Novel Fredericamycin Variant Overproduced by a Streptomycin-Resistant Streptomyces albus subsp. chlorinus Strain

Streptomycetes are an important source of natural products potentially applicable in the pharmaceutical industry. Many of these drugs are secondary metabolites whose biosynthetic genes are very often poorly expressed under laboratory cultivation conditions. In many cases, antibiotic-resistant mutants exhibit increased production of natural drugs, which facilitates the identification and isolation of new substances. In this study, we report the induction of a type II polyketide synthase gene cluster in the marine strain Streptomyces albus subsp. chlorinus through the selection of streptomycin-resistant mutants, resulting in overproduction of the novel compound fredericamycin C2 (1). Fredericamycin C2 (1) is structurally related to the potent antitumor drug lead fredericamycin A.


Introduction
The bacterial genus Streptomyces is well-known for producing a huge variety of bioactive secondary metabolites with potential pharmaceutical applications [1,2]. The genes responsible for their biosynthesis are generally clustered together. However, many of these biosynthetic genes are poorly or not expressed (silent) under laboratory cultivation conditions. Thus, the activation of silent secondary metabolite gene clusters is an essential step for the discovery of new natural drugs. A simple strategy to activate or upregulate the expression of these genes consists in screening for antibiotic-resistant mutants [3,4]. This approach is based on the work of Ochi and his team. They discovered that certain mutations in the rpsL and rpoB genes, which code for the ribosomal protein S12 and the β-subunit of RNA polymerase, respectively, lead to an altered gene product that confers resistance to streptomycin (str mutants) or to rifampicin (rif mutants), respectively. The mutations in these genes also increase the production of secondary metabolites in several Streptomyces strains [3][4][5][6][7]. Presumably, some of the str and rpoB mutations give rise to diverse metabolic changes which typically occur during the stringent response. The stringent response is triggered in E. coli [8] and other prokaryotic microorganisms [9,10] under amino acid starvation conditions by the signaling molecule, guanosine tetraphosphate (ppGpp). This molecule generates a series of physiological changes, including a significant decrease of protein synthesis [11], downregulation of rRNA transcription [12], and activation of stationary-phase metabolic processes, such as the expression of secondary metabolite biosynthetic genes [9]. Thus, it is suggested that the mutant gene products of rpsL and rpoB may be responsible for antibiotic overproduction, mimicking the role of ppGpp in the stringent response [3][4][5][6].
In previous studies, the marine strain Streptomyces albus subsp. chlorinus NRRL B-24108 has been shown to harbor genes for the production of several bioactive secondary metabolites, such as the herbicide albucidin or the antibiotic nybomycin [13,14]. Here, we report the induction of a type II polyketide synthase (PKS) gene cluster in S. albus subsp. chlorinus, leading to overproduction of the novel compound fredericamycin C 2 (1). This substance is structurally related to fredericamycin C (2), a secondary metabolite first isolated in 1981, together with the compounds fredericamycin A and fredericamycin B, from the culture broth of Streptomyces griseus ATCC 49344 [15]. All fredericamycin variants known to date (A, B, C, C 1 , and E) share a similar structure that involves two aromatic moieties, which in the case of fredericamycins A and E are linked by a rare stereogenic spiro carbon center [16][17][18]. Fredericamycin A displays strong in vivo anticancer activity against several mammal tumor cell lines [19], and it has been patented as an effective treatment for certain types of cancer in humans [20]. Additionally, fredericamycins A, B, C, and C 1 exhibit moderate antibacterial and antifungal activities [19]. In this study, we present the generation of the high-level streptomycin-resistant strain, Streptomyces albus subsp. chlorinus JR1, which overproduces the novel compound fredericamycin C 2 (1). We also describe the mutation likely causing this phenotype and propose the initial biosynthetic steps of fredericamycin C 2 (1), based on the gene cluster homology with that of fredericamycin A.

High-Level Streptomycin-Resistant Mutant S. albus subsp. chlorinus Overproduces the Novel Compound Fredericamycin C 2
The strain S. albus subsp. chlorinus NRRL B-24108 was cultured in MS agar medium containing increasing concentrations of streptomycin. After several rounds of selection, we isolated a streptomycin-resistant colony that displayed a distinctive violet color when cultured on a solid medium, in contrast to the characteristic white color of the parental strain ( Figure 1). We named the mutant strain S. albus subsp. chlorinus JR1. While the growth of the wild type of S. albus subsp. chlorinus was inhibited by a streptomycin concentration of 50 µg/mL, the mutant strain was able to grow in the presence of up to 200 µg/mL of antibiotic (Table S1). The metabolic profiles of both strains were analyzed using liquid chromatography and mass spectrometry, revealing the presence of a peak at t R = 12.7 min that corresponds to an [M + H] + ion of m/z 521.107 and displays UV absorption signals at λ max 195, 248, 274, 345, and 490 nm ( Figure 2). The peak area revealed an over 50-fold greater production of the corresponding compound by S. albus subsp. chlorinus JR1 compared to the parental strain's yield (Table S2, Figure S1). The search of this mass in a natural product database yielded no coincidences, suggesting a potentially undescribed metabolite overproduced by S. albus subsp. chlorinus JR1. The compound was extracted from a 2 l solid culture of S. albus subsp. chlorinus JR1 and purified by normal phase chromatography through a silica column followed by reverse phase chromatography.    O 11 ). In the proton NMR spectrum, two broad multiplets are visible at δ H 2.63 (t, 6'-H 2 ) and 2.79 (t, 7'-H 2 ) ppm, which are two neighbored methylene groups. All other signals are singlets: One methoxy group at δ H 3.94 (6-OCH 3 ), one additional methylene group at δ H 4.06 (4'-H 2 ), one methyl group at δ H 2.17 (1"-H 3 ), and two methines at δ H 6.52 (5'-H) and 6.88 (7-H). Moreover, broad singlets at δ H 5.5, 12.8, and 13.2 ppm indicate hydroxyl groups to be present in the molecule (Table 1). Typical quinone carbonyl signals at δ C 186.9 (C-4) and 187.6 (C-9) are visible in the 13 C-NMR spectrum. A signal at δ C 169.3 ppm (C-1') shows an acid functionality which is attached at C-8' proven from an HMBC (Heteronuclear Multiple Bond Coherence) correlation C-1'/ 5'-H. A very weak signal at δ H 204.1 ppm (C-3') shows a ketone supported by an HMBC cross-peak C-3'/1"-H 3 ( Table 1). This ketone is part of a propan-2-on sidechain which is attached at C-6' based on HMBC signals C-6'/4'-H 2 , C-6'/1"-H 3 , C-5'/4'-H 2 , and C-7'/4'-H 2 . The HMBC signal C-6/6-OCH 3 indicates the methoxy group to be attached at C-6 next to the aromatic proton 7-H. One of the rings contains a CH 2 -CH 2 moiety (C-6'-C7') which is in ring D proven from HMBC signals C-3/7'-H 2 , C-7a'/7'-H 2 , C-8'/7'-H 2 , C-7a'/6'-H 2 , and C-5a'/6'-H 2 ( Figure S2). Comparison to literature shows high similarity of the new compound named fredericamycin C 2 (1) to fredericamycin C (2) [17], which differs only in the sidechain attached at C-6' (fredericamycin C) and to KS-619-1 (3) [21] with a different substitution pattern of the pentacyclic ring system (KS-619-1) ( Figure 3).

Fredericamycin C 2 Is Biosynthesized by a Type II PKS Gene Cluster
Fredericamycin C 2 (1) is structurally related to other fredericamycin variants, such as C and A, which are biosynthesized by a type II PKS gene cluster [22]. This suggests the involvement of a type II PKS system in the production of fredericamycin C 2 (1). Although a subclass of type I PKSs (iterative type I PKSs) have also been reported to synthesize aromatic compounds [23][24][25][26], their products are structurally smaller and simpler than the complex multicyclic metabolites produced by type II PKSs [27]. The pentacyclic aromatic polyketide structure of fredericamycin C 2 (1) (Figure 3) further supports the assumption of type II PKS genes involved in its biosynthesis. The genome of S. albus subsp. chlorinus was screened for secondary metabolite genes, revealing the presence of one type II PKS gene cluster. Based on protein BLAST analysis, we assigned the genes putative functions, which are summarized in Table 2. To test whether the expression of these genes leads to fredericamycin C 2 (1) production, BAC 2P5 containing the identified type II PKS cluster was isolated from a genomic library of S. albus subsp. chlorinus NRRL B-24108 and transferred via intergeneric conjugation into the heterologous host Streptomyces albus Del14. HPLC-MS analysis of the extract from the resulting ex-conjugant S. albus 2P5 revealed the presence of a peak with identical retention time and m/z to those of fredericamycin C 2 (1), demonstrating that the type II PKS cluster from S. albus subsp chlorinus is responsible for fredericamycin C 2 (1) biosynthesis ( Figure S3, Figure 2). BAC 2P5 comprises a 35 kb genomic region containing a total of 37 open reading frames (ORFs), 10 of which share homology at protein level with the fredericamycin A gene cluster from S. griseus (Table 2, Figure 4), which is also responsible for fredericamycin C (2) biosynthesis [22]. The homologue genes include those coding for the minimal PKS ketosynthase subunits (KS α and KS β ), two polyketide cyclases, four tailoring enzymes, a transcriptional regulator, and a protein of unknown function. The gene similarity with fredericamycin A cluster, together with the heterologous expression results ( Figure S3), indicates the relevance of these genes in fredericamycin C 2 (1) biosynthesis. Frequently, streptomycin resistance results from a point mutation in the rpsL gene, which codes for the ribosomal protein S12 [3,4,[28][29][30]. Following genome sequencing of S. albus subsp. chlorinus JR1, the resulting reads were mapped to the genome of the wild type strain and single nucleotide mutations were searched in the sequence corresponding to the rpsL gene. No point mutations were found in this sequence, indicating that the mutation responsible for streptomycin resistance in S. albus subsp. chlorinus JR1 is located elsewhere in the genome. A total of fifteen point mutations were detected in the genome of S. albus subsp. chlorinus JR1 (Table S3). A nucleotide insertion was detected within the coding sequence of the jag gene (SACHL2_00217; position 3714884) that codes for a single-stranded DNA binding protein. This gene partially overlaps with the adjacent downstream gene rsmG (SACHL2_00216), indicating their co-transcription in a bicistronic operon. The insertion of a cytosine nucleotide in the sequence of jag creates a premature stop codon that truncates the protein translation, also affecting the expression of the co-translating rsmG gene. rsmG codes for a 16S rRNA methyltransferase, and it has been reported that point-nonsense mutations in its sequence, as well as the deletion of this gene, lead to increased resistance to streptomycin and enhanced production of secondary metabolites in different bacterial strains [31,32]. Therefore, we believe that a point mutation in the jag gene is responsible for high-level streptomycin resistance and upregulation of fredericamycin C 2 (1) biosynthetic gene expression in S. albus subsp. chlorinus JR1 by preventing the translation of the rsmG gene.

Biological Activity of Fredericamycin C 2
Fredericamycin C 2 (1) was tested for antibacterial activity against a Gram-positive (Bacillus subtilis) and two Gram-negative strains (Escherichia coli and Pseudomonas putida) through disk diffusion test. The new fredericamycin variant displays a growth inhibition zone against P. putida at a minimal concentration of 2.5 mg/mL ( Figure S4). Fredericamycin C 2 (1) shows no inhibitory activity against the growth of B. subtilis and E. coli.

Discussion
Fredericamycin variants constitute a family of aromatic polyketides with significant toxicity against tumor cells as well as moderate antibiotic and antifungal activity. Here, we present the novel variant fredericamycin C 2 (1), which is overproduced by the strain S. albus subsp. chlorinus JR1, a spontaneous streptomycin-resistant mutant derived from S. albus subsp. chlorinus NRRL B-24108. We suggest that a mutation in the jag gene affecting the translation of the adjacent gene rsmG is responsible for the phenotype of the mutant strain. The frame shift originated by the point insertion putatively generates a truncated non-functional Jag protein. In previous studies, deletion of a jag homologue in Streptococcus pneumoniae led to retarded growth and smaller cell size compared to the wild type strain, indicating that Jag is likely involved in cell division [33]. However, no association of jag deletion with increased antibiotic resistance or induction of secondary metabolite production has been reported before. The point mutation in the jag gene has a polar effect on the overlapping gene rsmG, preventing its transcription. The enhanced production of fredericamycin C 2 (1), as well as the increased streptomycin resistance observed in S. albus subsp. chlorinus JR1, is most likely derived from the lack of RsmG function. This is consistent with previous studies where rsmG deletion mutants showed higher resistance to streptomycin and improved yields of secondary metabolites at a late-growth phase [34,35]. rsmG encodes a methyltransferase that catalyzes the methylation of 16S rRNA at the residue G527 (E. coli numbering). This residue, together with C526 and the S12 protein, interacts with the antibiotic streptomycin [36]. These interactions tend to stabilize the tRNA-mRNA tandem, which affects the proof-reading process and results in misreading of the genetic code [37]. The absence of RsmG would generate 16S rRNA molecules non-methylated at residue G527, causing a weaker binding to streptomycin and making the strain resistant to the antibiotic. Although the mechanism by which the lack of 16S rRNA methyltransferase may induce the expression of fredericamycin C 2 (1) biosynthetic genes in S. albus subsp. chlorinus JR1 remains unknown, we hypothesize that the mutant experiences an increased protein synthesis rate at a stationary phase, as it has been previously observed [34,35]. The increased protein synthetic activity leads to expression of both pleiotropic and pathway-specific regulatory proteins, which eventually enhance the transcription of poorly expressed secondary metabolite gene clusters. Several attempts to complement the jag and rsmG gene functions in S. albus subsp. chlorinus JR1 resulted in no recombinant colonies, suggesting the genetic intractability of the strain (data not shown).
The structure of fredericamycin C 2 (1) presented in this paper only differs from that of fredericamycin C (2) in the polyketide chain length. While fredericamycin C (2) backbone consists of 30 carbon atoms, fredericamycin C 2 (1) contains a C 26 polyketide chain (Figure 3). Fredericamycin C (2) biosynthesis begins with the generation of a C 6 primer unit (hexadienyl-ACP) by the PKS initiation module. This starter unit is then transferred to the elongation module, which presumably catalyzes the sequential decarboxylative condensation of 12 malonyl-CoA molecules, delivering a C 30 polyketide chain [22,38]. In the case of fredericamycin C 2 (1), we propose that acetyl-CoA functions as the starter unit, which is elongated by the polyketide synthase through successive incorporation of malonyl-CoA extender units ( Figure 5). This process is most likely catalyzed by the minimal PKS enzymes encoded by the genes c2fdmB1, c2fdmA1, and c2fdmZ, and the ACP S-malonyltransferase encoded by c2fdmK1 ( Table 2). The resulting C 26 polyketide chain is subsequently modified by tailoring enzymes to eventually yield the product fredericamycin C 2 (1) ( Figure 5).

General Experimental Procedures
All strains and BACs (bacterial artificial chromosomes) used in this work are listed in Table S4. Escherichia coli strains were cultured in LB medium [39]. Streptomyces strains were grown on soy flour mannitol agar (MS agar) [40] for sporulation and conjugation and in liquid tryptic soy broth (TSB; Sigma-Aldrich, St. Louis, MO, USA). For metabolite expression, liquid DNPM medium (40 g/L dextrin, 7.5 g/L soytone, 5 g/L baking yeast, and 21 g/L MOPS, pH 6.8) or MS agar were used. The antibiotics kanamycin, apramycin, and nalidixic acid were supplemented when required.

Strain Generation, Metabolite Extraction, and Analysis
The spontaneous streptomycin-resistant mutant Streptomyces albus subsp. chlorinus JR1 was obtained after several rounds of selection of streptomycin-resistant colonies growing on MS agar medium containing increasing concentrations of the antibiotic. MICs of streptomycin were determined by spreading spores on MS agar plates containing 50, 100, and 200 µg/mL of streptomycin. Metabolites were extracted from the agar with ethyl acetate acidified with 100% acetic acid up to pH = 2.0, evaporated and dissolved in methanol. One µL of extract was separated using a Dionex Ultimate 3000 UPLC (Thermo Fisher Scientific, Waltham, MA, USA), a 10-cm ACQUITY UPLC®BEH C18 column, 1.7 µm (Waters, Milford, MA, USA), and a linear gradient of 0.1% formic acid solution in acetonitrile against 0.1% formic acid solution in water from 5% to 95% in 18 min at a flow rate of 0.6 mL/min. Samples were analyzed using an Orbitrap speed mass spectrometer (Thermo Scientific, Waltham, MA, USA). Data were collected and analyzed with the Thermo Xcalibur software, version 3.0 (Thermo Scientific, Waltham, MA, USA). The monoisotopic mass was searched in a natural product database.

Fredericamycin C 2 Purification and Quantification
S. albus subsp. chlorinus JR1 was grown for 8 days at 28 • C on 50 Petri dishes, each containing 40 mL of DNPM agar. Fredericamycin C 2 (1) was extracted from the solid agar using ethyl acetate acidified with 100% acetic acid up to pH = 2.0. The crude extract was first separated by normal phase chromatography on a prepacked silica cartridge (Biotage, Uppsala, Sweden) using hexane (solvent A), chloroform (solvent B), ethyl acetate (solvent C), and methanol (solvent D) (1:1:1:1) as the mobile phase, in a linear gradient from 0% to 100% of each pair of solvents (A-B, B-C, and C-D). Fractions containing fredericamycin C 2 (1) were detected by LC-MS analysis, pooled together, and further fractionated by semi-preparative HPLC (Dionex UltiMate 3000, Thermo Fisher Scientific, Waltham, MA, USA) using a C18 column (Synergi 10 µm, 250 × 10 mm; Phenomenex, Aschaffenburg, Germany) and a 0.1% formic acid solution in acetonitrile as the mobile phase in a linear gradient. UV spectra were recorded with a DAD detector at 274 nm and 525 nm. Finally, 0.8 g of fredericamycin C 2 (1) was collected in a single fraction. For quantification, a calibration curve with different fredericamycin C 2 (1) concentrations was constructed ( Figure S1).

H-NMR Spectroscopy
NMR data were measured using a Varian VNMR-S600 spectrometer equipped with 3 mm triple resonance inverse and 3 mm dual broadband probes. Fredericamycin C 2 (1) samples were dissolved in 150 µL DMSO-d6/Pyridine-d5 95:5 and measured at 35 • C. The residual solvent signal of DMSO was used as an internal reference.

Antimicrobial Susceptibility Test
Disk diffusion tests were performed according to [42]. Ten mL of LB soft agar (10 g/L tryptone, 10 g/L NaCl, 5 g/L yeast extract, 7 g/L agar) was inoculated with the strains Escherichia coli GB2005, Bacillus subtilis ATCC 6633, or Pseudomonas putida KT2440, and poured on LB agar plates. Five paper disks (Macherey and Nagel, Düren, Germany) were coated with 10, 5, 2.5, 0.5, or 0.25 mg/mL of fredericamycin C 2 (1) solved in methanol and placed onto the solidified soft agar. An additional disk loaded with methanol was used as a negative control and the antibiotics nalidixic acid, ampicillin, and chloramphenicol (50 µg/mL, respectively) were used as positive controls. The plates were incubated at 28 • C overnight.

Conclusions
Here, we demonstrate the significance of inducing poorly expressed secondary metabolite gene clusters for the identification of new microbial natural products. A mutation in the genome of the strain S. albus subsp. chlorinus JR1 obtained by selection of streptomycin-resistant colonies has led to overproduction and the discovery of the so far undescribed compound fredericamycin C 2 (1). This secondary metabolite expands the structural variability of the fredericamycin family, whose most prominent member, fredericamycin A, is a potent antitumor drug. Screening for antibiotic-resistant strains represents a simple and inexpensive approach that has enabled the improvement of secondary metabolite production in the genetically intractable bacterial strain S. albus subsp. chlorinus.

Conflicts of Interest:
The authors declare no conflict of interest.