Genome Mining and Metabolic Profiling Uncover Polycyclic Tetramate Macrolactams from Streptomyces koyangensis SCSIO 5802

We have previously shown deep-sea-derived Streptomyces koyangensis SCSIO 5802 to produce two types of active secondary metabolites, abyssomicins and candicidins. Here, we report the complete genome sequence of S. koyangensis SCSIO 5802 employing bioinformatics to highlight its potential to produce at least 21 categories of natural products. In order to mine novel natural products, the production of two polycyclic tetramate macrolactams (PTMs), the known 10-epi-HSAF (1) and a new compound, koyanamide A (2), was stimulated via inactivation of the abyssomicin and candicidin biosynthetic machineries. Detailed bioinformatics analyses revealed a PKS/NRPS gene cluster, containing 6 open reading frames (ORFs) and spanning ~16 kb of contiguous genomic DNA, as the putative PTM biosynthetic gene cluster (BGC) (termed herein sko). We furthermore demonstrate, via gene disruption experiments, that the sko cluster encodes the biosynthesis of 10-epi-HSAF and koyanamide A. Finally, we propose a plausible biosynthetic pathway to 10-epi-HSAF and koyanamide A. In total, this study demonstrates an effective approach to cryptic BGC activation enabling the discovery of new bioactive metabolites; genome mining and metabolic profiling methods play key roles in this strategy.


Introduction
Microbially produced natural products (NPs) play an important role in drug discovery, especially with respect to initiatives aimed at developing new agents for combatting infectious disease [1]. Increasing numbers and expanding structural diversity of new bioactive NPs identified from marine-derived Streptomyces suggest the increasing importance of marine-derived microbes as an important reservoir for new NPs [2]. Additionally, the rapid development of whole-genome sequencing technology has made it clear that microorganisms have great, though previously underappreciated potential to produce new bioactive NPs by virtue of biosynthetic gene clusters (BGCs) that remain "silent" under standard laboratory culture conditions [3]. For instance, around a dozen types of secondary metabolites had been discovered from the model organism Streptomyces coelicolor A3(2) before the publication of its genome sequence in 2002; however, seven additional metabolites were discovered by genome mining [4]. In light of this realization, genome mining has become a powerful tool by which to identify BGCs of interest which might then be activated to produce NPs of interest [5]. Indeed, many strategies to identify new natural products have benefitted from genome mining [6]; these include OSMAC (one strain-many compounds) [7], heterologous BGC expression [8], as well as manipulation of global or biosynthetic pathway-specific regulator strategies [9]. Recently, we investigated genetic inactivation of the biosynthetic genes of a microbe's predominant products as a means generating new otherwise absent NPs. This approach also employs genome mining and has been demonstrated to be a good strategy to direct metabolic flux towards other biosynthetic routes, thus activating a silent gene cluster. For instance, genetic inactivation of the biosynthetic genes responsible for rishirilide and xiamycin production in Streptomyces olivaceus SCSIO T05 activated the production of lobophorin CR4 [10]. In another example, the production of nocardamine [11] and atratumycin [12] in Streptomyces atratus SCSIO ZH16 was achieved by inactivation of ilamycin biosynthesis.
Polycyclic tetramate macrolactams (PTMs) are a unique class of natural products with a variety of valuable biological activities such as antibiotic, antifungal, antiprotozoal, and antitumor properties [13]. PTMs consist of a tetramate-containing macrocyclic lactam ring and a polycyclic carbocycle with a 5/5, 5/6, 5/5/6, or 5/6/5 ring system [14] ( Figure  S1). Multiple asymmetric centers in the PTMs greatly enhance their structural diversity and contribute to a wide range of biological activities [13,15,16]. The significant biological activities and intriguing chemical structures of PTMs have attracted a great deal of interest from synthetic and biosynthetic chemists. However, owing to the multiple stereocenters in representatives of the family, biosynthesis has thus far proven the most cost effective and efficient means of accessing the PTM scaffold. Detailed analysis of the genome sequence of the deep-sea-derived Streptomyces koyangensis SCSIO 5802 revealed the presence of at least 21 gene clusters including one putative PTM-encoding gene cluster. Previous chemical investigations of this strain have enabled the discovery of two types of active secondary metabolites, abyssomicins and candicidins; however, the PTM representatives have not been detected. In this paper, to activate the PTM gene cluster, the biosynthetic gene clusters for abyssomicin/candicidin were genetically inactivated. As a result, two PTMs, the known 10-epi-HSAF (1) and a new compound, koyanamide A (2), were isolated from the large scale fermentation of the mutant strain and their structures were unambiguously determined by extensive analyses of HRESIMS, 1D and 2D NMR, electronic circular dichroism (ECD) spectroscopic data. In addition, we also performed detailed analysis of the PTM gene cluster and proposed a biosynthetic pathway of PTMs in S. koyangensis SCSIO 5802.

Genome Features and Annotation of the Strain S. koyangensis SCSIO 5802
Whole genome sequencing is an important tool in efforts to discover new NPs [10,17]. To understand the full biosynthetic landscape of S. koyangensis SCSIO 5802, the microbe's complete genome has been sequenced in previous study [18]. We reported and analyzed the genome data in this study. The SCSIO 5802 genome consists of a 6,861,302-bp circular chromosome and a 241,660-bp circular plasmid with an average G + C content of 73.29%. This complete genome was predicted to contain 6,419 protein-coding genes, 21 rRNA genes, and 65 tRNA genes ( Figure 1 and Table 1). To identify putative gene clusters within the SCSIO 5802 genome, the complete sequence was interrogated for secondary metabolite BGCs using online antiSMASH software [19]. The analysis revealed that a total of 21 putative gene clusters are contained within the SCSIO 5802 genome, including three polyketide synthase (PKS, Type I and Type III), four nonribosomal peptide synthetase (NRPS), two hybrid PKS-NRPS, three terpene, two bacteriocin, three saccharide, one lanthipeptide BGCs and two other categories of BGCs (Table 2). This finding suggests that SCSIO 5802 has the potential to produce at least 21 categories of NPs. Among the 21 predicted BGCs, cluster 18 was proposed to be involved in the biosynthesis of PTMs, since it shows high similarity to previously reported PTM gene clusters. However, only two types of secondary metabolites, abyssomicins and candicidins ( Figure S2), have been isolated to date [20][21][22]. Thus, these analyses suggested that S. koyangensis SCSIO 5802 represented an outstanding candidate strain to mine PTM-type compounds. To identify putative gene clusters within the SCSIO 5802 genome, the complete sequence was interrogated for secondary metabolite BGCs using online antiSMASH software [19]. The analysis revealed that a total of 21 putative gene clusters are contained within the SCSIO 5802 genome, including three polyketide synthase (PKS, Type I and Type III), four nonribosomal peptide synthetase (NRPS), two hybrid PKS-NRPS, three terpene, two bacteriocin, three saccharide, one lanthipeptide BGCs and two other categories of BGCs (Table 2). This finding suggests that SCSIO 5802 has the potential to produce at least 21 categories of NPs. Among the 21 predicted BGCs, cluster 18 was proposed to be involved in the biosynthesis of PTMs, since it shows high similarity to previously reported PTM gene clusters. However, only two types of secondary metabolites, abyssomicins and candicidins ( Figure S2), have been isolated to date [20][21][22]. Thus, these analyses suggested that S. koyangensis SCSIO 5802 represented an outstanding candidate strain to mine PTM-type compounds.   [20,21] (Figure 2, traces i-ii). We hypothesized that this large metabolic background might obscure the presence of other, less abundant secondary metabolites. In our previous study, we had characterized AbmB1 as the PKS responsible for the assembly of the abyssomicin polyketide backbone. To enhance both the presence and detectability of other secondary metabolites, we have therefore generated an in-frame deletion mutant designated S. koyangensis SCSIO 5802A, devoid of the β-ketoacyl synthase (KS) and acyl transferase (AT) domains of abmB1 [18]. HPLC analyses have revealed that the SCSIO 5802A mutant fails to produce abyssomicin [18]. Instead, the SCSIO 5802A mutant generated the polyenic antibiotics, candicidins (7-9) [22][23][24], as the major metabolites ( Figure 2, traces iii-iv). To further minimize the metabolic background of the genetically engineered mutant S. koyangensis SCSIO 5802A, 1.8 kb of the PKS gene canD [22] responsible for the construction of the polyketide chain of candicidins was deleted using the in-frame format, thus affording the in-frame "double-deletion" mutant designated S. koyangensis SCSIO 5802AC ( Figure S3). HPLC analyses of the fermentation extract revealed that this genetically engineered strain produced neither the candicidins nor the abyssomicins ( Figure 2, traces v-vi). Notably, a group of secondary metabolites with UV spectra distinct from those of the abyssomicins and candicidins now constituted the major metabolites produced by S. koyangensis SCSIO 5802AC. The ablation of both candicidin and abyssomicin biosynthesis enabled the production and visualization of these newly produced compounds in quantities sufficient for structure elucidation.
S. koyangensis SCSIO 5802 predominantly produces abyssomicins (3-6) under standard laboratory culture conditions [20,21] (Figure 2, traces i-ii). We hypothesized that this large metabolic background might obscure the presence of other, less abundant secondary metabolites. In our previous study, we had characterized AbmB1 as the PKS responsible for the assembly of the abyssomicin polyketide backbone. To enhance both the presence and detectability of other secondary metabolites, we have therefore generated an in-frame deletion mutant designated S. koyangensis SCSIO 5802A, devoid of the β-ketoacyl synthase (KS) and acyl transferase (AT) domains of abmB1 [18]. HPLC analyses have revealed that the SCSIO 5802A mutant fails to produce abyssomicin [18]. Instead, the SCSIO 5802A mutant generated the polyenic antibiotics, candicidins (7-9) [22][23][24], as the major metabolites ( Figure 2, traces iii-iv). To further minimize the metabolic background of the genetically engineered mutant S. koyangensis SCSIO 5802A, 1.8 kb of the PKS gene canD [22] responsible for the construction of the polyketide chain of candicidins was deleted using the in-frame format, thus affording the in-frame "double-deletion" mutant designated S. koyangensis SCSIO 5802AC ( Figure S3). HPLC analyses of the fermentation extract revealed that this genetically engineered strain produced neither the candicidins nor the abyssomicins ( Figure 2, traces v-vi). Notably, a group of secondary metabolites with UV spectra distinct from those of the abyssomicins and candicidins now constituted the major metabolites produced by S. koyangensis SCSIO 5802AC. The ablation of both candicidin and abyssomicin biosynthesis enabled the production and visualization of these newly produced compounds in quantities sufficient for structure elucidation. . Due to differences in the UV absorption wavelengths between 1-6 and 7-9, all chromatograms employed UV detections at both 254 nm and 400 nm. Compounds 1 and 2 are 10-epi-HSAF and koyanamide A. Compounds 3-9 were previously identified as abyssomicin 4, neoabyssomicin B, abyssomicin 2, neoabyssomicin A, candicidin A3, candicidin D, and candicidin A1, respectively.

Fermentation, Isolation, and Structural Elucidation of PTMs From S. koyangensis SCSIO 5802AC
To elucidate the two newly generated NPs, a 21 L scale fermentation of the S. koyangensis SCSIO 5802AC mutant was performed with a two-step fermentation process as previously reported [20]. The culture broth was extracted with butanone, and the mycelial cake was extracted with acetone. Silica gel chromatography and RP-HPLC purification methods afforded analytically pure samples of 1 and 2.

Fermentation, Isolation, and Structural Elucidation of PTMs from S. koyangensis SCSIO 5802AC
To elucidate the two newly generated NPs, a 21 L scale fermentation of the S. koyangensis SCSIO 5802AC mutant was performed with a two-step fermentation process as previously reported [20]. The culture broth was extracted with butanone, and the mycelial cake was extracted with acetone. Silica gel chromatography and RP-HPLC purification methods afforded analytically pure samples of 1 and 2.
The molecular formula of 1 was determined as C 29 Figure S4). The planar structure of 1 was determined to be the same as that of 10-epi-HSAF [25,26], on the basis of NMR spectroscopic data comparisons for 1 (Table S3; Figures S5-S11) and 10-epi-HSAF [25,26]. The relative configurations of 1 were assigned by NOESY correlations and then compared with 10-epi-HSAF [25,26]. Very obvious NOESY correlations of H-5/H-13, H-6/H-8, H-6/H-14, H-6/H-16, H-8/H-10, and H-10/H-12 were observed; H-10 and H-11 were assigned to be trans-oriented for 1. In parallel, we also measured the electronic circular dichroism (ECD) spectrum of 1, which was found to be in excellent agreement with the calculated ECD spectrum of 5S, 6R, 8R, 10R, 11S, 12S, 13R, 14S, 16S, 23S, 25S and 10-epi-HSAF ( Figure S12). Thus, the structure of 1 was consistent with 10-epi-HSAF ( Figure 3A) [25,26].  Figure S13). A comprehensive analysis of the NMR data for 2 revealed that the structure of 2 was similar to that of 1 (Figures S14-S20). A careful comparison suggested that the C2-C3 double bond (δH 5.70 and 5.89; δC 124.1 and 139.1) and C17-C18 double bond (δH 6.57 and 6.86; δC 150.2 and 121.3) in 1 were absent in 2; instead, the C3-C4 double bond (δH 5.46 and 5.56; δC 125.9 and 138.3) and C16-C17 double bond (δH 5.75 and 5.67; δC 138.2 and 126.1) were found in 2. The coupling constants (J3,4 = 9.1 Hz and J16,17 = 8.8 Hz) indicated that these double bonds are in the Z configurations. The 1 H-1 H COSY spectrum of 2 indicated the presence of carboncarbon bonds from C2 to C18 and from C5 to C15 ( Figure 3B). Thus, we deduced that 2 contains a cyclooctadiene unit. The rest of the NMR data for both 1 and 2 were almost identical. These analyses suggested that 2 is likely derived from intermediate IV via an intramolecular photochemical [4 + 4] cycloaddition, as has been reported for alteramide A ( Figure 4A and 4C) [27]. From a biosynthetic perspective, absolute configurations of the first two five-membered rings, C14, C23 and C25 in 2 should be the same as those found in 1. The absolute configurations of the four remaining stereogenic centers at C2, C5, C15 and C18 in 2 were established with the use of NOESY spectra; correlations of H-2/H-18, H-5/H-15, and H-5/H-6 allowed the determination of the absolute configurations at these centers to be as 2S, 5S, 15S, 18S ( Figure 3B). Thus, compound 2 was named koyanamide A.  Figure S13). A comprehensive analysis of the NMR data for 2 revealed that the structure of 2 was similar to that of 1 (Figures S14-S20). A careful comparison suggested that the C2-C3 double bond (δ H 5.70 and 5.89; δ C 124.1 and 139.1) and C17-C18 double bond (δ H 6.57 and 6.86; δ C 150.2 and 121.3) in 1 were absent in 2; instead, the C3-C4 double bond (δ H 5.46 and 5.56; δ C 125.9 and 138.3) and C16-C17 double bond (δ H 5.75 and 5.67; δ C 138.2 and 126.1) were found in 2. The coupling constants (J 3,4 = 9.1 Hz and J 16,17 = 8.8 Hz) indicated that these double bonds are in the Z configurations. The 1 H-1 H COSY spectrum of 2 indicated the presence of carboncarbon bonds from C2 to C18 and from C5 to C15 ( Figure 3B). Thus, we deduced that 2 contains a cyclooctadiene unit. The rest of the NMR data for both 1 and 2 were almost identical. These analyses suggested that 2 is likely derived from intermediate IV via an intramolecular photochemical [4 + 4] cycloaddition, as has been reported for alteramide A ( Figure 4A,C) [27]. From a biosynthetic perspective, absolute configurations of the first two five-membered rings, C14, C23 and C25 in 2 should be the same as those found in 1. The absolute configurations of the four remaining stereogenic centers at C2, C5, C15 and C18 in 2 were established with the use of NOESY spectra; correlations of H-2/H-18, H-5/H-15, and H-5/H-6 allowed the determination of the absolute configurations at these centers to be as 2S, 5S, 15S, 18S ( Figure 3B). Thus, compound 2 was named koyanamide A.  [28]. (C) The photochemical [4 + 4] cycloaddition converting alteramide A to intramolecularly cyclized alteramide A has been confirmed [29].

Bioinformatic Analysis of the PTM Gene Cluster and the Proposed Biosynthetic Pathway of PTMs in S. koyangensis SCSIO 5802
Bioinformatic analysis revealed that cluster 18 (termed the sko gene cluster) was proposed to be the gene cluster responsible for the biosynthesis of compounds 1 and 2, since it shows high similarity to previously reported PTM gene clusters, such as SGR810-815 BGC from Streptomyces griseus [28] (Figure 5). The sko cluster was found to contain six genes including a hybrid PKS/NRPS gene skoA, two FAD-dependent oxidoreductases genes skoB1 and skoB2, an alcohol dehydrogenase skoC, a sterol desaturase (SD) gene skoD, and a cytochrome P450 gene skoE ( Figure 5 and Table 3).

Bioinformatic Analysis of the PTM Gene Cluster and the Proposed Biosynthetic Pathway of PTMs in S. koyangensis SCSIO 5802
Bioinformatic analysis revealed that cluster 18 (termed the sko gene cluster) was proposed to be the gene cluster responsible for the biosynthesis of compounds 1 and 2, since it shows high similarity to previously reported PTM gene clusters, such as SGR810-815 BGC from Streptomyces griseus [28] (Figure 5). The sko cluster was found to contain six genes including a hybrid PKS/NRPS gene skoA, two FAD-dependent oxidoreductases genes skoB1 and skoB2, an alcohol dehydrogenase skoC, a sterol desaturase (SD) gene skoD, and a cytochrome P450 gene skoE ( Figure 5 and Table 3).  The center of the gene cluster is the hybrid PKS/NRPS encoding gene skoA; SkoA comprises one PKS module and one NRPS module, making up a total of nine domains (KS-AT-DH-KR-ACP-C-A-PCP-TE). As reported, this type of hybrid PKS/NRPS is conserved among PTM producing bacterial species [32] (Figure 5). The single-module PKS of the hybrid PKS/NRPS enzyme is iteratively used to produce two separate polyketide chains. These chains are, respectively, linked with the αand δamino groups of an Lornithine that is tethered in the NRPS portion to generate a common polyene tetramate intermediate [25,26]. SkoA shows high similarity to the previously characterized PKS/NRPS SGR814 (identity/similarity, 71%/79%) encoded in the SGR810-815 BGC from Streptomyces griseus and the PKS/NRPS enzyme (identity/similarity, 62%/73%) encoded within the HSAF cluster from Lysobacter enzymogenes C3 (Table 3). Accordingly, SkoA is proposed to be responsible for the assembly of the polyene tetramate precursor in the biosynthetic pathway to compounds 1 and 2.  The center of the gene cluster is the hybrid PKS/NRPS encoding gene skoA; SkoA comprises one PKS module and one NRPS module, making up a total of nine domains (KS-AT-DH-KR-ACP-C-A-PCP-TE). As reported, this type of hybrid PKS/NRPS is con-served among PTM producing bacterial species [32] (Figure 5). The single-module PKS of the hybrid PKS/NRPS enzyme is iteratively used to produce two separate polyketide chains. These chains are, respectively, linked with the αand δamino groups of an L-ornithine that is tethered in the NRPS portion to generate a common polyene tetramate intermediate [25,26]. SkoA shows high similarity to the previously characterized PKS/NRPS SGR814 (identity/similarity, 71%/79%) encoded in the SGR810-815 BGC from Streptomyces griseus and the PKS/NRPS enzyme (identity/similarity, 62%/73%) encoded within the HSAF cluster from Lysobacter enzymogenes C3 (Table 3). Accordingly, SkoA is proposed to be responsible for the assembly of the polyene tetramate precursor in the biosynthetic pathway to compounds 1 and 2.
Downstream of the skoA are two oxidoreductases encoding genes (skoB1 and skoB2). SkoB1 and SkoB2 strongly correlate to SGR813 (similarity/identity, 82%/89%) and SGR812 (similarity/identity, 84%/90%) enzymes encoded by the SGR810-815 gene cluster, respectively. It has been demonstrated that formation of the diverse carbocyclic ring in PTMs, after formation of the tetramate polyene precursor, is catalyzed by a set of oxidoreductases. Generally, the presence of a single FAD-dependent oxidoreductase (such as IkaB in the ikarugamycin biosynthetic pathway) is suggested to catalyze the formation of the 5/6 ring system in PTMs [31]. In contrast, the presence of two or three oxidoreductases (such as SGR812 and SGR813 in the alteramide A biosynthetic pathway, OX1-OX3 [30] in HSAF biosynthesis, as well as PtmB1 and PtmB2 [8] in pactamide A assembly) is thought necessary for the formation of 5/5 ring systems [25] (Figure 5). SGR812 or PtmB2 or OX3 was demonstrated to catalyze the formation of the first five-membered ring [8,30], whereas the formation of the second five-membered ring appeared to be catalyzed by SGR813 or PtmB1 or OX1/OX2 [8,30]. Notably, it has been confirmed that formation of the second ring is coupled to C14 hydroxylation during HSAF biosynthesis [33]. The presence of two oxidoreductases in the sko gene cluster is consistent with the 5/5 ring system seen in 1 and 2. Thus, SkoB2 is predicted to catalyze the formation of the first five-membered ring, whereas SkoB1 is proposed to install the second five-membered ring coupled with C14 hydroxylation en route to 1 and 2.
Flanking the left of the PKS/NRPS encoding skoA is a sterol desaturase (SD) gene, skoD. SkoD corresponds well to the SD found encoded in the HSAF gene cluster [35], SGR815 encoded in SGR810-815 gene cluster ( Figure 5, Table 3) and FtdA encoded in the frontalamide gene cluster [32]. The SD and FtdA have been demonstrated to be responsible for the C25-hydroxylation in the corresponding 5/5/6 ring system [32,35]; SGR815 has been confirmed to hydroxylate C25 of the 5/5 ring system. Importantly, such hydroxylases have been shown to exhibit a reasonable amount of substrate promiscuity [32,36]. Thus, SkoD is proposed to catalyze the C25-hydroxylation of two different substrates in the biosynthetic pathway to compounds 1 and 2.
To the right of the sko gene cluster is encoded a cytochrome P450 enzyme (SkoE); this enzyme shows high similarity to PtmE'(similarity/identity, 71%/77%) encoded in the ptm' gene cluster and FtdF (similarity/identity, 60%/72%) encoded in the frontalamide cluster. PtmE' was predicted to catalyze hydroxylation at C12 whereas FtdF was predicted to catalyze 14-dehydrogenation [26,32]; however, neither of these proposed functions have been experimentally verified thus far.
Based on the above bioinformatics analysis of gene functions and organization of the sko gene cluster, we propose putative biosynthetic pathways leading to 10-epi-HSAF (1) and compound 2 ( Figure 4A). We envision the following route to assembly for both 1 and 2. First, iterative catalysis of the single-module PKS of SkoA produces two separate polyketide chains. The NRPS portion of SkoA then catalyzes the formation of two amide bonds employing the α and δ-amino groups of the same amino acid (ornithine) via two Claisen condensations and the subsequent formation of the tetramate moiety via a Dieckmann-type reaction to generate a polyene tetramate precursor I ( Figure 4A). The FAD-dependent oxidoreductase SkoB2 catalyzes formation of the cyclopentane ring in II, and II is then acted upon by SkoB1 to install the second five-membered ring of the fused 5/5 system while also installing the C14 OH moiety to afford intermediate III. Subsequently, the alcohol dehydrogenase SkoC catalyzes formation of the six-membered necessary to yield 10-epi-deOH-HSAF. Finally, the sterol desaturase SkoD catalyzes hydroxylation at C25 of 10-epi-deOH-HSAF to form 10-epi-HSAF (1). Meanwhile, due to the functional promiscuity of SkoD as repoted for other characterized sterol desaturases (e.g., SGR815 [28]) ( Figure 4B), SkoD may also hydroxylate C25 of intermediate III to generate intermediate IV. Intermediate IV might be further transformed into the hexacyclic 2 spontaneously as reported for altermide A [29] ( Figure 4C) and combamide C [37]. It has been reported that alteramide A is susceptible to a unique intramolecular photochemical [4 + 4] cycloaddition to generate hexacyclic products [29]. Similar spontaneous reactions have also been proposed in the biosynthesis of combamide C [37]. Notably, cytochrome P450 SkoE showed high similarity with FtdF; however, the function of FtdF has not been experimentally verified thus far. Based on the inspiration of P450 BvnD [38], which was reported to set the stage for the following spontaneous [4 + 2] Diels-Alder reaction, we speculated that SkoE might be involved in accelerating the spontaneous [4 + 4] cycloaddition; however, this hypothesis requires further study.
To demonstrate the validity of the putative sko gene cluster in the biosynthesis of 10-epi-HSAF (1) and koyanamide A (2), the 1.6 kb KS domain coding region in the PKS/NRPS gene skoA was deleted using PCR-targeting methods; the resulting mutant strain, S. koyangensis SCSIO 5802ACM, was verified by both phenotypic and genotypic means ( Figure S21). Fermentation and extraction of the SCSIO 5802ACM strain was carried out enabling follow up metabolomics. Not surprisingly, the HPLC-based metabolite analyses of the SCSIO 5802ACM mutant fermentation revealed the complete absence of 1 and 2, thereby confirming the indispensability of the sko BGC for the biosynthesis of compounds 1 and 2 ( Figure 2, traces vii-viii).

Genome Sequencing, Annotation, and Analysis
Sequencing of the complete genome was accomplished in the prior study [18]. The complete chromosome and plasmid sequences for SCSIO 5802 have been deposited in the GenBank database with accession numbers CP049945 and CP049946, respectively. Genes involved in secondary metabolic pathways were predicted using online antiSMASH software (http://antismash.secondarymetabolites.org/). The deduced ORFs were analyzed using online FramePlot 4.0beta software (http://nocardia.nih.go.jp/fp4/, accessed on 5 July 2021) and their functional predictions were accomplished with an online BLAST program (http://blast.ncbi.nlm.nih.gov/, accessed on 5 July 2021). The PKS architectures were analyzed using an NRPS-PKS online website (http://nrps.igs.umaryland.edu/nrps/, accessed on 5 July 2021). The identity and similarity values between proteins encoded in sko gene cluster and their homolog proteins were analyzed using an online Protein BLAST program (https://blast.ncbi.nlm.nih.gov/Blast.cgi?PROGRAM=blastp, accessed on 5 July 2021).

Construction of an In-Frame "Double-Deletion" Mutant SCSIO 5802AC and Inactivation of skoA in SCSIO 5802AC
In-frame gene deletions were achieved by following the REDIRECT protocol [40]. The S. koyangensis SCSIO 5802 genomic cosmid library was constructed as previously [18]. The apramycin resistance gene oriT-aac(3)IV fragment was obtained by using specific primers (Table S1) that contain additional SpeI restriction sites, and used to replace the target gene can in the cosmid 7-8F. Restriction digest of the mutant cosmid with SpeI and subsequent relegation predictably abolished the apramycin resistance gene oriT-aac(3)IV fragment. The second round of PCR-targeting was performed to replace the kanamycin resistance gene on SuperCos I with another apramycin resistance gene oriT-aac(3)IV fragment obtained by a primer pair ARK [41]. The constructed mutant cosmid was introduced into nonmethylating Escherichia coli ET12567/pUZ8002 and then transferred into S. koyangensis SCSIO 5802A, an in-frame deletion mutant, which has been constructed previously [22]. The single-crossover mutant strain was selected from antibiotic selection (apramycin R ). To obtain the double-crossover mutant strain, the single-crossover mutant strain was first cultured on antibiotic-free ISP-4 medium. The double-crossover mutant was then selected by showing the apramycin S phenotypes. The double-crossover mutant was finally confirmed by PCR verification ( Figure S3). The in-frame "double-deletion" mutant was designated SCSIO 5802AC.

Secondary Metabolites Analyses of S. koyangensis SCSIO 5802 and Related Derivative Strains
The wild-type S. koyangensis SCSIO 5802 and relevant gene-inactivated mutants were first grown on A1 medium agar [20] at 28 • C for 4-5 days to achieve sporulation. A portion of mycelium and spores (1 cm 2 ) for each strain was added to 250 mL flasks containing 50 mL of RA medium. Fermentations were then carried out at 28 • C on rotary shakers (200 rpm) for 8 days. After fermentation, each fermentation culture was washed with 100 mL butanone, and the butanone solvent was removed under reduced pressure to afford an oily residue. Residues were each dissolved into 1 mL MeOH and centrifuged at 13,000× g for 10 min; supernatants were subjected to HPLC-UV analyses, each of which was performed using an Agilent Technologies 1260 Infinity system using a Phenomenex ODS column (150 × 4.6 mm, 5 µm), eluting with a linear gradient of 5 to 65% solvent B (solvent B: CH 3 CN + 0.1% trifluoroacetic acid (TFA); solvent A: H 2 O + 0.1% TFA) over 20min, followed by 65% to 100% solvent B in 2 min, and then 100% solvent B for 5 min, at a flow rate of 1 mL/min. Since the UV absorption wavelengths of compounds 1-6 and compounds 7-9 are different, chromatograms employed UV detection at both 254 nm and 400 nm to ensure comprehensive detection of new compounds.

Conclusions
In this study, we analyzed the complete genome of S. koyangensis SCSIO 5802 and employed antiSMASH to unveil 21 putative BGCs including one putative PTM-encoding gene cluster. Ablation of the abyssomicin and candicidin biosynthetic pathways [22], major products of SCSIO 5802, enabled us to see two previously absent peaks in HPLC analyses of fermentation extracts from the S. koyangensis SCSIO 5802AC mutant strain. The two new HPLC signals were isolated and identified as PTM-type compounds, 10-epi-HSAF (1) and koyanamide A (2). Within the genome of SCSIO 5802 we also were able, using bioinformatics, to identify the sko BGC responsible for the production of 1 and 2; gene inactivation experiments verified 1 and 2 as direct products of the sko biosynthetic machinery. Finally, enabled by our knowledge of the full sko cluster and structural elucidations of 1 and 2, we have formulated a cogent biosynthetic pathway leading to these otherwise unobserved NPs. This work highlights metabolic engineering and genome mining as effective tools by which to turn on putatively silent BGCs; the result is expanded NP chemical diversity for drug discovery initiatives.

Data Availability Statement:
The authors declare that all relevant data supporting the findings of this study are available within the article and its Supplementary Materials file, or from the corresponding authors upon request.