Functional Analysis of P450 Monooxygenase SrrO in the Biosynthesis of Butenolide-Type Signaling Molecules in Streptomyces rochei

Streptomyces rochei 7434AN4 produces two structurally unrelated polyketide antibiotics lankacidin and lankamycin, and their biosynthesis is tightly controlled by butenolide-type signaling molecules SRB1 and SRB2. SRBs are synthesized by SRB synthase SrrX, and induce lankacidin and lankamycin production at 40 nM concentration. We here investigated the role of a P450 monooxygenase gene srrO (orf84), which is located adjacent to srrX (orf85), in SRB biosynthesis. An srrO mutant KA54 accumulated lankacidin and lankamycin at a normal level when compared with the parent strain. To elucidate the chemical structures of the signaling molecules accumulated in KA54 (termed as KA54-SRBs), this mutant was cultured (30 L) and the active components were purified. Two active components (KA54-SRB1 and KA54-SRB2) were detected in ESI-MS and chiral HPLC analysis. The molecular formulae for KA54-SRB1 and KA54-SRB2 are C15H26O4 and C16H28O4, whose values are one oxygen smaller and two hydrogen larger when compared with those for SRB1 and SRB2, respectively. Based on extensive NMR analysis, the signaling molecules in KA54 were determined to be 6′-deoxo-SRB1 and 6′-deoxo-SRB2. Gel shift analysis indicated that a ligand affinity of 6′-deoxo-SRB1 to the specific receptor SrrA was 100-fold less than that of SRB1. We performed bioconversion of the synthetic 6′-deoxo-SRB1 in the Streptomyces lividans recombinant carrying SrrO-expression plasmid. Substrate 6′-deoxo-SRB1 was converted through 6′-deoxo-6′-hydroxy-SRB1 to SRB1 in a time-dependent manner. Thus, these results clearly indicated that SrrO catalyzes the C-6′ oxidation at a final step in SRB biosynthesis.

The cytochrome P450 comprises a ferric heme-containing enzyme superfamily that catalyzes the oxygenation reaction on vast array of substrates including antibiotics, lipids, and steroids, and is widely distributed across the Kingdoms from bacteria to human [24][25][26]. One of the typical chemical reactions catalyzed by P450 enzymes is an oxidation of non-reactive C-H bonds [27]. Monooxygenase reactions are involved in the macrolide biosynthesis of erythromycin [28,29], oleandomycin [30], and lankamycin [19]. Location of srrO (orf84) gene at the neighbor of srrX (orf85) indicates that srrO may be involved in SRB biosynthesis in S. rochei. We here analyzed the function of srrO by gene inactivation and enzymatic bioconversion experiments, the results of which will be described in this paper.

Strains, Reagents, and Culture Conditions
All strains and plasmids used in this study were listed in Table S1. Strain 51252 that harbors pSLA2-L in addition to the chromosome was used as a parent strain [12]. Strain KA20, a double mutant of srrX and the transcriptional repressor gene srrB, was used as signaling-molecule indicator strain [11,21]. YM medium (0.4% yeast extract, 1.0% malt extract, and 0.4% D-glucose, pH The cytochrome P450 comprises a ferric heme-containing enzyme superfamily that catalyzes the oxygenation reaction on vast array of substrates including antibiotics, lipids, and steroids, and is widely distributed across the Kingdoms from bacteria to human [24][25][26]. One of the typical chemical reactions catalyzed by P450 enzymes is an oxidation of non-reactive C-H bonds [27]. Monooxygenase reactions are involved in the macrolide biosynthesis of erythromycin [28,29], oleandomycin [30], and lankamycin [19]. Location of srrO (orf84) gene at the neighbor of srrX (orf85) indicates that srrO may be involved in SRB biosynthesis in S. rochei. We here analyzed the function of srrO by gene inactivation and enzymatic bioconversion experiments, the results of which will be described in this paper.

Strains, Reagents, and Culture Conditions
All strains and plasmids used in this study were listed in Table S1. Strain 51252 that harbors pSLA2-L in addition to the chromosome was used as a parent strain [12]. Strain KA20, a double mutant of srrX and the transcriptional repressor gene srrB, was used as signaling-molecule indicator strain [11,21]. YM medium (0.4% yeast extract, 1.0% malt extract, and 0.4% D-glucose, pH 7.3) was used for signaling-molecule synthesis and bioassay. For protoplast preparation and SrrO-protein expression, Streptomyces strains were grown in YEME liquid medium [31]. Protoplasts were regenerated on R1M solid medium [32]. To construct targeting plasmids for srrO mutation, E. coli strains were grown in Luria-Bertani medium supplemented with ampicillin (100 µg/mL) when necessary. Genetic manipulations for Streptomyces [31] and E. coli [33] were performed according to the described procedures.

Spectroscopic Instruments
NMR spectra were recorded on ECA-500 and/or ECA-600 spectrometers (JEOL, Tokyo, Japan) equipped with a field gradient accessory. Deuteriochloroform (99.8 atom% enriched; Kanto Chemical Co., Ltd., Tokyo, Japan) was used as a solvent. Chemical shifts were recorded as a δ value based on a resident solvent signal (δ C = 77.0), or an internal standard signal of tetramethylsilane (δ H = 0). High resolution electrospray ionization (ESI) mass spectra were measured by a LTQ Orbitrap XL mass spectrometer (Thermo Fisher Scientific, Waltham, MA, USA). High resolution gas chromatography-time of flight mass spectra (ionization mode; CI) were acquired on a JMS-T100 GCV 4G (JEOL, Tokyo, Japan). Optical rotations were measured using a DIP-370 polarimeter (JASCO Cooperation, Tokyo, Japan). IR spectra were recorded on a IRAffiniy-1 spectrometer (Shimadzu Cooperation, Kyoto, Japan) using the ATR (Attenuated Total Reflection) method.

Construction of the srrO Mutant KA54
A 3.8-kb StuI-EcoRI fragment (nt 143,844,147,674 of pSLA2-L) containing an srrO gene was ligated with 2.8-kb StuI-EcoRI fragment of Litmus 28i to give pKAR3041. To a ClaI restriction site at the 5'-terminal region of srrO was introduced a 1.0-kb ClaI fragment of aac(3)IV gene cassette conferring apramycin resistance. The resulting plasmid pKAR3043 was digested with EcoRI and StuI, and the vector of which was replaced with an EcoRI-SmaI fragment of pRES18, an E. coli-Streptomyces shuttle vector [34], to afford pKAR3044. This plasmid was transformed into protoplast of strain 51252, and then an srrO mutant KA54 was obtained through homologous recombination according to our protocol [18] ( Figure S2).

Metabolites in the srrO Mutant KA54
Antibiotic production in strain KA54 was analyzed by high performance liquid chromatography (HPLC) and thin layer chromatography (TLC) in comparison with that in parent strain 51252. The crude extract was applied on a COSMOSIL CHOLESTER column (4.6 × 250 mm, Nacalai Tesque, Kyoto, Japan) and eluted with a mixture of acetonitrile-10 mM sodium phosphate buffer (pH 8.2) (3:7, v/v) at a flow rate of 1.0 mL/min. The eluate was monitored at 230 nm with a MD-2010 multiwavelength photodiode array detector (JASCO Cooperation, Tokyo, Japan). TLC was developed with a mixture of CHCl 3 -methanol (15:1, v/v) and baked after spraying with anisaldehyde-H 2 SO 4 .

Isolation of Signaling Molecules from the srrO Mutant KA54
Seed culture (20 mL) of strain KA54 was inoculated into to 2 L YM medium in 5-L Erlenmeyer flask, which was grown at 28 • C for 36 h. A total of 30 L culture broth was extracted with equal volume of EtOAc twice. The combined organic phase was dried (Na 2 SO 4 ), filtered, and concentrated to dryness. The resulting crude extracts were purified by Sephadex LH-20 (GE Healthcare, Chicago, IL, USA) with methanol. Each fraction was subjected to a bioassay using strain KA20 as a test organism according to our previous report [11]. Active fractions were collected and purified by a series of silica gel column chromatography with two different solvent systems of CHCl 3 -MeOH (50:1, v/v) and toluene-EtOAc    Nevertheless, the low quantity of active components (100 µg from 30 L culture), compounds 1 and 2 were separated by repeated runs of HPLC (25% aqueous acetonitrile containing 0.1% trifluoroacetic acid) at 10.6 min and 17.1 min, respectively. Their C-1' stereochemistry was confirmed by chiral HPLC using synthetic 6 -deoxo-SRBs (details were shown in Sections 2.7 and 3.3).
Compound 13 (696 mg, 4.09 mmol) and L-menthyloxy-butenolide (962 mg, 3.81 mmol) were treated in the same manner as described for the preparation of 10 to give a 2:1 mixture of 14a and 14b (806 mg, 50%) as a colorless oil, which was also further separated by flash chromatography.
Compound  The compound 14a (19 mg, 45 µmol) was treated in the same manner as described for the preparation of 1a to give 6 -deoxo-SRB2a (2a) (

Gel Shift Assay
Preparation of an SrrA protein (SRB receptor) and a DNA probe containing the promoter region of srrY (srrYp), a target gene of SrrA, was described previously [22,38]. The reaction mixture contained the binding buffer (20 mM Tris-HCl [pH 8.0], 100 mM NaCl, 1 mM dithiothreitol, 0.1 mg of bovine serum albumin and 5% glycerol), 0.35 nM labeled DNA, and 2 µM SrrA protein. To analyze the ligand affinity of 6 -deoxo-SRB on the binding of SrrA, various concentrations of synthetic 6 -deoxo-SRBs and SRBs were added to the reaction mixture.

Bioconversion of 6 -deoxo-SRB1 in the SrrO Recombinant
To a 2-day-growth culture (100 mL) of S. lividans TK64/pNTT01 (+SrrO) was added 2 µmol of the substrate, and the fed cultures were further incubated for 0-5 h periods. The culture supernatant was extracted with EtOAc twice, and the combined organic phase was dried (Na 2 SO 4 ), filtered, and concentrated to dryness. The crude extracts were analyzed by ESI-MS and TLC. The cell culture of S. lividans TK64/pHSA81 (control) [39] was used for negative control.

Construction and Metabolite Analysis of an srrO Mutant KA54
The P450 monooxygenase gene srrO was inactivated to analyze its function in SRB biosynthesis. An apramycin resistance gene cassette was introduced into a 5'-terminal region of srrO, and double crossover mutants were obtained through homologous recombination. Gene replacement was confirmed by Southern hybridization experiment. As shown in Figure S1, a 4.4-kb BspEI-XhoI fragment in parent 51252 was changed into two fragments with 2.3-kb and 3.1-kb in mutant KA54. A metabolite profile in strain KA54 was analyzed in comparison with that in parent 51252. KA54 produced lankacidin and lankamycin in a comparative level to 51252 (Figure 2A). There are two possibilities to explain that the disruption of the srrO showed no effect on the production of lankacidin and lankamycin; (1) SrrO is not involved in SRB biosynthesis, and (2) the signaling molecule(s) in the srrO mutant has(have) an ability to induce lankacidin and lankamycin production.

Structural Elucidation of Signaling Molecules in KA54
To investigate these possibilities mentioned above, we carried out the isolation of signaling molecules in KA54 (termed as KA54-SRBs). Cultivation of KA54 was stopped at a 36 h period to reduce the accumulation of lankacidin and lankamycin, which would disturb the SRB bioassay. A 30-L culture was extracted with ethyl acetate (EtOAc), and the resulting oil was purified by Sephadex LH20 with methanol. Each fraction was subjected to bioassay to check its antibiotic-inducing activity. The srrX mutant was used for an indicator strain that could restore lankacidin and lankamycin production in the presence of SRBs. The active fractions were combined and purified by series of silica gel chromatography with CHCl 3 -methanol (50:1, v/v) and toluene-EtOAc (3:1, v/v).
The active fractions were further analyzed by ESI-MS analysis. ESI-MS indicated the presence of two active components (KA54-SRB1 and KA54-SRB2) in the ratio 1:1 ( Figure 2B). The molecular formulae for KA54-SRB1 and KA54-SRB2 were established to be C 15 H 26 O 4 and C 16 H 28 O 4 , whose values were one oxygen smaller and two hydrogen larger when compared with those for SRB1 and SRB2, respectively. Due to the low amounts of active components KA54-SRB1 and KA54-SRB2, we further analyzed their structural assignments as a mixture (compound with C 15 H 26 O 4 was termed as KA54-SRB1, while C 16 H 28 O 4 as KA54-SRB2).
To confirm the C-1' stereochemistry in compounds 1 and 2, chiral HPLC analysis was performed ( Figure 3). The retention times of 1 and 2 (21.1 and 53.3 min, respectively) on a chiral HPLC column were identical to those of the synthetic (1'R)-isomers 1a and 2a, whereas the synthetic (1'S)-isomers 1b and 2b eluted slightly earlier at 20.2 and 50.0 min. Thus, the C-1' stereochemistry in 6 -deoxo-SRBs was the same as with SRBs.

Ligand Affinity of 6′-deoxo-SRBs
To compare an antibiotic-inducing activity of SRBs and 6′-deoxo-SRBs, a bioassay method using the srrX mutant was not suitable since 6′-deoxo-SRBs could be converted to SRBs in some extent by SrrO protein expressed in the srrX mutant. Therefore, the antibiotic-inducing activity of 6′-deoxo-SRBs was evaluated by minimum concentration to dissociate SrrA protein from a main

Ligand Affinity of 6 -deoxo-SRBs
To compare an antibiotic-inducing activity of SRBs and 6 -deoxo-SRBs, a bioassay method using the srrX mutant was not suitable since 6 -deoxo-SRBs could be converted to SRBs in some extent by SrrO protein expressed in the srrX mutant. Therefore, the antibiotic-inducing activity of 6 -deoxo-SRBs was evaluated by minimum concentration to dissociate SrrA protein from a main target gene srrY, which encodes SARP and activates lankacidin and lankamycin production in S. rochei [22]. Gel shift assay was performed using a recombinant SrrA protein and a 32 P-labeled DNA probe containing srrY-promoter (srrYp) region in the presence of either 6 -deoxo-SRB1 or SRB1 ( Figure 4A). As shown in Figure 4B, minimum concentration of 6 -deoxo-SRB1 was 100-fold higher than that of SRB1. This finding indicated that C-6 keto group is important for improvement of antibiotic-inducing activity in S. rochei.  Figure 1B) [11] or synthetic 6′-deoxo-SRB1 ((1'R)-isomer; Figure 1B) was added.
place at the final step in SRB biosynthesis. One of the plausible reasons is an increase of hydrophilic property in SRB to improve antibiotic-inducing activity. To our knowledge, this is a first report to obtain biosynthetic intermediates of signaling molecules in Streptomyces species.

Discussion
The function of the P450 monooxygenase gene srrO in SRB biosynthesis was analyzed through gene disruption, gel-shift assay, and by in vivo enzymatic conversion. The srrO disruptant produced lankacidin and lankamycin in a comparative yield with the parent 51252, and accumulated novel signaling molecules, 6 -deoxo-SRB1 (1 = 1a) and 6 -deoxo-SRB2 (2 = 2a). Based on a ligand activity of signaling molecules for dissociation of their receptor SrrA, 6 -deoxo-SRB1 exhibits the 100-fold less binding activity compared with SRB1. Nevertheless, 6 -deoxo-SRBs could also bind to the specific SRB receptor SrrA to induce lankacidin and lankamycin production in S. rochei. At this stage, we have no answer why oxidation of C-6 methylene to ketone by SrrO took place at the final step in SRB biosynthesis. One of the plausible reasons is an increase of hydrophilic property in SRB to improve antibiotic-inducing activity. To our knowledge, this is a first report to obtain biosynthetic intermediates of signaling molecules in Streptomyces species.
Several possible genes for SRB biosynthesis and antibiotic regulation were found around srrX (orf85) on pSLA2-L; the NAD-dependent dehydrogenase gene srrG (orf81), the phosphatase gene srrP (orf83), the P450 monooxygenase gene srrO (orf84), and the thioesterase gene srrH (orf86), together with repressor genes srrA (orf82) and srrB (orf79). The possible biosynthetic pathway of SRBs were shown in Figure 7 based on the biosynthetic pathways for other signaling molecules, A-factor in S. griseus [8] and virginia butanolides in S. virginiae [43], and antifungal butenolide gladiofungin in Burkholderia gladioli HKI0739 [44]. Medium-chain (C 12 or C 13 ) β-keto acid was derived through fatty acid biosynthesis pathway; four units of malonyl CoA are condensed with either an isobutyryl CoA unit for SRB1 or a 2-methylbutyryl CoA unit for SRB2. These β-keto acids are condensed with a C 3 unit (a hydrate form of glyceraldehyde 3-phosphate) by SrrX, then followed by spontaneous dephosphorylation, dihydroxylation, and intramolecular aldol condensation to generate the butenolide skeleton. The C-1' ketone moiety in butenolide intermediate will be reduced by dehydrogenase SrrG to synthesize 6 -deoxo-SRBs, which accept two-stage oxidation by SrrO to form SRBs. Several possible genes for SRB biosynthesis and antibiotic regulation were found around srrX (orf85) on pSLA2-L; the NAD-dependent dehydrogenase gene srrG (orf81), the phosphatase gene srrP (orf83), the P450 monooxygenase gene srrO (orf84), and the thioesterase gene srrH (orf86), together with repressor genes srrA (orf82) and srrB (orf79). The possible biosynthetic pathway of SRBs were shown in Figure 7 based on the biosynthetic pathways for other signaling molecules, A-factor in S. griseus [8] and virginia butanolides in S. virginiae [43], and antifungal butenolide gladiofungin in Burkholderia gladioli HKI0739 [44]. Medium-chain (C12 or C13) β-keto acid was derived through fatty acid biosynthesis pathway; four units of malonyl CoA are condensed with either an isobutyryl CoA unit for SRB1 or a 2-methylbutyryl CoA unit for SRB2. These β-keto acids are condensed with a C3 unit (a hydrate form of glyceraldehyde 3-phosphate) by SrrX, then followed by spontaneous dephosphorylation, dihydroxylation, and intramolecular aldol condensation to generate the butenolide skeleton. The C-1' ketone moiety in butenolide intermediate will be reduced by dehydrogenase SrrG to synthesize 6′-deoxo-SRBs, which accept two-stage oxidation by SrrO to form SRBs. Streptomyces signaling molecules have a crucial role to induce secondary metabolite biosynthesis. Many Streptomyces strains have more than 30 secondary metabolites biosynthetic gene clusters in their genome; however, many of them (around 80-90%) are silent in normal culture conditions. In S. rochei, 40 biosynthetic gene clusters (35 in the chromosome and 5 in pSLA2-L) are found; however, only six compounds are detected [17,45]. Although we do not yet know the reason why many of them are silent or poorly expressed, a lack of specific signaling molecules is one of a plausible possibility [46,47]. Streptomyces strains generally have multiple signaling-molecule receptor genes, suggesting that they may have a potential to recognize heterologous signaling molecules. Thus, signaling molecules may contribute as "genetic engineering-free" genome mining tools to act as communication signals between actinomycetes, between different bacteria, and/or between interkingdom.

Conclusions
We here investigated the role of a P450 monooxygenase gene srrO in SRB biosynthesis. Two signaling molecules in the srrO-disruptant KA54 were determined to be 6′-deoxo-SRB1 and 6′-deoxo-SRB2 by ESI-MS, NMR, and HPLC analyses from 30 L culture extract. A ligand affinity of 6′-deoxo-SRB1 to SrrA was shown to be 100-fold less than that of SRB1. We further performed bioconversion of the synthetic 6′-deoxo-SRB1 in the S. lividans recombinant carrying SrrO-expression plasmid. Substrate 6′-deoxo-SRB1 was converted through Streptomyces signaling molecules have a crucial role to induce secondary metabolite biosynthesis. Many Streptomyces strains have more than 30 secondary metabolites biosynthetic gene clusters in their genome; however, many of them (around 80-90%) are silent in normal culture conditions. In S. rochei, 40 biosynthetic gene clusters (35 in the chromosome and 5 in pSLA2-L) are found; however, only six compounds are detected [17,45]. Although we do not yet know the reason why many of them are silent or poorly expressed, a lack of specific signaling molecules is one of a plausible possibility [46,47]. Streptomyces strains generally have multiple signaling-molecule receptor genes, suggesting that they may have a potential to recognize heterologous signaling molecules. Thus, signaling molecules may contribute as "genetic engineering-free" genome mining tools to act as communication signals between actinomycetes, between different bacteria, and/or between interkingdom.

Conclusions
We here investigated the role of a P450 monooxygenase gene srrO in SRB biosynthesis. Two signaling molecules in the srrO-disruptant KA54 were determined to be 6 -deoxo-SRB1 and 6 -deoxo-SRB2 by ESI-MS, NMR, and HPLC analyses from 30 L culture extract. A ligand affinity of 6 -deoxo-SRB1 to SrrA was shown to be 100-fold less than that of SRB1. We further performed bioconversion of the synthetic 6 -deoxo-SRB1 in the S. lividans recombinant carrying SrrO-expression plasmid. Substrate 6 -deoxo-SRB1 was converted through 6 -deoxo-6 -hydroxy-SRB1 to SRB1 in a time-dependent manner. Thus, these results clearly indicated that SrrO catalyzes the C-6 oxidation at a final step in SRB biosynthesis.