serine/alanine and valine activating NRPS subunits TrsII and TrsIV of S. echinatus by anion exchange chromatography. A, anion exchange chromatography of the high Mr fraction of gel filtration fractionation of S. echinatus on ResourceQ. Shown are the serine binding activity (•) and valine binding activity () of the triostin synthetases II and IV (TrsII and IV). Dotted line, protein. The serine binding activity was concomitantly associated with alanine binding activity (not shown here), which indicated that the protein harbors both modules for the first two amino acids of the quinoxalinoyl tetrapeptide chain. B, analysis of the serine- and alanine-thioester complex: lane a, addition of formic acid to serine-thioester; b, addition of performic acid to serine-thioester; c, addition of formic acid to alanine-thioester; d, addition of performic acid to alanine-thioester. 100 µl of (•) peak fraction was incubated with 1 µCi of 14C-labeled alanine or serine in the presence of MgATP for 30 min at 30 °C. Enzyme thioester was precipitated with trichloroacetic acid and after washing was subjected to analysis. C, analysis of the valine-thioester complex: lane a, formic acid to valine-thioester; b, performic acid to valinethioester; c, performic acid to valine-thioester formed in the additional presence of S-adenosyl-L-methionine. 100 µl of enzyme fraction of the () peak fraction were incubated with 1 µCi of [14C]valine, MgATP for 30 min at 30 °C with and without 1 mM S-adenosyl-L-methionine.

Detection of activity catalyzing cysteine-protein thioester in S. echinatus protein fractions was hampered by the nonspecific binding of [³⁵S]cysteine to proteins, presumably via disulfide bonds (not shown). Applying fractions from Ultrogel AcA34 gel filtrations of S. echinatus protein extracts incubated with [³⁵S]cysteine and ATP onto SDS-PAGE gels with subsequent autoradiography and inspection of the various lanes for ATP-dependent radiolabeled bands revealed a 250-kDa protein that bound [³⁵S]cysteine in an ATP-dependent manner (Fig. 4). The protein resided in the fractionation M_r range of 230,000–300,000 of the Ultrogel AcA34 gel filtration column. In aging protein preparations the 250-kDa band gradually converted into a 120-kDa band, indicating proteolysis. Nevertheless, still in this case the native activity still resided in the M_r 230,000–280,000 fractionation range of the column, suggesting that the proteolytic fragment still held together with other fragments of the protein or protein complex. The identification of the cysteine binding activity prompted a similar analysis for the serine, alanine, and valine activating proteins. Fig. 4 shows, in fact, that [¹⁴C]serine and [¹⁴C]alanine both labeled the same 245-kDa band. This indicated that the protein is bimodular. It was designated triostin synthetase II (TrsII). Inspection of the valine peak from the gel filtrations did not show the expected band of more than 200 kDa. Instead, the enzyme showed labeled bands of 110 and 80 kDa, which were unreproducible in their abundance (not shown). This enzyme was difficult to handle because of its susceptibility to proteolysis, which could not be inhibited. The cysteine activating activity was designated triostin synthetase III (TrsIII) and the valine-activating protein was designated triostin synthetase IV. Whether TrsIII and -IV are proteolytic fragments of a larger enzyme could not be determined. TrsII, with a size of 245 kDa, is in the same size range as its counterpart, ACMSII, from the biosynthesis of actinomycin, which has a molecular mass of 280 kDa (25) and therefore was regarded as an intact species. TrsII never changed its size in other separation steps and always stayed as a 245-kDa band that could be labeled with [¹⁴C]serine and [¹⁴C]alanine.

View larger version (31K): [in this window] [in a new window]   FIG. 4. Size determinations of serine/alanine and cysteine activating NRPS subunits TrsII and TrsIII of S. echinatus. Shown are autoradiographs of SDS-PAGE separations of loading reactions involving TrsII or TrsIII with [14C]serine, [14C]alanine, and [35S]cysteine, respectively. The fractions used for these loading reactions were from Ultrogel AcA34 gel filtrations using a column calibrated with ferritin (450 kDa), catalase (250 kDa), and aldolase (154 kDa). The indicated Ve/Vo ratios of the NRPS represent the peak maxima of binding activities for serine/valine or cysteine. The left inset shows the labeled band of TrsII after incubation with [14C]serine or [14C]alanine. The right inset shows the ATP-dependent labeling of the band representing TrsIII with [35S]cysteine. Sizes of the denatured bands are 245 and 250 kDa, respectively. The incubations were carried out as described in the legend to Fig. 3 except that the reaction was stopped by the addition of acetone and the precipitate after dissolution in buffer B with 2% SDS was applied to SDS-PAGE gels. The 350-kDa marker in the marker lanes is enniatin synthetase.

View larger version (22K): [in this window] [in a new window]   FIG. 5. Substrate specificity determining residues in the active site adenylate binding pocket of TrsI. Alignment of amino acid residues in positions of the substrate binding pockets of various aromatic carboxylic acid activating enzymes (A-domains). The residues were extracted from A-domain sequences according to Refs. 38 and 39 on the basis of the crystal structure of gramicidin S synthetase 1 A-domain complexed with phenylalanine-adenylate. The following are shown in the figure: TrsI, this work, SnbA (29), EntE (30), YbtE (37), ACMSI (20).

Expression of trsA—To further characterize the trsA gene product, its gene was expressed as a His₆ fusion protein in E. coli M15 or S. lividans TK64. Cloning in pQE32 or pQE70 with subsequent expression in E. coli afforded both the amino-terminal and carboxyl-terminal His₆-tagged versions of the protein in active form (as tested by the QA-dependent ATP-pyrophosphate exchange). By contrast, in S. lividans expression gave active protein, also, but with loss of the carboxyl-terminal His₆ tag. The amino-terminal His₆ fusion from E. coli was used throughout this study. Purification to homogeneity of the protein from E. coli M15 harboring pTRSN1 was achieved by Ni-NTA affinity chromatography, Superdex 75 gel filtration, and chromatography on -aminohexyl-Sepharose (Fig. 6). The purification from the eluate of the Ni-NTA column was 4.87-fold with an overall yield of 78% (1.4 mg liter^–1 culture). The specific activity of the homogeneous enzyme was 7.8 nanokatal mg^–1 protein, which drastically contrasts the specific activity of the same enzyme obtained from S. triostinicus, which was determined to be 1.1 nanokatal mg^–1. This can be explained, at least in part, by the fact that purification of enzyme from the streptomycete requires seven purification steps compared with two with the His₆-tagged protein, suggesting that during purification from the streptomycete a considerable portion of enzyme might have been inactivated. Like wild type enzyme, the recombinant enzyme adenylated all of the structural analogues of QA with the same relative activity compared with QA (14).

View larger version (116K): [in this window] [in a new window]   FIG. 6. Purification of amino terminal His6-tagged triostin synthetase I. Shown are SDS-PAGE separations of individual purification steps. Lane M, Mr markers; 1, crude extract; 2, Ni-NTA affinity chromatography; 3, Superdex 75 gel filtration; 4, -aminohexyl-Sepharose anion exchange chromatography.

The gene of the QA-binding protein was cloned by screening various restriction blots of chromosomal DNA of S. echinatus with an oligonucleotide probe derived from part of the amino-terminal sequence of the binding protein. Sequencing of the hybridizing 3.6-kb BglII/PstI fragment revealed in its middle an open reading frame of 249 bp encoding a protein of 8.91 kDa. The gene had a G + C content of 62.3%, which increases to 90.4% in the third codon positions. The derived amino acid sequence perfectly fitted the determined amino-terminal sequence of the QA-binding protein, confirming the correctness of the cloned sequence. Alignment of the protein sequence with known sequences in the data base revealed that as had already been obvious from the amino-terminal sequence, the protein was nearly identical over its whole length with FAS acyl carrier proteins from S. coelicolor (98% identity), S. glaucescens (98% identity), and S. avermitilis (96% identity). This suggested that this gene encodes the fatty acyl carrier protein of FAS in S. echinatus, which of course clearly explained why the protein was much more abundant in S. echinatus than the peptide synthetase TrsII.

Further proof for the identity of the binding protein with FAS ACP came from analysis of its flanking regions. The gene 5'-flanking to the acyl carrier protein gene was found to be identical to the gene coding for 3-oxoacyl-[acyl carrier protein] synthase III (FabH, SCO2388) of S. coelicolor. The gene 3' to the acyl carrier protein was identical with the gene encoding 3-oxoacyl-[acyl carrier protein] synthase II (FabF, SCO2390) of S. coelicolor. In S coelicolor, these two genes directly flank the acyl carrier protein gene (AcpP, SCO2389) as in S. echinatus with the same orientations (Fig. 7). The same positioning is present in S. avermitilis (42) and S. glaucescens (43). Therefore, we conclude that the QA-binding protein is the ACP of fatty acid synthase of S. echinatus. The gene and protein were designated acpPSE and AcpPSE, respectively. Testing the ACPs of FAS of other streptomycetes not producing quinoxaline antibiotics such as S. coelicolor M145, S. lividans TK64, S. griseoviridus DSM40229 or S. chrysomallus ATCC11523 in cell extracts from these strains supplemented with purified TrsI revealed in all cases QA binding in comparable specific activities as in S. echinatus (not shown). Binding of QA was not observed in these extracts when TrsI was not added, which leaves no doubt that Streptomyces FAS ACP is a specific substrate of TrsI. Repeated attempts using different experimental approaches to find other QA-binding proteins besides AcpPSE in S. echinatus or S. triostinicus always ended in the isolation of the same protein, AcpPSE. Moreover, hybridization analysis of S. echinatus DNA revealed that the gene acpPSE was present in one single copy.

View larger version (16K): [in this window] [in a new window]   FIG. 7. Fatty acid synthase genes in various streptomycete genomes and location of acpPSE on the chromosome of S. echinatus.

View larger version (82K): [in this window] [in a new window]   FIG. 8. Loading of recombinant AcpPSE with [14C]QA. Purified amino-terminal His6 fusion of AcpPSE on a 12.5% SDS-polyacrylamide gel (lane 1). Incubation of protein with TrsI, [14C]QA, and MgATP with subsequent SDS-PAGE and autoradiography reveals exclusive labeling of the monomer (lane 1'), whereas the AcpPSE dimer (moving as a about 28-kDa band) is not labeled. M denotes the markers and M' the location of that lane on the autoradiogram.

View larger version (11K): [in this window] [in a new window]   FIG. 9. Kinetic analysis of the initiation modules TrsI/AcpPSE and ACMSI/AcmACP. Time courses of loading of AcpPSE () and AcmACP (•) by TrsI with [14C]QA is shown in the inset of panel A. Time courses of loading AcmACP (•) and AcpPSE () by ACMSI with [14C]p-toluic acid is shown in the inset of panel B. Velocity-substrate curves for TrsI/AcpPSE (panel A ()) and ACMSI/AcmACP (panel B (•)) were obtained by measuring initial velocities of carrier protein loading reactions at various carrier protein concentrations with fixed total concentration [Et] of the A-domain present. Values for kcat/Km derived from initial velocities were 13.5 and 3.5 µM–1 min–1 for TrsI/AcpPSE and ACMSI/AcmACP, respectively. Concentrations of adenylating enzymes were 30 (TrsI) and 7.5 nM (ACMSI). Note that the substrate p-toluic acid is not the natural substrate for ACMSI. With natural substrate, higher kcat/Km values are expected based on the results of total cell-free synthesis studies.2

View larger version (24K): [in this window] [in a new window]   FIG. 10. Synthesis of quinoxalinoyl-D-serine catalyzed by TrsI and TrsII in conjunction with AcpPSE. The various lanes show autoradiographs of TLC separations of reaction products formed in the presence of TrsI (15 nM), TrsII (0.05 units), and AcpPSE (0.6 µM) bound as thioester to TrsII. The incubations were with 0.5 µCi of [14C]L-serine, 2mM QA or TPA, 10 mM ATP, and 20 mM MgCl2 in a total volume of 250 µl (buffer B). After incubation, protein was precipitated by addition of 7% (w/v) trichloroacetic acid. Washing of the precipitate was with 7% trichloroacetic acid and finally with ethanol. Reaction products were released from protein by saponification with NaOH. After acidification, product was extracted with ethyl acetate and subjected to TLC. Panel A, formation of quinoxalinoyl serine (compound denoted a) dependent on the presence of the indicated components. Panel B shows formation of thieno[3,2-b]pyridine-5-carboxyloyl serine (denoted b) with TPA replacing QA. Panel C shows the separation on a chiral plate of an acid hydrolysate of the enzymatically formed radioactive quinoxalinoyl serine (a), which yields back [14C]serine. The major portion of the serine has the D-configuration. Solvent systems were III (panels A and B) and IV (panel C).

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

The QA tetrapeptide chains of the quinoxaline antibiotics from their structures resemble the aroyl peptide chains of other chromopeptide lactones from streptomycetes such as the actinomycins or the mikamycin B II group of antibiotics, which points to similar organizations of the catalytic domains of their assembly systems (21, 29). In fact, TrsI from S. triostinicus and S. echinatus, which activate QA as adenylate, is similar in size and sequence to its counterparts in the syntheses of actinomycin (ACMSI), pristinamycin I (SnbA), and etamycin (MkmsI) (33, 44). It also has significant similarity to the various aryl carboxylic acid activating A-domains from the biosynthetic systems for the bacterial catechol/hydroxyphenyl-capped siderophore peptides such as enterobactin (EntE), yersiniabactin (YbtE), or vibriobactin (VibE) (17). Analysis of the specificity determining amino acid residues in the putative substrate-binding pocket of TrsI derived by comparison with the binding pocket structure of gramicidin synthetase I (Fig. 5) indicated the highest similarity with the binding pocket of SnbA, which activates 3-hydroxypicolinic acid, the chromophore of pristinamycin I (29). Although TrsI does not activate 3-hydroxypicolinic acid, the similarity agrees with the finding that it prefers quinoline-derived carboxylic acids as substrates where the ring nitrogen (like in 3-hydroxypicolinic acid) is proximal to the carboxyl group (14). Both TrsI and MkmsI (the SnbA orthologue) from etamycin-producing S. griseoviridus, albeit weakly, activate the compound pyrazine-2-carboxylic acid, which may indirectly confirm the similarity seen in the substrate binding pockets of TrsI and SnbA.²

All of the free-standing A-domains that activate aromatic carboxylic acids in the biosynthesis of the various catechol- or hydroxyphenyl-capped peptide siderophores acylate specific carrier proteins (ArCPs) (17). These are similar to the PCP domains of NRPS and are fused to isochorismate lyase, an enzyme involved in the synthesis of the aromatic carboxylic acid 2,3-dihdroxybenzoic acid used as starter residue in these syntheses. In other cases, such as in the synthesis systems of yersiniabactin or pyochelin, the carrier domain is fused amino-terminal to the NRPS lying downstream in the biosynthetic sequence, HMWP2 or PchE, respectively (17, 19, 37). By contrast, in the case of quinoxaline antibiotic biosynthesis described here, a stand-alone ArCP domain has been identified as the binding protein for QA. This protein, AcpPSE, was identified as the fatty acyl ACP of S. echinatus by comparison of its sequence to those of FAS ACPs of S. coelicolor, S. avermitilis, or S. glaucescens. Further confirmation came from the analysis of the 5'- and 3'-flanking regions of acpPSE, which contain genes encoding putative 3-oxoacyl-ACP synthase III and 3-oxoacyl-ACP synthase II enzymes, both components of the fatty acid synthase complex. Analysis of AcpPSE using the A-domains TrsI, ACMSI, and MkmsI showed exclusive acylation by TrsI with QA (Fig. 8), suggesting intimate functional linkage between the two proteins. The k_cat/K_m values obtained from the loading reactions were for the pairs TrsI/AcpPSE and ACMS I/AcmACP between 5 and 15 µM^–1 min^–1 and therefore comparable with data obtained with the pairs EntE/EntB (18) or VibE/VibB (45), which were interpreted as the result of specific protein-protein interactions. Clearly, these values were several orders higher than those obtained when measuring acylation of coenzyme A or pantetheine with aryl carboxylic acids, which may possibly be nonenzymatic trapping of acyladenylate by these highly reactive sulfhydryl compounds (45, 49).

A further contribution to proving the involvement of AcpPSE in triostin synthesis was the identification of the NRPS modules for QA tetrapeptide assembly (Fig. 11). The data show that the triostin assembly line is analogous to those for actinomycin (Fig. 2) and pristinamycin I, indicating an evolutionary relationship in the synthetic strategies for formation of these compounds although quite different in their final structures. TrsII, as the enzyme carrying the first two modules (Fig. 4), has its counterparts in ACMSII of actinomycin or SnbC of pristinamycin I biosy