Identification of Middle Chain Fatty Acyl-CoA Ligase Responsible for the Biosynthesis of 2-Alkylmalonyl-CoAs for Polyketide Extender Unit*

Background: Fatty acyl-CoA ligases involved in polyketide biosynthesis remain uncharacterized. Results: RevS classified in fatty acyl-AMP ligase clade was the middle chain fatty acyl-CoA ligase. Conclusion: RevS was responsible for 2-alkylmalonyl-CoA biosynthesis through enzyme coupling with RevT carboxylase/reductase. Significance: 2-Alkylmalonyl-CoA biosynthesis was strongly supported by the function of RevR and RevS, which utilized fatty acids derived from de novo biosynthesis and degradation products, respectively. Understanding the biosynthetic mechanism of the atypical polyketide extender unit is important for the development of bioactive natural products. Reveromycin (RM) derivatives produced by Streptomyces sp. SN-593 possess several aliphatic extender units. Here, we studied the molecular basis of 2-alkylmalonyl-CoA formation by analyzing the revR and revS genes, which form a transcriptional unit with the revT gene, a crotonyl-CoA carboxylase/reductase homolog. We mainly focused on the uncharacterized adenylate-forming enzyme (RevS). revS gene disruption resulted in the reduction of all RM derivatives, whereas reintroduction of the gene restored the yield of RMs. Although RevS was classified in the fatty acyl-AMP ligase clade based on phylogenetic analysis, biochemical characterization revealed that the enzyme catalyzed the middle chain fatty acyl-CoA ligase (FACL) but not the fatty acyl-AMP ligase activity, suggesting the molecular evolution for acyl-CoA biosynthesis. Moreover, we examined the in vitro conversion of fatty acid into 2-alkylmalonyl-CoA using purified RevS and RevT. The coupling reaction showed efficient conversion of hexenoic acid into butylmalonyl-CoA. RevS efficiently catalyzed C8–C10 middle chain FACL activity; therefore, we speculated that the acyl-CoA precursor was truncated via β-oxidation and converted into (E)-2-enoyl-CoA, a RevT substrate. To determine whether the β-oxidation process is involved between the RevS and RevT reaction, we performed the feeding experiment using [1,2,3,4-13C]octanoic acid. 13C NMR analysis clearly demonstrated incorporation of the [3,4-13C]octanoic acid moiety into the structure of RM-A. Our results provide insight into the role of uncharacterized RevS homologs that may catalyze middle chain FACL to produce a unique polyketide extender unit.

Microorganisms harbor structurally diverse natural products, including polyketides, peptides, and terpenoids (1). Their secondary metabolites with strong biological activity have been utilized as antibiotics, antitumor drugs, immunosuppressants, hypercholesterolemia drugs, and insecticides (2). Unique chemical structures are linked to a variety of biological activities; therefore, understanding the biosynthetic machinery is important for future combinatorial biosynthesis (3)(4)(5).
Fatty acyl chains found in microbial natural products are responsible for biological activity and physiological function in microorganisms. For instance, daptomycin binds to the cell membranes of Gram-positive bacteria through its lipid moiety, followed by calcium-dependent insertion and oligomerization (6). Dimerization of alkyl resorcinol results in the generation of potent proteasome inhibitors, cylindrocyclophanes (7). The fatty acyl chain of mycolic acid plays an important role in membrane construction in Mycobacterium tuberculosis (8).
FAAL, 2 a member of the adenylate-forming enzyme superfamily (9), catalyzes ATP-dependent formation of acyl-AMP and loading onto the phosphopantetheine arm of the acyl carrier protein (ACP). FAALs are associated with the polyketide synthase (PKS) and nonribosomal peptide synthetase gene clusters, and the resulting fatty acyl-ACP is incorporated into the starter unit of the PKS and nonribosomal peptide synthetase assembly line (7, 10 -21). In contrast to FAAL, fatty acyl-* This work was supported in part by Japan Society for the Promotion of Laboratories, Inc. (Cambridge, MA). All commercially available chemicals for chemical synthesis were used without further purification.
Analytical Methods-The 1 H and 13 C NMR spectra of compound 1 and CoA derivatives were recorded on JEOL JNM-ECA-500 and JNM-AL-400 spectrometers (JEOL, Ltd., Tokyo, Japan). Chemical shifts (in ppm) were referenced against the residual solvent for 1 H NMR and 3-(trimethylsilyl)propionic-2,2,3,3d 4 acid sodium salt (TSP-d 4 ) as an external standard for 13 C NMR. UV absorbance was measured using a Jasco V-630 BIO spectrophotometer (Jasco Co., Tokyo, Japan). Fast atom bombardment-mass spectra were obtained using a JMS-700 mass spectrometer. Electrospray ionization (ESI) mass spectra of CoA derivatives were obtained on a BioApex-II Fourier transform ion cyclotron resonance mass spectrometer (Bruker Daltonics Japan, Ltd., Tokyo, Japan) and Synapt G2 (Waters, MA). ESI-MS analysis of RM derivatives and RevS and RevT reaction products were performed using a Waters Alliance HPLC system equipped with a mass spectrometer (Q-TRAP, Applied Biosystems, CA).
All chemical syntheses were carried out under a nitrogen atmosphere and monitored by TLC with 0.25-mm pre-coated Silica Gel plates 60F254 Art 5715 (Merck, Darmstadt, Germany). Visualization was achieved using UV light and 10% ethanol solution of phosphomolybdic acid, followed by heating. For column chromatography, Silica Gel 60 N (spherical, neutral, 0.04 -0.05 mm; Kanto Chemical Co, Inc., Tokyo, Japan) and Cosmosil 75C18-DNP (Nacalai Tesque Inc., Kyoto, Japan) were utilized.
Plasmid Construction for Gene Disruption and Complementation-The revR, revS, and revT gene disruptions were performed by PCR-targeted gene replacement using the plasmids pKD46, pKD13, and pCP20 and E. coli BW25113 (41,42). Ampicillin-resistant pKD46 containing the Red-mediated recombination functions was used with the chloramphenicolresistant pCC1FOS fosmid clone 3C11 containing the revR, revS, and revT genes. The plasmid pKD13 was used as a template for the FRT-flanked kanamycin-resistant gene cassette (42,43).
To construct the revR, revS, and revT gene replacement plasmids, DNA fragments containing the disrupted gene were amplified by PCR using PrimeSTARHS DNA polymerase (Takara Bio). PCR conditions were as follows: 98°C for 10 s, 25 cycles of 98°C for 10 s, 62°C for 5 s, and 68°C for 5 min. The amplified fragments were ligated to the HindIII site of the pIM vector (38). For Southern analysis (Fig. 2), the AlkPhos Direct Labeling and Detection System (GE Healthcare, Little Chalfont, UK) was used. The primers used for gene disruptions and Southern analysis are shown in Table 1.
To obtain the gene complementation plasmid, the revR, revS, and revT genes were ligated into the BamHI and HindIII sites of pTYM19 with an aphII promoter (38,44). Because the revR gene includes a BamHI site in the sequence, a silent mutation was introduced using the QuikChange protocol (Stratagene, La Jolla, CA). The revR gene inserted into the NdeI/XhoI restriction sites of the pET28b vector was amplified using PfuTurbo DNA polymerase (Stratagene) and the following primer pairs (with the mutation sites underlined): 5Ј-TCCAGCGAC-GAGTGGATACGCCGGCACTCCGGGAT-3Ј and 5Ј-ATC-CCGGAGTGCCGGCGTATCCACTCGTCGCTGGA-3Ј. PCR conditions were as follows: 98°C for 30 s, 20 cycles of 98°C for 30 s, 60°C for 1 min, and 68°C for 13 min. The resultant pET28b-revR-BamHI_mut vector was used as a template for the complementation plasmid (Table 1).
Plasmid Construction for Heterologous Gene Expression in E. coli-The revS (1740 bp) and revT (1332 bp) genes were amplified from the fosmid clone (PCC1FOS-3C11) using PrimeSTARHS DNA polymerase under the following conditions: 98°C for 10 s and 25 cycles of 98°C for 10 s, 62°C for 5 s, and 68°C for 1.5 min. The SRE2849 gene (243 bp) coding ACP of Streptomyces sp. SN-593 was amplified from PCC1FOS-4B1 under the following conditions: 98°C for 10 s and 25 cycles of 98°C for 10 s, 60°C for 5 s, and 68°C for 25 s. The Bacillus subtilis 168 phosphopantetheinyl transferase gene (sfp) (654 bp) was amplified from pUC19-sfp under the following conditions: 98°C for 10 s and 25 cycles of 98°C for 10 s, 62°C for 5 s, and 68°C for 30 s (45). The primers used for amplification are shown in Table 1. After restriction enzyme digestion, the revS, revT, SRE2849, and sfp gene fragments were inserted into the NdeI and XhoI sites of pET28b(ϩ) or pACYCDuet-1 to construct pET28b-revS, pET28b-revT, pET28b-SRE2849, and pACYCDuet-1-sfp, respectively.
Plasmid Construction for Heterologous Gene Expression in S. lividans TK23-The ␤-lactamase gene in the pWHM3 vector was replaced with the aphI gene using Red recombination to form pWK. Next, the tipA promoter (P tipA ) was inserted into the EcoRI and BamHI sites to construct pWK-P tipA for expres-sion in S. lividans TK23. To facilitate enzyme purification, the revT gene containing a His 8 tag coding sequence was amplified from pET28b-revT using the primers shown in Table 1. After restriction enzyme digestion, the fragment was inserted into the BamHI and HindIII sites of pWK-P tipA to construct pWK-P tipA -revT.
Heterologous Gene Expression and Purification of Enzymes-To obtain recombinant RevS, the expression plasmid was transformed into E. coli BL21 Star TM (DE3) cells. The resultant transformants were grown on LB medium containing 50 g ml Ϫ1 kanamycin overnight. To 200 ml of TB containing 50 g ml Ϫ1 kanamycin, the preculture was added at an initial absorbance of 0.1 at 600 nm (A 600 ). Cells were cultured at 18°C. When the A 600 reached 0.6, isopropyl ␤-D-1-thiogalactopyranoside was added to a final concentration of 0.5 mM. After further growth for 18 h at 18°C, the cells were harvested by centrifugation and frozen at Ϫ80°C.
After thawing on ice, the cells were suspended in 20 ml of extraction buffer (50 mM Tris-HCl, pH 7.5, 500 mM NaCl, and 20% glycerol, 5 mM imidazole, 0.5 mg ml Ϫ1 lysozyme, and 6.25 units ml Ϫ1 benzonase (Sigma)). For stabilization of recombinant RevS, octanoic acid was added to the cell suspension at a final concentration of 1 mM. The cell suspension was sonicated 10 times on ice for 10 s with 1-min interval between each sonication treatment (UD-200, TOMY, Tokyo, Japan). Cell debris was removed by centrifugation (10,000 ϫ g for 30 min) (SRX-201, TOMY), and then the supernatant was applied to a nickelnitrilotriacetic acid (Ni-NTA)-agarose (Qiagen, Hilden, Germany) column (2 ϫ 4 cm) that had been equilibrated with buffer A (50 mM Tris-HCl, pH 7.5, 500 mM NaCl, and 20% glycerol) containing 5 mM imidazole. After washing with the same buffer, non-specifically bound proteins were removed by washing with 20 ml of buffer A containing 5 mM imidazole and 0.2% Tween 20. After removing Tween 20 with 40 ml of buffer A containing 5 mM imidazole, the column was further washed with 40 ml of buffer A containing 40 mM imidazole. The His tag fusion protein was eluted with 25 ml of buffer A containing 250 mM imidazole. The eluted fraction was concentrated using an Amicon ultracentrifugal filter (Merck, Darmstadt, Germany), and the buffer was exchanged for stock buffer (50 mM Tris-HCl, pH 7.5, 200 mM NaCl, 20% glycerol) on the centrifugal filter. The purity of RevS was confirmed by SDS-PAGE. Size exclusion chromatography was performed as described previously (46).
To obtain recombinant RevT, pWK-P tipA -revT was introduced into S. lividans TK23 using a polyethylene glycol-mediated protoplast method (47). The resulting transformants were grown in SK2 medium containing 15 g ml Ϫ1 kanamycin for 2 days, and then 1 ml of preculture was inoculated into 70 ml of SK2 medium containing 15 g ml Ϫ1 kanamycin. After 1 day of culture, thiostrepton was added at a final concentration of 50 g ml Ϫ1 to express RevT. After another 1 day of culture, the cells were harvested by centrifugation and frozen at Ϫ80°C. Purification of RevT was performed as described for RevS. The His tag fusion protein was eluted with 5 ml of buffer A containing 250 mM imidazole and 1 mM NADPH and used for biochemical characterization. The purity of RevT was confirmed by SDS-PAGE. Size exclusion chromatography was performed using 50 mM Tris-HCl, pH 7.5, 10 mM NaHCO 3 , 200 mM NaCl, and 10% glycerol.
To obtain apo-ACP, pET28b-SRE2849 was transformed into E. coli BL21 Star TM (DE3) cells. The transformants were selected on LB medium containing 50 g ml Ϫ1 kanamycin. Heterologous expression and purification were performed as described for RevS. The fraction eluted from the Ni-NTA column was concentrated using an Amicon ultracentrifugal filter, and the buffer was exchanged for stock buffer.
To obtain holo-ACP, pET28b-SRE2849 and pACYCDuet-1sfp were co-transformed into E. coli BL21 Star TM (DE3) cells. Transformants were selected on LB medium containing 50 g ml Ϫ1 kanamycin and 30 g ml Ϫ1 chloramphenicol. Heterologous gene expression, Ni-NTA column chromatography, and buffer exchange were performed as described for apo-ACP. To eliminate ␣-N-gluconoylated holo-ACP (48) from the Ni-NTA fraction, holo-ACP was further purified using a Mono Q 5/50 column (GE Healthcare) that had been pre-equilibrated with 50  NOVEMBER 6, 2015 • VOLUME 290 • NUMBER 45 mM BisTris, pH 6.5, containing 0.2 M NaCl and 20% glycerol. After loading the Ni-NTA fraction (42.2 mg), the column was washed with equilibration buffer for 30 min at a flow rate of 0.2 ml min Ϫ1 and eluted with a linear gradient of 0.2-0.5 M NaCl for 60 min. Subsequently, 400 l of the Mono Q fraction (5.7 mg) containing holo-ACP was collected based on MALDI-TOF/MS analysis and used in the FAAL assay for RevS.

Middle Chain FACL Involved in RM-A Biosynthesis
FACL Assay of RevS-For HPLC analysis of the RevS reaction product, an enzyme reaction was performed in a final reaction volume of 100 l containing 100 mM HEPES, pH 7.5, 20% glycerol, 10 mM MgCl 2 , 5 mM TCEP, 0.5 mM fatty acid, 2 mM CoA, 3 mM ATP, and 1 M RevS. After preincubation of the reaction mixture at 25°C for 3 min, the reaction was initiated by the addition of RevS and allowed to proceed for 60 min. The reaction was terminated by the rapid addition of 0.5 l of formic acid and then centrifuged at 15,000 ϫ g for 10 min at 4°C (Kubota 3700, Kubota Co., Tokyo, Japan) to remove the protein. The supernatant was subjected to liquid chromatography (LC)/ESI-MS (Table 2).
To determine the substrate specificity of RevS, we quantified fatty acyl-CoA formation by HPLC and NADH oxidation using a spectrophotometer. Because the stoichiometry of the product amount in both assays was identical, we performed a spectrophotometric assay to determine the kinetic parameters of RevS After preincubation of the reaction mixture at 25°C for 3 min, the reaction was initiated by adding RevS, and the initial rate of oxidation of NADH at 340 nm was measured using a spectrophotometer. The kinetic constant was calculated by a nonlinear regression fit to the Michaelis-Menten equation using Sigma-Plot11 ( Table 3).

Summary of RevS reaction products
RevS reaction products were analyzed by LC/ESI-MS as described under "Experimental Procedures" All mass spectra were collected in the ESI-negative mode.
Enzyme Assay of RevT-To detect the RevT reaction product, the assay was performed in a final volume of 100 l containing 100 mM Tris-HCl, pH 7.5, 10% glycerol, 10 mM NaHCO 3 , 11 mM MgCl 2 , 1 mM EDTA, 0.5 mM (E)-2-enoyl-CoA, 4 mM NADPH, and 1 M RevT. After preincubation of the reaction mixture at 25°C for 3 min, the reaction was initiated by the addition of RevT and allowed to proceed for 10 -30 min. The reaction was terminated by the rapid addition of 0.5 l of formic acid and centrifuged at 15,000 ϫ g for 10 min at 4°C to remove the protein. The supernatant was subjected to LC/ESI-MS analysis.
To determine the kinetic parameters of RevT, we quantified both butylmalonyl-CoA formation by HPLC and NADPH oxidation using a spectrophotometer. Because the stoichiometry of the reaction product in both assays was identical, we conducted a spectrophotometric assay. The assay was performed in a 1-ml quartz cuvette in a final volume of 200 l containing 100 mM Tris-HCl, pH 7.5, 10% glycerol, 10 mM NaHCO 3 , 11 mM MgCl 2 , 1 mM EDTA, 0.5 mM (E)-2-enoyl-CoA, 0.3 mM NADPH, and 1 M RevT. NADPH concentration was varied from 0.01-0.3 mM. The substrate concentration for (E)-2-enoyl-CoA varied from 0.1-2 mM. After preincubation of the reaction mixture at 25°C for 3 min, the reaction was initiated by adding RevT, and NADPH oxidation at 340 nm was measured using a spectrophotometer. The kinetic constant was calculated by a nonlinear regression fit to the Michaelis-Menten equation using SigmaPlot11 (Table 4).
In Vitro Reconstruction of Butylmalonyl-CoA-An enzyme coupling assay was performed for 1 h at 25°C in 100 mM Tris-HCl, pH 7.5, buffer containing 20% glycerol, 10 mM NaHCO 3 , 11 mM MgCl 2 , 1 mM EDTA, 5 mM TCEP, 0.5 mM (E)-2-hexenoic acid, 2 mM CoA, 3 mM ATP, 4 mM NADPH, 1 M RevS, and 1 M RevT in a final volume of 100 l. The reaction was terminated by the rapid addition of 0.5 l of formic acid and centrifuged at 15,000 ϫ g for 10 min at 4°C to remove the protein.
The supernatant was subjected to LC/ESI-MS analysis.
LC/ESI-MS Analysis of Enzyme Reaction Products-To analyze the reaction products of RevS, an LC/ESI-MS analysis was carried out. An Applied Biosystems Q-TRAP was connected to a Waters Alliance 2965 with a 2996 photodiode array detector and an XTerraMS C 18 column (5 m, 2.1 ϫ 150 mm (Waters)). The HPLC conditions were as follows: 0.25 ml min Ϫ1 flow rate; solvent A, water containing 40 mM ammonium acetate (pH 6.8); solvent B, acetonitrile. The conditions of the linear gradient elution were optimized to detect each RevS reaction product. The column was equilibrated with 2% solvent B. After injection of 1 l of the sample into the equilibrated column, the column was developed using a linear gradient from 2 to 20% for saturated acyl-CoA, 2 to 50% for (E)-2-enoyl-CoA, and 2 to 65% acetonitrile for (E)-3-enoyl-CoA over 45 min. After the gradient, the column was washed with 100% solvent B for 10 min. All mass spectra were collected in the ESI-negative mode ( Table 2).
To analyze the reaction products of RevT, LC/ESI-MS analysis was carried out. The HPLC conditions were as follows: 5 m (2.1 ϫ 150 mm) column, XTerraMS C 18 ; 0.25 ml min Ϫ1 flow rate; solvent A, water containing 40 mM ammonium acetate (pH 5); solvent B, acetonitrile. After injection of a 1-l sample into the column that had been equilibrated with 2% solvent B, the column was developed over a linear gradient of 2-25% solvent B for 30 min and washed with 100% solvent B for 10 min. Mass spectra were collected in the ESI-negative mode.

Feeding of Labeled Precursors and Isolation of Compound 1-
The ⌬revR mutants were grown in 70 ml of SY medium for 2 days at 28°C, and then 1 ml of the preculture was inoculated into 70 ml of PV8 medium. After 2 days of culture at 28°C, [1-13 C]hexanoic acid or [1,2,3,4-13 C]octanoic acid were added to the culture at a final concentration of 0.3 mM. Five days after inoculation, an equal volume of acetone was added to the broth and filtered to remove the mycelia. The filtrate was evaporated in vacuo to obtain an aqueous solution. The solution was adjusted to pH 4 using acetic acid and extracted twice with the same volume of ethyl acetate. The organic layer was dried over Na 2 SO 4 and concentrated under reduced pressure to yield a brown oil. 1.5 g of extract from 2 liters of [1-13 C]hexanoic acid feeding culture and 0.5 g of extract from 1 liter of [1,2,3,4-13 C]octanoic acid feeding culture were obtained. The 1.5-g extract from [1-13 C]hexanoic acid feeding was subjected to silica gel chromatography with stepwise elution from 100:0 to 0:100 with CHCl 3 /methanol. Compound 1 was eluted 100:10 with CHCl 3 /methanol. The fraction was separated using a Waters 600 HPLC system equipped with a 2996 photodiode array detector and a Senshu Pak Pegasil ODS column (20 ϫ 250 mm) using an acetonitrile, 0.05% aqueous formic acid isocratic

Middle Chain FACL Involved in RM-A Biosynthesis
system (48/52) at a flow rate of 9 ml min Ϫ1 to yield 7.7 mg of compound 1.
The 0.5-g extract from the [1,2,3,4-13 C]octanoic acid feeding was separated on a Sephadex LH-20 column (GE Healthcare) and eluted with methanol. The fractions containing compound 1 were applied to a 12-g RediSep Rf column (Teledyne Isco, Inc., Lincoln, NE) for medium pressure liquid chromatography with linear gradient from 20:80 to 100:0 of ethyl acetate/hexane for 35 min at a flow rate 30 ml min Ϫ1 to obtain 28 fractions. The fraction containing compound 1 was separated on a Senshu Pak Pegasil ODS (10 ϫ 250 mm) column using a methanol, 0.05% aqueous formic acid isocratic system (67/33) at a flow rate of 4 ml min Ϫ1 to yield 1.6 mg of compound 1.

Disruption of revR, revS, and revT Genes and Analysis of
Metabolite Profiles-In the compound 1 gene cluster, the revR, revS, and revT genes were in the same transcriptional unit. Based on BLAST searching, revT gene belongs to the CCR homologs that are distributed in gene clusters producing polyketide compounds with atypical extender units (27)(28)(29)(30)(31)(32). In addition to the revT gene, homologs of the revR and revS genes were also distributed in various polyketide gene clusters (Fig. 1, panels I-IV). Therefore, to understand the role of revR, revS, and revT genes, we conducted gene disruptions (Fig. 2, A, C, and E). After confirmation of gene disruption (⌬revR, ⌬revS, and ⌬revT) by Southern hybridization (Fig. 2, B, D, and F), the metabolite profile of each gene disruptant was analyzed by LC/ESI-MS (Fig. 3). In the wild-type strain, compound 1 was the major product among the RM derivatives (Figs. 1A and 3A). Interestingly, ⌬revR mutants selectively decreased the produc-tion of compound 1 (Fig. 3B). Reintroduction of the revR gene under the control of the aph promoter restored compound 1 production to wild-type levels (Fig. 3C). ⌬revS mutants showed an ϳ50% reduction of all RM derivatives (Figs. 1A and 3D), and the phenotype was recovered by reintroduction of the revS gene (Fig. 3E), suggesting that RevS is involved in the formation of C18 alkyl residues in RM derivatives. Additionally, ⌬revT mutants completely abolished the production of RMs (Fig. 3F), which was restored by complementation with the revT gene (Fig. 3G), indicating that RevT is essential for the production of RMs.
Purification and Characterization of RevS-Based on BLAST searching, RevS belongs to the uncharacterized adenylateforming enzyme family. Together with the gene disruption phenotype (Fig. 3, D and E), we hypothesized that RevS should function as an FACL or FAAL. We heterologously expressed His 8 -tagged RevS in E. coli and purified the protein to homogeneity using Ni-NTA column chromatography (Fig. 4A). Both monomeric (65 kDa) and dimeric (123 kDa) RevS proteins were observed following gel filtration analysis (Fig. 4B). After gel filtration chromatography, we evaluated the specific activity of the monomeric and dimeric fractions. Based on the FACL assay, the specific activities of the monomer and dimer fractions of RevS were the same as that of the purified RevS before gel filtration (Fig. 4C). Therefore, the multimeric status of RevS on gel filtration chromatography was speculated to be the monomer-dimer equilibrium.
To examine FACL activity, purified RevS was incubated with various fatty acids in the presence of ATP and CoA, and the reaction products were analyzed by LC/ESI-MS. RevS activated both saturated and unsaturated fatty acids with carbon chain lengths of C4 -C11 into fatty acyl-CoAs (Table 2) but did not activate fatty acids with chain lengths of more than C13. Efficient acyl-CoA formation was observed when middle chain fatty acids (C8 -C10) were used as the substrate. Moreover, to examine FAAL activity, we heterologously expressed holo-ACP in E. coli and purified the protein to homogeneity using Ni-NTA column chromatography (Fig. 4D). Because the His tag was not attached in Sfp, only holo-ACP was purified using Ni-NTA column chromatography after co-expression of Sfp and ACP in E. coli. The holo-ACP was further purified by Mono Q column chromatography, and its modification was confirmed by MALDI-TOF/MS analysis (Fig. 4E). Next, purified RevS was incubated with various fatty acids in the presence of ATP and holo-ACP. Although the FAAL and FACL activities of RevS showed the same substrate specificity, the specific activity of FAAL was significantly lower than that of FACL (Fig. 4F).
Analysis of Kinetic Properties of FACL Activity of RevS-To determine the kinetic parameters of RevS for each substrate, we performed spectrophotometric assays. The RevS reaction apparently followed Michaelis-Menten kinetics. High catalytic efficiency was observed when middle chain fatty acids (C8 -C10) were used for the RevS assay (Table 3). In addition, RevS accepted both saturated and unsaturated fatty acids as substrates with the same catalytic efficiency. We also calculated kinetic constants for ATP and CoA. The K m and k cat values for ATP were 0.039 Ϯ 0.004 mM and 56.7 Ϯ 2.2 min Ϫ1 , respectively, giving a catalytic efficiency of 1456 min Ϫ1 mM Ϫ1 . The K m and k cat values of CoA were 0.05 Ϯ 0.01 mM and 39.6 Ϯ 0.8 min Ϫ1 , respectively, giving a catalytic efficiency of 861 min Ϫ1 mM Ϫ1 . The kinetic parameters were comparable with those for other FACLs reported previously (Table 3).
Phylogenetic Analysis of RevS-To obtain insight into the molecular evolution of RevS, we performed phylogenetic analysis of the RevS using the amino acid sequences of FACL, FAAL, and a variety of CoA ligases (Fig. 5A). Consistent with the presence of a FAAL-specific insertion motif (Fig. 5B) (14), RevS was classified in an FAAL clade. However, biochemical analyses clearly demonstrated that RevS is an FACL. Interestingly, RevS homologs such as CinT and SamR0482 associated in the biosynthetic gene cluster of cinnabaramide and stambomycin (Fig. 1B), respectively, were found in the same branch, suggesting that CinT and SamR0482 may be middle chain FACLs. These RevS homologs may have evolved from the FAAL family and were adapted for the production of fatty acyl-CoAs to efficiently convert them into 2-alkylmalonyl-CoAs for the PKS extender unit.
Purification and Biochemical Characterization of RevT-To determine the function of RevT, His 8 tag RevT was heterologously expressed in S. lividans TK23 and purified using Ni-NTA column chromatography. The molecular mass of the RevT was estimated to be 48 kDa by SDS-PAGE (Fig. 6A) and 178 kDa by gel filtration chromatography (Fig. 6B), suggesting that RevT exists as a tetramer. Incubation of purified RevT in the presence of (E)-2-hexenoyl-CoA, NaHCO 3 , and NADPH was performed for 30 min. The time-dependent production of butylmalonyl-CoA was observed for 20 min (Fig. 6C).
Because the understanding of substrate specificity and kinetic properties of RevT is important for the synthesis of polyketide compounds, we also characterized the kinetic parameters of (E) than that of CinF (31). Catalytic efficiency of SalG was significantly higher than that of RevT (27).
RevS and RevT Are Responsible for Butylmalonyl-CoA Biosynthesis-Because of efficient incorporation of fatty acids into compound 1, we also investigated the successive conversion of (E)-2-hexenoic acid into (E)-2-hexenoyl-CoA by RevS and the carboxylase/reductase reaction by RevT to give butylmalonyl-CoA (Fig. 8A). As expected, (E)-2-hexenoic acid was efficiently converted into butylmalonyl-CoA in the RevS-RevT coupling reaction (Fig. 8B). This is the first report of in vitro reconstruction of 2-alkylmalonyl-CoA using physiological enzymes derived from a secondary metabolite gene cluster.

Discussion
In this study, we characterized the revR, revS, and revT genes, which are involved in 2-alkylmalonyl-CoA formation in RM biosynthesis. The revR gene disruption demonstrated that RevR was involved in the selective production of butylmalonyl-CoA, which was consistent with the high production of compound 1 in wild-type culture (Fig. 3). RevR showed 55% amino acid identity to BenQ that is reported to be crucial for providing the hexanoate PKS starter unit for benastatin biosynthesis (54,55). In addition, it has been reported that KASIII is a determinant of the fatty acid priming unit producing straight or branched chain, as well as the even or odd number chain (56,57). RevR is likely to select butyryl-CoA as a priming substrate to yield 3-oxo-hexanoyl-ACP. Next, the product may be converted into (E)-2-hexenoyl-ACP by ␤-keto processing enzymes (Fig. 1C). In this de novo fatty acid biosynthetic pathway, an unknown transacylase may be responsible for the conversion of (E)-2-hexenoyl-ACP into (E)-2-hexenoyl-CoA, as RevT did not accept (E)-2-enoyl-ACP as a substrate (Fig. 8D). The revS gene disruption and complementation analysis indicated the nonessentiality of RevS for the biosynthesis of RMs (Fig. 3, D and E). We speculate that RM biosynthesis in ⌬revS mutants was mainly supported by de novo fatty acid biosynthesis (Fig. 1C). Another possibility is the involvement of unidentified CoA ligase, which may support medium chain acyl-CoA formation. We found three genes showing 30 -43% amino acid identity to RevS in the draft genome of Streptomyces sp. SN-593. Although three were not in the FACL clade of RevS (Fig. 5A), gene disruption and biochemical analysis will be essential for final conclusions.
Overall, the structure of the class I adenylate-forming enzymes is composed of large N-terminal and small C-terminal domains connected by a flexible hinge region (9). Crystal structure analysis of FAAL28 (FadD28) from M. tuberculosis revealed that FAAL has a specific insertion motif containing 22 amino acids (Fig. 5B). The insertion motif modulates C-terminal domain movement and disrupts access of the acyl-AMP intermediate to the CoA-binding motif (14). Amino acid alignments and phylogenetic analysis of adenylate-forming enzymes indicated that RevS belongs to the FAAL clade. Unexpectedly, biochemical analysis of RevS revealed FACL activity even in the presence of the FAALspecific insertion motif (Fig. 5). Arora et al. (14) also reported that deletion of the insertion motif found in FadD28 resulted in the conversion of FAAL activity into FACL activity, indicating that FAAL originally contains CoA binding pocket. Because phylogenetic analysis suggested that FAAL was derived from the CoA ligase, the FACL activity of RevS may have re-evolved from the FAAL clade by generating access to the CoA binding pocket by optimizing the orientation of the N-and C-terminal domains. Interestingly, in contrast to RevS, FadD10, which does not possess the FAAL-specific insertion motif, from M. tuberculosis showed FAAL activity. This was explained by the orientation of its N-and C-terminal domains, which are quite different from those of other adenylate-forming family members (58). Therefore, crystal structure analysis of RevS is essential for fully understanding its biosynthetic mechanism.
Metabolite profiling of revS gene disruptants (Figs. 2 and 3) and kinetic analysis of RevS ( Fig. 4 and Tables 2 and 3) also suggested the presence of available free fatty acids (C6 -C10) during the late stationary phase and the close relationship between fatty acid metabolism and polyketide biosynthesis. Streptomyces typically utilizes de novo-synthesized fatty acids not only for building membrane phospholipids but also to accumulate neutral lipid storage compounds such as triacylglycerol (TAG) (59 -61). TAG is one of the important carbon sources for energy production and a precursor for secondary metabolism. A link between TAG degradation and actinorhodin production has been suggested (59). Because TAG found in S. lividans consists of fatty acids ranging from C14 to C18 (59), the fatty acids derived from TAG degradation should mainly be utilized for the production of acetyl-CoA, but not as RevS substrates. Therefore, other mechanisms may generate medium chain fatty acids to support the RevS reaction. A possible mechanism is termination of fatty acid biosynthesis by chain lengthspecific acyl-ACP thioesterase. Interestingly, it has been reported that the seed of Cuphea palustris possessed C8-and C14-specific acyl-ACP thioesterases, CpFatB1 and CpFatB2, respectively, which is consistent with the accumulation of middle chain fatty acids in the seed (62). However, we found no homologous gene in Streptomyces sp. SN-593. Additional detailed studies are needed to understand fatty acid homeostasis during secondary metabolite biosynthesis.
In summary, 2-alkylmalonyl-CoA biosynthesis was strongly supported by the functions of RevR and RevS. These enzymes effectively utilized de novo fatty acid biosynthesis and fatty acid degradation products, respectively. Decreased supply of atypical building blocks in the polyketide assembly line is critical for secondary metabolite biosynthesis. Therefore, the presence of RevR and RevS may be a back-up system to ensure the production of atypical extender units. Our results explain why homologs of RevR and RevS are distributed in polyketide biosynthetic gene clusters, which utilize atypical extender units (Fig. 1).