Identification of a butenolide signaling system that regulates nikkomycin biosynthesis in Streptomyces

Butenolides are an emerging family of signaling molecules in Streptomyces. They control complex physiological traits, such as morphological differentiation and antibiotic production. However, how butenolides regulate these processes is poorly investigated because of obstacles in obtaining these signaling molecules. This study reports the identification of a butenolide-type signaling system for nikkomycin biosynthesis in Streptomyces ansochromogenes with distinct features. We identified a gene cluster, sab, consisting of three genes, sabAPD, for butenolide biosynthesis and two regulator genes, sabR1 and sabR2, and characterized three butenolides (SAB1, -2, and -3) by heterologous expression of sabAPD. sabA disruption abolished nikkomycin production, which could be restored by the addition of SABs or by deletion of sabR1 in ΔsabA. Electrophoretic mobility-shift assays and transcriptional analyses indicated that SabR1 indirectly represses the transcription of nikkomycin biosynthetic genes, but directly represses sabA and sabR1. In the presence of SABs, the SabR1 transcriptional regulator dissociated from its target genes, verifying that SabR1 is the cognate receptor of SABs. Genome-wide scanning with the conserved SabR1-binding sequence revealed another SabR1 target gene, cprC, whose transcription was strongly repressed by SabR1. Intriguingly, CprC positively regulated the pleiotropic regulatory gene adpA by binding to its promoter and, in turn, activated nikkomycin biosynthesis. This is the first report that butenolide-type signaling molecules and their cognate receptor SabR1 can regulate adpA via a newly identified activator, CprC, to control nikkomycin production. These findings pave the way for further studies seeking to unravel the regulatory mechanism and functions of the butenolide signaling system in Streptomyces.

Improper use of antibiotics as well as the horizontal transfer of antibiotic resistance genes has led to the continuous appear-ance of antibiotic-resistant pathogens and the loss of antibiotic native efficiency. Most commercially important natural antibiotics are produced by Streptomyces. Their biosynthesis is precisely controlled by cluster-situated regulators (CSRs) 4 or global regulators, most of which can respond to small molecules including hormone-like signaling molecules, also known as autoregulators, and some specialized metabolites (1). Defining these signal transmission pathways is desirable to improve the efficiency of antibiotic production yield and to identify new gene clusters to discover previously unknown bioactive products of interest.
Although signaling molecules are widely distributed in Streptomyces, their biosynthesis is usually under stringent control, resulting in low yields in producing strains. This impedes the discovery of new signaling molecules and elucidation of their function. Since the first ␥-butyrolactone (GBL) signaling molecule, A-factor, was discovered in Streptomyces griseus, only 33 autoregulators have been identified in Streptomyces (1)(2)(3). They are classified into five groups, GBLs, furans, butenolides, PI factor, and N-methylphenylalanyl-dehydrobutyrine diketopiperazine, based on their structures. GBL is the largest family of autoregulators, and its regulatory mechanism and biosynthesis have been intensively investigated, whereas studies on other family of autoregulators are very rare (4). Recently, a butenolide signaling molecule, avenolide, was found to be essential for biosynthesis of the clinically important antibiotic avermectin, inspiring more interest in this family of regulators (5). Seven butenolides associated with antibiotic biosynthesis have been discovered so far, including avenolide from Streptomyces avermitilis and Streptomyces albus (3,5) and SRBs from Streptomyces rochei (6). Butenolide contains a five-member ring backbone, but with an unsaturated bond at C3-C4 and diverse side chains at C3, C4, or C5, which confer unique activities.
Signaling molecules and their receptors can exert considerable impact on the onset and production of antibiotic biosynthesis. A typical autoregulator system has been exemplified in S. griseus and portrayed as a pyramid-like network. A-factor and its receptor ArpA constitute the apex of the pyramid, controlling a series of diverse downstream pathways via a pivot regulator, AdpA. ArpA directly represses the transcription of adpA, whereas binding of A-factor to ArpA results in derepression of adpA transcription and consequently the expression profile switch of numerous target genes of AdpA (7). Subsequently, more autoregulators and their cognate receptors have been characterized. They have intimate association with antibiotic production and morphological differentiation in various Streptomyces species, and some of these are involved in the AdpA regulatory network, a pivotal pleiotropic regulator widespread in Streptomyces. Although the target genes of AdpA and their regulation have been comprehensively illustrated in many Streptomyces species, how AdpA expression itself is maintained under delicate and precise control is poorly investigated (8,9). Signaling molecules taking part in the AdpA regulatory cascade provide an important mechanism for cells to respond to environmental and physiological changes, and also signal amplification can be achieved via AdpA transmission (10).
Streptomyces ansochromogenes 7100 produces nikkomycin under the control of global regulators, such as WblA, AdpA, GBL receptor-like regulator SabR (11)(12)(13), and the pathwayspecific regulator SanG (14,15). AdpA positively regulates nikkomycin biosynthesis via binding to the promoter region of sanG, but negatively regulates oviedomycin production by repressing ovmZ/ovmW (16). An autoregulator biosynthetic gene cluster sab (KF170348) was revealed in this strain by genome mining (17). BLAST search suggested that the homolog of sabA, the core gene in sab, is widely distributed in Streptomyces. What autoregulators may be synthesized by sab and how they coordinate with AdpA to regulate the secondary metabolism in S. ansochromogenes are of great interest. In this work, we report the characterization of a novel butenolide signal transduction pathway, in which signal input is transmitted to nikkomycin biosynthesis via a newly discovered activator, CprC, of adpA.

Characterization of butenolide signal molecules (SABs) triggering nikkomycin production
It is well-known that AfsA-like proteins are key enzymes in autoregulator biosynthesis. By genome mining of S. ansochromogenes, an autoregulator biosynthetic gene cluster (sab) was identified (Fig. 1A), and sab is located about 1.87 megabases away from the nikkomycin biosynthetic gene cluster (nik) in the chromosome of S. ansochromogenes. In sab cluster, sabA encodes an AfsA-like enzyme (31% identity with AfsA from S. griseus), whereas sabP and sabD encode phosphatase and dehydrogenase enzymes, respectively, with putative tailoring functions. sabR1 and sabR2 encode TetR family regulators. SabR1 belongs to the GBL receptor family and shows 40% identity with ScbR from S. coelicolor (18), and SabR2 shows 31% identity with pseudo-GBL receptor JadR2 from S. venezuelae (19). sabR1 and sabR2 are situated downstream of sabD in sab cluster, implying their potential correlation with the signal molecules as receptors.
To understand what kind of signal molecules can be synthesized by sab and whether they can affect nikkomycin biosynthesis, a sabA disruption mutant (⌬sabA) was constructed. Nikkomycin was determined by HPLC analysis and bioassays against Candida albicans. No nikkomycin was detected in the culture supernatant of ⌬sabA after 5 days' incubation. When a copy of sabA with its promoter region was introduced into ⌬sabA, nikkomycin production was almost restored to the level of the WT strain. Furthermore, nikkomycin in ⌬sabA was restored with the addition of ethyl acetate extracts of WT, but not with the extract of ⌬sabA as expected (Fig. 1, B and C). These results indicated that compounds synthesized by SabA are closely related to nikkomycin biosynthesis.
Signaling molecules usually work at nanomolar concentrations, and it is difficult to obtain a large enough quantity from

Butenolide-mediated regulation of nikkomycin biosynthesis
native producer strains for structural determination. To overcome this problem, heterologous expression of sabAPD was carried out in Escherichia coli and in Streptomyces, respectively, and the products were detected by HPLC (Fig. 2, A and B). It was shown that nikkomycin production in ⌬sabA was restored by the addition of extracts from E. coli C41 containing sabAPD (CpAPD) or S. coelicolor M1146 containing sabAPD (MpAPD) but not recovered by the addition of extracts from strains E. coli C41/pET23b and S. coelicolor M1146/pKC1139 as controls (data not shown). Clearly, sabAPD were expressed in both E. coli C41 and S. coelicolor M1146, and the resulting products triggered nikkomycin production.
Compounds inducing nikkomycin production were isolated from CpAPD and MpAPD strains and subsequently purified using HPLC by tracking the activity inducing nikkomycin biosynthesis in ⌬sabA. A total of 30 mg of purified SAB1 was gained from 3 liters of culture broth of CpAPD. High-resolution electrospray ionization MS showed a molecular ion peak at m/z 187.0964 [M ϩ H] ϩ (Fig. S1A), and the molecular formula was deduced as C 9 H 14 O 4 . By comparing the NMR spectroscopic data with those of known compounds (compounds 1 and 3 in S. antibioticus) (20), the only difference indicated was in the side chain at C3. On 1 H NMR, one methyl group signal (␦ H 0.79; ␦ C 15.0) at C2Ј in compound 3 was absent, and CH was changed to CH 2 (␦ H 1.8/1.6; ␦ C 38.2) ( Fig. S1 and Table S1). Further analyses of the 1 H-1 H COSY, HSQC, and HMBC confirmed the side chain at C3 in SAB1 to be hydroxyl-butyl, and the structure of SAB1 was thus determined as 5-hydroxy-3-(1Јhydroxyl-butyl)-4-methyl-2(5H)-furanone (Fig. 2C), a novel member of butenolide autoregulators in Streptomyces.
HPLC analysis revealed that two other molecules, SAB2 and -3, were produced by MpAPD in addition to SAB1 (Fig. 2B). A
To confirm that purified SABs are functional in vivo, SAB1, -2, or -3 was added into ⌬sabA at concentrations ranging from 1 nM to 10 M, and nikkomycin production was detected after incubation for 5 days. HPLC showed that nikkomycin production was elicited in ⌬sabA with increasing concentration of SAB1, -2, or -3. SAB1 and SAB3 were effective at 5 nM, and nikkomycin production was fully restored to WT strain levels at 100 nM. In contrast, the minimum effective concentration of SAB2 required for eliciting nikkomycin biosynthesis was 500 nM (Fig. 2D), suggesting that the side chain of SABs is key in structure-activity relationships.

SABs and their receptor SabR1 coordinately regulate nikkomycin biosynthesis
Autoregulators usually perform their regulatory functions on secondary metabolism and morphological differentiation through specific receptor proteins. By multiple-sequence alignment and phylogeny evolution analysis, SabR1 appears to belong to GBL receptor family.
To verify this hypothesis, sabR1 was disrupted in WT S. ansochromogenes to generate ⌬sabR1 and also in SABs-defective mutant (⌬sabA) to generate ⌬sabA/⌬sabR1. Nikkomycin production was not affected in ⌬sabR1, where SABs are present. However, nikkomycin production was abolished in ⌬sabA due to the absence of SABs, which could be reversed by further disruption of sabR1 in ⌬sabA. The complementation of sabR1 in ⌬sabA/⌬sabR1 (⌬sabA/⌬sabR1c) completely repressed nikkomycin biosynthesis (Fig. 3, A and B). Accordingly, qRT-PCR revealed that the transcription of key regulatory genes adpA and sanG for nikkomycin biosynthesis as well as the structural genes sanF, sanN, and sanO in nik cluster was reduced in ⌬sabA, but restored in ⌬sabA/⌬sabR1 (Fig. 3C). These experiments clearly demonstrated that the inhibition of nikkomycin production in ⌬sabA resulted from SabR1-dependent repression of the transcription of nikkomycin biosynthetic genes. Similarly, in the absence of SABs, SabR1 repressed the transcription of sabR1, sabR2, and sabA in ⌬sabA but not in the WT strain, where SABs are present (Fig. S5). Taken together, it was suggested that SABs are associated with SabR1 to regulate nikkomycin biosynthesis.
To verify the recognition of SABs by SabR1, electrophoretic mobility-shift assays (EMSAs) were performed using SabR1-His 6 and potential target gene promoter regions as probes. Complexes of SabR1 with the upstream regions of sabA, sabR1, or sabR2 were formed in a concentration-dependent manner (Fig. 4A), whereas binding of SabR1 to the upstream regions of sanF, sanG, or adpA was not observed (data not shown), implying that SabR1 regulates nikkomycin biosynthesis indirectly. When SABs were added into the EMSA reaction mixture, the complex could not be formed, demonstrating that sabA, sabR1, and sabR2 are the direct targets of SabR1, and SABs can interact with SabR1 to cause its dissociation from the target DNA (Fig.  4B). Thus, SabR1 was determined as the cognate receptor of SABs in vitro.
To identify the precise binding sequences of SabR1 on its target genes, DNase I footprinting experiments were performed. The results showed that SabR1 protected regions from Ϫ137 to Ϫ111 bp relative to the sabA translation start site (tss), Ϫ118 to Ϫ88 bp relative to the sabR1 tss, and Ϫ282 to Ϫ254 bp relative to the sabR2 tss ( Fig. 4, C-E). Analysis of the three binding sites using the MEME program (21) revealed a conserved consensus sequence (5Ј-AAWAVAAACCRGDBRD-TYSGTWY-3Ј) (Fig. 4F). In agreement with EMSAs, this consensus sequence was not found in the promoters of sanF, sanG, and adpA, confirming that SabR1 regulates nikkomycin biosynthesis indirectly.

cprC is directly repressed by SabR1 and positively regulates nikkomycin biosynthesis
As mentioned above, SabR1 regulates nikkomycin production indirectly via unknown intermediates. One candidate is sabR2, a target gene of SabR1, encoding a TetR family regulator, which is potentially an intermediate protein between SabR1 and nikkomycin biosynthetic genes. However, sabR2 showed no impact on nikkomycin production after its disruption ( Fig.  S6), implying that SabR1 exerts its regulatory function on nikkomycin biosynthesis through other means. To explore more SabR1 potential target genes, S. ansochromogenes genome scanning was undertaken using the conserved 23-bp SabR1 binding consensus. This revealed another TetR-family regulator gene, cprC, whose encoded protein CprC shows 68% identity with CprA and 70% identity with CprB from S. coelicolor (22). The promoter region of cprC (P cprC ) contains a SabR1 conserved binding motif 5Ј-AAAACAAACCGCATGTCCT-GTTCTTTTG-3Ј (Fig. 5A). EMSA experiments confirmed that P cprC could be recognized by SabR1 to form a complex. When P cprC was mutated to 5Ј-GCGGCCGCCCGCATGTCCT-GTTCTTTTG-3Ј at the specific sites shown in boldface type, SabR1 no longer bound the mutated P cprC (Fig. 5, B and C). Thus, cprC is confirmed as a direct target gene of SabR1.
Analyses by qRT-PCR showed that the transcription of cprC was dramatically reduced in ⌬sabA but restored in ⌬sabA/ ⌬sabR1, suggesting that SabR1 is a repressor of cprC (Fig. 5D). Along with the EMSA results, it was verified that SabR1 represses the transcription of cprC by directly binding to the promoter region, which was also further illustrated by the

Butenolide-mediated regulation of nikkomycin biosynthesis
expression of gusA encoding a ␤-glucuronidase reporter system (Fig. 5E). It can be hypothesized that the decreased transcription of cprC may lead to abolition of nikkomycin production in ⌬sabA.
To ascertain the effect of CprC on nikkomycin production, disruption of cprC was performed to generate ⌬cprC. The yield of nikkomycin in ⌬cprC was notably decreased compared with that in WT, suggesting that CprC is an important activator for nikkomycin production. In addition, cprC was complementarily expressed in ⌬sabA under the control of hrdB promoter (P hrdB ), and the resulting strain (⌬sabA/cprCoe) efficiently restored nikkomycin production (Fig. 6). Thus, CprC was confirmed to be an activator for nikkomycin biosynthesis as the target of SabR1.

CprC directly activates the transcription of adpA
To understand how CprC activates nikkomycin biosynthesis, EMSAs were performed using purified CprC-His 6 and the promoters of potential target genes, adpA, sanG, and sanF. It was shown that CprC binds to the upstream region of adpA (Fig. 7A), but not to the promoter regions of sanG and sanF (data not shown). DNase I footprinting experiments revealed that the protected sequence on adpA promoter by CprC is 5Ј-CACCC-GGGGCCAGACACCGGTCGACTGACCTGTTTTC-3Ј, from Ϫ406 to Ϫ369 bp relative to the tss of adpA (Fig. 7, B and C). Therefore, adpA is verified to be a target gene of CprC.
Analyses by qRT-PCR revealed that the transcription of adpA was significantly decreased in ⌬cprC (Fig. 7D), confirming that CprC is an activator of adpA in S. ansochromogenes.

Butenolide-mediated regulation of nikkomycin biosynthesis
Remarkably, upon adpA overexpression under the control of P hrdB in ⌬cprC (⌬cprC/adpAoe), nikkomycin production was restored compared with ⌬cprC (Fig. 7, E and F), demonstrating that CprC activation on adpA transcription is an important route in regulating nikkomycin biosynthesis.
Taken together, these findings enable us to propose a plausible regulatory model of the SABs-SabR1 signaling system in S. ansochromogens. sabR1, encoding a receptor of signaling molecule, is situated in a minicluster sab, in which sabAPD are signaling molecule biosynthetic genes. SabR1 can repress the transcription of cprC, which can be relieved by signaling molecules synthesized by SabAPD. CprC is newly identified as an activator of adpA, whereas AdpA positively regulates nikkomycin production by activating the cluster-situated regulatory gene sanG. In this complicated regulatory system, CprC and AdpA are key factors that transmit butenolide signals to secondary metabolism and thereby regulate nikkomycin biosynthesis (Fig. 8).

Discussion
The dramatic rise in drug-resistant pathogens creates an urgent need for effective alternatives. Bacteria of the genus Streptomyces are a particularly abundant source of antibiotics, and their signaling systems are appealing for their critical roles in the regulation of secondary metabolism, particularly in antibiotic biosynthesis. The discovery of new signaling molecules and elucidation of their regulatory roles would have great significance for their use. In this work, a novel member of the butenolide autoregulators, SAB1, along with two analogues, SAB2 and SAB3, were characterized. A cascade regulation of nikkomycin biosynthesis mediated by SABs and its receptor SabR1 via a newly discovered adpA activator CprC was revealed for the first time, which may be applicable for other Streptomyces, as SabA and CprC homologs are widely distributed. Moreover, butenolide regulation of global regulator AdpA indicates their potential pleiotro-

Butenolide-mediated regulation of nikkomycin biosynthesis
pic effect. These findings provide new insights into the butenolide signaling system and also facilitate the application of signaling molecules in natural product discovery.
Genome sequencing has revealed the presence of a large number of cryptic secondary metabolic gene clusters in Streptomyces. Such pathways are usually under stringent control by a variety of different regulatory mechanisms, in which signaling molecules might play important roles. A lack of specific signaling molecules, such as ␥-butyrolactones and butenolides, may be a factor causing the silence of gene clusters (23). Characterization of signaling molecules or autoregulators is essential to evaluate their significance in antibiotic production and morphological differentiation. However, this task is hindered by the low yield of autoregulators in producing strains, which makes their study by large-scale fermentation rather burdensome and inefficient (1). It has been evidenced that AfsA and its homologs as well as tailoring enzymes are responsible for the biosynthesis of various signaling molecules, such as GBLs, furans, and some butenolides (4,6,24). In this study, heterologous expression of afsA homologue sabA and two tailoring enzyme genes, sabP and sabD, of S. ansochromogenes significantly improved the signal molecule yield, and three different butenolides were characterized. SAB1, SAB2, and SAB3 share a common ring skeleton but with different side chains at the C-3 position. In addition, more than one configurational isomer of SABs was visible on NMR spectra. SAB1 and SAB3 are more effective than SAB2 in nikkomycin production, reflecting the structure-activity relationship. Thus, heterologous expression of the butenolide biosynthetic genes proved to be an effective and feasible approach for producing higher yield and diverse structures of signal molecules, which can confer on them different receptor binding affinity and regulatory activity on antibiotic biosynthesis. Interestingly, genome mining revealed that the homologs of SabA are present in Streptomyces lavenduligriseus, Streptomyces achromogenes, Streptomyces sp. NRRL S-31, and so on, suggesting that SAB-like autoregulators may exist widely in Streptomyces.
Autoregulators usually function via binding to cognate receptor proteins, and their effects can be transmitted in cascade The underlined red letters are the predicted binding site of SabR1 on P cprC (ARE-C) . B, the underlined ARE-C site in P cprC was mutated to generate the mutant promoter region (P cprCM ) of cprC. C, EMSA of SabR1 binding to fluorescently labeled P cprC and P cprCM . Each lane contains 50 ng of labeled probes and 1 g of poly(dI-dC). S, unlabeled specific probe P cprC (30-fold) was added; N, unlabeled nonspecific probe P hrdB (30-fold) was added. D, qRT-PCR transcriptional analysis of cprC in WT, ⌬sabA, and ⌬sabA/⌬sabR1. Error bars, S.D. calculated from three independent experiments. E, GusA activity in derivatives of S. coelicolor M1146 containing various promoters fused with gusA. P C G and P CM G, S. coelicolor M1146 containing P cprC or P cprCM fused with gusA. P C G-BR1 and P CM G-BR1, strains P C G and P CM G containing P hrdB fused with sabR1. P C G-R1 and P CM G-R1, strains P C G and P CM G containing sabR1 with its own promoter.

Butenolide-mediated regulation of nikkomycin biosynthesis
fashion to specific pathways by a series of regulators. The wellexemplified representative signaling system is GBLs. For butenolides, a total of 10 members (including SAB1, -2, and -3 identified in this study) of this family of signaling molecules from five Streptomyces species have been discovered to date, but their signal transduction system has been the subject of little investigation. In S. avermitilis, avenolide receptors AvaR1 and AvaR2 were verified, and AvaR2 directly repressed the transcription of aveR, a CSR activator gene for avermectin production (25)(26)(27). Recently, four compounds structurally resembling avenolide were found in Streptomyces albus strain J1074 with different avenolide-like activity (3). In S. rochei, SRB1 and SRB2 activated the production of lankacidin and lankamycin but showed a negative effect on morphological differentiation. Their receptor, SrrA, controls two SARP regulators, SrrY and SrrZ, but the exact molecular mechanisms of these regulators activating antibiotic production and morphological differentiation still remain obscure (28,29). Moreover, four butenolide compounds were isolated from S. antibioticus DSM40725, but their activities on antibiotic biosynthesis have not yet been reported in detail (20). It is noteworthy that the regulation mediated by the above butenolides, avenolide or SRBs, is AdpA-independent. Interestingly, our work indicated that SabR1 does not target to the CSRs, but can regulate adpA via an activator CprC, and the established signaling system might be significant for elucidating complicated regulatory network in antibiotic biosynthesis.
CprC, a new activator of adpA discovered in this work, is a member of the TetR-family regulators. BLAST searches revealed that its homologs are widespread in Streptomyces, but only CprA and CprB have been studied. CprA and CprB are two ArpA-like proteins in S. coelicolor and possess regulatory func-   genes (sabR1, sabR2,  and sabA). CprC is a new activator of adpA, and AdpA positively regulates nikkomycin production by activating the cluster-situated regulatory gene sanG. In WT strain, binding of SABs to SabR1 causes the dissociation of SabR1 from cprC promoter and increases crpC transcription, which in turn activates adpA transcription to trigger nikkomycin biosynthesis. When sabA is disrupted or SABs are absent, binding of SabR1 to cprC would result in the repression of cprC and consequently cause the transcriptional reduction of adpA and nikkomycin biosynthetic genes.

Butenolide-mediated regulation of nikkomycin biosynthesis
tions in both secondary metabolism and morphogenesis in S. coelicolor A3 (2). CprA stimulates actinorhodin and undecylprodigiosin biosyntheses and sporulation, whereas CprB represses actinorhodin biosynthesis and sporulation (22). However, their mechanism of action has not been revealed. Here we demonstrated that CprC directly binds to the adpA promoter region to activate its transcription, which may be also applicable for CprC homologs in other Streptomyces species. Interestingly, a signal binding domain at the C terminus of these proteins is present, but how they coordinate with small molecules to influence the expression of target genes is unclear.
It will be of interest to verify the ligand of CprC and dissect the coordinated regulatory network in future studies.

Bacterial strains, plasmids, and growth conditions
Bacteria strains and plasmids used in this study are listed in Table 1. S. ansochromogenes and its derivative strains were grown at 28°C on MS medium agar for sporulation and in SP medium for nikkomycin production (11). E. coli strains were grown in lysogeny broth medium at 37°C containing corresponding antibiotics (16).

Construction of disruption and complementation mutants
Gene disruption mutants were constructed via homologous recombination. The plasmids for disruption, complementation, and overexpression were first constructed in E. coli JM109 with appropriate primers for individual genes as shown in Table  S2 and then conjugally transferred into Streptomyces through ET12567/pUZ8002.
To inactivate sabA, the fragment corresponding to the upstream region of sabA was amplified using the primer pairs sabALF/R, and the resulting product was digested with EcoRI/ XbaI. The DNA fragment corresponding to the downstream region of sabA was amplified using primer pair sabARF/R and followed by XbaI/HindIII digestion. The two fragments were then ligated into the EcoRI/HindIII-digested sites of pKC1139 to generate plasmid pKC1139AD. Primers sabAcF/R were used to verify the mutant ⌬sabA. To complement sabA, a 1.55-kb XbaI/BamHI-digested fragment containing the intact sabA and its own promoter region was inserted into pSET152 to give plasmid pSET152::sabA. Primers sabAcF/R were used to confirm the complementary strain ⌬sabAc. The construction of disruption strains of sabR1, sabR2, and cprC and the complement strains of sabR1 and cprC was performed as mentioned above using appropriate primer pairs as shown in Table S2.
To overexpress cprC in ⌬sabA, a 714-bp fragment containing the coding sequence of cprC was amplified using primers cprCoF/R, and the constitutive promoter P hrdB of S. coelicolor was amplified using primer pair PhrdBF/R. After digestion with NdeI/EcoRI and XbaI/NdeI, respectively, the two fragments were inserted into the EcoRI/XbaI-digested sites of pSET152 to generate pSET152::hcprC. Primers PhrdBF and cprCoR were used to confirm the strain ⌬sabA/cprCoe. The overexpression strains of sabR2 in ⌬sabA and adpA in ⌬cprC were constructed as mentioned above using appropriate primer pairs as shown in Table S2.

Expression and purification of SabR1 and CprC
SabR1 coding region was amplified by PCR using primers SabR1F/R (Table S2). After digestion with NdeI and XhoI, the amplified fragments were inserted into the same sites of pET-23b to give plasmid pET23b::sabR1. The plasmid pET23b::cprC for expression and purification of CprC-His 6 was constructed as mentioned above with suitable primers as shown in Table  S2. Plasmids pET23b::sabR1 and pET23b::cprC were transferred into E. coli C41 for expression of SabR1-His 6 and CprC-His 6 . Protein purification was performed with a nickel-nitrilotriacetic acid-agarose column as described previously (30).

Construction of plasmids pET23b::sabAPD and pKC1139::hsabAPD
To construct the plasmids for heterologous expression of sabAPD, a 0.96-kb fragment containing sabA was amplified using the primer pairs sabAF/R, and a 1.56-kb fragment containing sabPD was amplified using the primer pairs sab-PDF/R. The two fragments were digested with NdeI/XbaI and XbaI/EcoRI, respectively, and then cloned into the NdeI/EcoRI-digested sites of pET23b to generate plasmid pET23b::sabAPD. It was transferred into E. coli C41 to generate CpAPD. A 2.5-kb fragment containing sabAPD digested with NdeI/EcoRI from pET23b::sabAPD and P hrdB of S. coelicolor digested with XbaI/NdeI were inserted into the EcoRI/XbaI-digested sites of pKC1139 to generate plasmid pKC1139::hsabAPD, which was transferred into S. coelicolor M1146 to generate MpAPD.

Purification and structural analyses of SABs
Seed broth (60 ml) of CpAPD grown in lysogeny broth medium at 37°C for 12 h was inoculated into 3 liters of M9 medium. After incubation at 37°C for 4 h, isopropyl 1-thio-␤-D-galactopyranoside was added at a final concentration of 0.1 mM and further incubated at 28°C for 4 h, followed by another 12 h at 37°C with an airflow rate of 3 liters/min and agitation speed of 200 rpm. The culture filtrate was extracted twice with equal volumes of ethyl acetate, and the organic phase was dried and redissolved in methanol. SAB1 was isolated from the extract by three rounds of HPLC separation (Zorbax, SB-C18, 9.4 ϫ 250 mm, 5 m) with a flow rate of 3 ml/min at 210-nm detection wavelength. The elution gradient was set as follows: 31% methanol for 15 min on first HPLC, followed by a linear gradient of 5-20% acetonitrile in 40-min elution on second HPLC, and finally 5-100% methanol in a 30-min linear gradient elution.
Seed broth (60 ml) of MpAPD grown in YEME medium at 28°C for 48 h was inoculated into 3 liters of modified AlaMM medium (24) (supplemented with 10 g/liter mannitol and 3 g/liter casaminoacids, pH 6.0) with an airflow rate of 0.8 liter/min and agitation speed of 200 rpm. After incubation at 28°C for 5 days, the culture filtrate was processed with the same procedure as that of CpAPD. To separate SAB1, -2, and -3, the elution profile was set as follows: 10 -40% methanol for 30 min and then 40 -60% methanol for 10 min. SAB1, -2, and -3 were eluted at 25.7-27 min, 24.2-25.7 min, and 34.7-36.2 min, respectively. The collected fractions of SAB1 were further purified with the elution profile of 14% acetonitrile in 20 min, SAB2 with a linear gradient of 14 -16% acetonitrile in 20 min, and SAB3 with a linear gradient of 22-23% acetonitrile in 20 min.
Mass spectral analysis was performed on a triple quadrupole LC/MS/MS system (Agilent 1260/6460) in positive mode with an Agilent ZORBAX SB-C18 column (3.5 m, 2.1 ϫ 100 mm). 1 H and 13 C NMR spectra as well as 1 H-1 H COSY, 1 H-13 C HMBC, and 1 H-13 C HSQC were recorded on a 500-MHz Bruker spectrometer using CDCl 3 as solvent.

EMSAs and DNase I footprinting
The EMSAs and DNase I footprinting assays were performed as described (16,31). All probes for EMSAs were amplified by PCR using the corresponding primer pairs listed in Table S2. For EMSAs with unlabeled probes, SabR1 was incubated with 20-ng probes in a 20-l reaction mixture at 25°C for 30 min, and then the samples were loaded on 4% (w/v) native polyacrylamide gels for electrophoresis. The gel was stained with SYBR Gold nucleic acid gel stain for 30 min and photographed under UV transillumination using Quantity One. EMSAs with fluorescently labeled probes were carried out as follows. To obtain the fluorescently labeled DNA, the promoter regions of target genes were individually inserted into the EcoRV site of plasmid M13. The resulting plas-Butenolide-mediated regulation of nikkomycin biosynthesis mids were used as templates for probe amplification using fluorescently labeled primers FAM-F/HEX-R. 50-ng labeled probes were incubated with SabR1 in a 20-l reaction mixture containing 1 g of poly(dI-dC) at 25°C for 30 min. The fluorescently labeled DNA was detected by Tanon-5200 Multi. For competition assays, 1.5 g of unlabeled specific probe or the nonspecific probe P hrdB was added into the binding reaction mixture. The EMSAs of CprC binding to DNA probes were performed similarly as mentioned above with appropriate probes.

RNA isolation and real-time quantitative PCR
Total RNA was isolated from cultures of S. ansochromogenes and its derivative strains at various time points. RNA isolation, genomic DNA removal, reverse transcription, and qRT-PCR were performed as described previously (30). All of the primers used were listed in Table S2. 23S rRNA of S. ansochromogenes was used as internal control.

Bioassay and HPLC analysis of nikkomycin
The bioassay and HPLC of nikkomycin in culture filtrates of S. ansochromogenes and its derivative strains was carried out as described previously (12).

Bioassays of SABs
The activity of SABs in vivo was determined by the ability to restore nikkomycin production in ⌬sabA. Different concentrations of SABs were added into 50-ml cultures of ⌬sabA at the beginning of fermentation. After incubation for 5 days, nikkomycin production was detected by HPLC as described above.

gusA transcriptional fusion assays
To confirm the binding activity of SabR1 to P cprC , two reporter plasmids containing P cprC or P cprCM -gusA fusions were constructed. P cprC was first amplified using the primers PGcprCF/ PGcprCR. To obtain the template of P cprCM (mutant of P cprC at the binding motif), two fragments amplified by PCR using PcprCF/ PcprCMR and PcprCMF/PcprCR were digested with NotI and then ligated by T4-ligase. The resulting mixture was diluted 10 times and then used as the template of P cprCM . P cprCM was amplified using the primers PGcprCF/PGcprCR. gusA was amplified from pGUS using primer pair gusA-F/R. Then the P cprC or P cprCM digested with SpeI/NdeI and gusA with NdeI/XhoI were cloned into the SpeI/XhoI-digested sites of pIJ10500 to generate plasmid pIJ10500::cgusA and pIJ10500::cmgusA. Then the two constructs were respectively introduced and integrated into the ⌽BT1 attB site of S. coelicolor M1146. Subsequently, pSET152::sabR1 and pSET152::hsabR1 were respectively introduced into the two S. coelicolor M1146 derivatives containing different reporter constructs and integrated into the ⌽C31 attB site. GusA activity was detected as described previously (32).