Characterization of Biosynthetic Genes of Ascamycin/Dealanylascamycin Featuring a 5′-O-Sulfonamide Moiety in Streptomyces sp. JCM9888

Ascamycin (ACM) and dealanylascamycin (DACM) are nucleoside antibiotics elaborated by Streptomyces sp. JCM9888. The later shows broad spectrum inhibition activity to various gram-positive and gram-negative bacteria, eukaryotic Trypanosoma and is also toxic to mice, while ascamycin is active against very limited microorganisms, such as Xanthomonas. Both compounds share an unusual 5′-O-sulfonamide moiety which is attached to an adenosine nucleoside. In this paper, we first report on the 30 kb gene cluster (23 genes, acmA to acmW) involved in the biosynthesis of these two antibiotics and a biosynthetic assembly line was proposed. Of them, six genes (AcmABGKIW) are hypothetical genes involved in 5′-O-sulfonamide formation. Two flavin adenine dinucleotide (FAD)-dependent chlorinase genes acmX and acmY were characterized which are significantly remote from acmA-W and postulated to be required for adenine C2-halogenation. Notably gene disruption of acmE resulted in a mutant which could only produce dealanylascamycin but was blocked in its ability to biosynthesize ascamycin, revealing its key role of conversion of dealanylascamycin to ascamycin.

Dealanylascamycin (DACM) (1) and ascamycin (ACM) (2) [20] (also identified as AT-265 [21]) ( Fig. 1) are two adenosine antibiotics produced by Streptomyces sp. JCM9888 whose chemical scaffold resemble that of the fluorine-containing antibiotic nucleocidin (3) [3] (Fig. 1). Ascamycin, dealanylascamycin and nucleocidin are closely related 59-O-sulfonamide ribonucleosides. The ascamycins have C2-chloroadenine as the base on C-19, whereas nucleocidin uses adenine, which lacks the chlorine. Nucleocidin also has a unique fluorine atom on its C-49 position. Ascamycin and dealanylascamycin differ by N-alanylation of the 59-Osulfonamide moiety in the former but not the latter ( Fig. 1) [3]. Although ascamycin and dealanylascamycin have similar chemical structures, they possess very different bactericidal activities. Ascamycin exhibited toxicity only to a few bacterium genus, such as Xanthomonas [22], while dealanylascamycin is broadspectrum and active against both gram-positive, gram-negative bacteria and even some eukaryotic cells [22]. The later also showed unusual trypanocidal (such as Trypanosoma equiperdum) activity and is an anti-amoebae (such as Endamoeba histolytica) agent and thus it has been found valuable in treating animal diseases caused by these microorganisms [23]. Their differences in biological activity has been attributed to an aminopeptidase that cleaves the alanyl group from ascamycin. Since ascamycin is only active against a few microorganisms (Xanthomonas) in which this aminopeptidase was found, it was proposed that dealanylascamycin is the active form of this antibiotic, and most microorganisms (such as E. coli etc.) lack this aminopeptidase and are therefore not susceptible to ascamycin [22,24]. Ascamycin and dealanylascamycin both possess a chlorine atom on the C2 position of adenine. It is often the case that incorporation of halogen into natural products plays an important role in increasing their biological scope and activity [25]. Some naturally occurring halogenated secondary metabolites, such as vancomycin, chlorotetracycline are of clinical importance. The mechanisms of biohalogenation, particularly for chlorination and bromination were widely discussed [25] [26] [27]. Enzymes involved in halogenation of aromatic groups have been extensively described, which require a two-component flavin-dependent halogenase/reductase as follows: A flavin reductase uses NADH (Nicotinamide adenine dinucleotide) to reduce FAD (flavin adenine dinucleotide) and produces FADH 2 (reduced flavin adenine dinucleotide) and a halogenase utilize the FADH 2 formed and O 2 as cofactors to perform electrophilic halogenation via an intermediate halonium ion equivalent [25]. The mechanism of dealanylascamycin inhibition has been studied in the analogous antibiotic nucleocidin, which demonstrated that it inhibits amino acid incorporation into human liver cell protein in vivo and in vitro [28]. Due to their special biological activities, organic chemists have been attracted to synthesize ascamycin, dealanylascamycin, nucleocidin and related nucleoside derivatives [29] [30] [31].
In this paper we first report the entire gene cluster (acmA-acmW) responsible for the biosynthesis of two nucleoside antibiotics ascamycin/dealanylascamycin and a biosynthetic machinery was purposed from in silico analysis. Intriguingly, six genes (AcmABGKIW) are postulated to be involved in 59-O-sulfonamide formation. Two FADH 2 -dependent halogenases (acmX and acmY) which are involved in the pathway were found not clustered with acmA-acmW and we postulate that they are required for chlorination on the C2-position of the adenine ring. Notably, inactivation of an esterase acmE resulted in a mutant which only produce dealanylascamycin but blocked in its ability to biosynthesize ascamycin which suggest its role of converting dealanylascamycin to ascamycin. These results provide biosynthetic profiles for the 59-O-sulfonamide containing antibiotic ACM/DACM and lay a solid foundation for target improvement of their production via synthetic biology strategy..

Results
Identification and characterization of the ascamycin/dealanyl-ascamycin biosynthetic genes Streptomyces sp. JCM9888 is unusual in its ability to biosynthesize two nucleoside antibiotics: dealanylascamycin (1) and ascamycin (2) [20] (Fig. 1). As they are both featured with a 59-O-sulfonamide moiety, we suspected sulfate metabolite related genes are required for the antibiotic production. Partial genome sequencing of Streptomyces genome revealed a 30,488 bp contiguous DNA sequence with an overall GC content of 66.7% (GenBank accession number KJ817374). Bioinformatics analysis of the sequence revealed 23 ORFs (acmA-W) and the deduced functions of individual ORFs are annotated and summarized in Table 1. Of them, 6 genes are proposed to be related to 59-O-sulfonamide biosynthesis (acmA, acmB, acmG, acmI, acmK, acmW) ( Fig. 2A). Notably, no halogenases were obvious around the gene cluster while BLASTp searches highlighted that two hypothetical flavin-dependent chlorinases (acmX and acmY) (GenBank accession number KJ817375) lying adjacent to each other on the chromosome in a distance of approx. 1 million basepair from acmA-W (Fig. 2B). AcmX and AcmY were shown to share high homology to all known FADdependent chlorinases, such as ChlB4 (accession number AAZ77674) from Streptomyces antibiotics which participate in chlorothricin biosynthesis (65% and 57% protein sequence identity respectively) [32]. Their analogy to proteins required for aromatic moiety chlorination suggested that they are potential candidates of the ascamycin/dealanylascamycin chlorinases as both two antibiotics possess a chlorine atom at C2-position of adenine.

59-O-sulfonamide and 59-(N-alanyl-O-sulfonamide) groups formation
Ascamycin and dealanylascamycin are decorated with unusual sulfonamide groups at the 59-hydroxyl of the ribose ring which are rare paradigms in secondary metabolites [33]. Six genes are postulated to be involved in 59-O-sulfonamide moiety activation, loading and transfer in the ascamycin/dealanylascamycin gene cluster (acmA, B, G, I, K, W). Sulfate ions(SO 4 22 ) are not utilised directly but must first be activated by the formation of adenosyl phosphosulfate (APS) or phosphoadenosine phosphosulfate (PAPS). These are activated forms of sulfate generated by coupling inorganic sulfate onto ATP (adenosine triphosphate) (  Fig. 3B). Sulfurylation of ATP is probably catalysed by adenylsulfate kinase (AcmB): SO 4 22 + ATP R PAPS + PP i . Next, AcmK (sulfotransferase) [34] or AcmA and AcmW (sulfate adenyltransferase) [35] act upon the 59 position of adenosine, utilizing PAPS as an activated sulfate donor to generate adenosine 59sulfonate. The manner in which AcmG (sulfatase) and AcmI (acylsulfatase) participate in this complex sulfate metabolite pathway is still elusive although bioinformatics analyses suggest that they have relevant functions (Fig. 3B). An aminotransferase AcmN was postulated to generate the 59-O-sulfonamide moiety by replacing a hydroxyl group of the sulfonate with an amine group (Fig. 3B).

Verification of the ascamycin/dealanylascamycin biosynthetic genes
To verify our hypothesis that this sulfate related pathway is responsible for ascamycin/dealanylascamycin biosynthesis, a genomic cosmid library of JCM9888 was constructed on shuttle vector SuperCos1. Using acmG gene as a DNA probe, four positive clones were identified from some 4,000 packaged cosmids. Gene inactivation was carried out to disrupt acmG (sulfatase) or acmK (sulfotransfer- ase) respectively (See material and methods) and the resulted mutants were validated by PCR analysis and subsequent sequencing using primers listed in Table S2 (See results in Fig. S1). When the Streptomyces mutants were fermented on petri dishes, as anticipated they lose their ability to elaborate both ascamycin and dealanylascamycin, as showed in HPLC-UV analysis (Fig. 4A). The molecular formula of compounds 1 and 2 were determined to be C 10  [M+H] + ) on the basis of ESI-HRMS, the isotopic pattern indicated ascamycin and dealanylascamycin each contained one Cl atom (Fig. 4BC). Detailed MS/MS pattern of ascamycin and dealanylascamycin are shown (Fig. 4D). The detected data is closely consistent with anticipated fragmentation ions and every mass of debris have been elucidated (Fig. 4D1: dealanylascamycin; Fig. 4D2: ascamycin).
AcmE is an esterase which is responsible for the conversion of dealanylascamycin to ascamycin Analysing ascamycin/dealanylascamycin gene cluster also revealed two alanyl-tRNA synthetases (acmD and acmF). Intriguingly, ascamycin has an alanyl-cap while dealanylascamycin lacks it. Gene acmE located between acmD and acmF whose translated product weakly resemble an aminopeptidase from Xanthomonas campestris (accession number BAA11623, 20% identity) which have been demonstrated in vitro to catalyse a hydrolysis reaction of ascamycin to dealanylascamycin causing the host to become susceptible to ascamycin [36]. Most bacteria lack this enzyme and therefore are resistant to ascamycin and this explain the very different bacterial activity spectrum of ascamycin and dealanylascamycin [37]. The location of acmCDE in the biosynthetic pathway suggests their role in converting dealanylascamycin to ascamycin via a tRNAdependent manner, in which acmE may catalyse a condensation reaction between alanyl-tRNA and dealanylascamycin to generate ascamycin [38].
To verify above hypothesis, gene inactivation was carried out on acmE and the Streptomyces mutant was validated by PCR on appropriate primers (Fig. S1-A). Intriguingly, acmE-null mutant could only produce dealanylascamycin but was blocked in its ability to biosynthesize ascamycin, as illustrated in HPLC-UV analysis ( Fig. 4A(iv)), confirming the role of acmE to catalyse the condensation reaction of dealanylascamycin and L-alanine to generate ascamycin. (Fig. 3A)

A proposed model for ACM/DACM biosynthesis
Thus we propose a model for ascamycin/dealanylascamycin biosynthesis as illustrated in Fig. 3A. Firstly, a phosphatase (AcmT) mediates phosphorylation at C1 of D-ribose to generate D-ribose-1-phosphate [39]. The next enzyme, a phosphoribosyltransferase (AcmV) catalyze the displacement of the phosphate group by adenine to form adenosine [40]. 59-O-sulfonamide biosynthesis related enzymes (AcmABGIKW) activate inorganic sulfate anions and install a sulfonate group onto the 59 position of adenosine. Then AcmN, an aminotransferase transfers an amine group onto the sulfonate moiety to generate the corresponding O-sulfonamide, which is dealanylascamycin. Finally, AcmE catalyse an amide formation between the carboxyl end of L-alanine and the NH 2 on the Osulfonamide group of dealanylascamycin in a tRNA-dependent manner to furnish ascamycin [38].

Discussion
Nucleoside antibiotics are a large family of compounds with significant activities and some of them have been developed into antibacterial, antifungal and antiparasite agents. The pace of discovering their biosynthetic pathways were increasing in the past decade, revealing new enzymes and paving the way to new ''unnatural'' natural products by synthetic biology. Ascamycin/dealanylascamycin are unusual nucleoside antibiotics which are elaborated by Streptomyces sp. They share a special 59-O-sulfonamide moiety on adenosine which is not present in other antibiotics to the best of our knowledge. In this study we identified the putative genes which are involved in their biosynthetic pathway, including those genes required for sulfonamide formation, chlorination and N-alanylation. The details of how dealanylascamycin is converted into ascamycin is an intriguing question that has yet to be answered. Here we present in vivo evidence to prove our hypothesis that an esterase AcmE are responsible for that. Bioinformatics analysis showed that an esterase AcmE was present which also had some similarity to the only aminopeptidase identified to-date that cleaves ascamycin to generate dealanylascamycin. The host only becomes (Xanthomonas sp.) susceptibile to ascamycin [24] when they have this aminopeptidase activity. We suspect acmE is involved in a reversible N-alanylation enzymatic reaction in the ascamycin/ dealanylascamycin pathway.
Ascamycin and dealanylascamycin are rare secondary metabolites decorated with sulfonamide groups at the 59-hydroxyl of the ribose ring. Six organosulfur metabolism genes (AcmABGKIW) and aminotransferase AcmN are candidate genes required for 59-O-sulfonamide formation. These genes showed little homology to other genes of sulfur-groups metabolism [eg. The caprazamycin metabolite pathway which involves a two-step sulfation:A triketide pyrones that are sulfated by an PAPS-dependent sulfotransferase (Cpz8) to generate phenolic sulfate esters which play a role as sulfate donors for a PAPS-independent arylsulfate sulfotransferase (Cpz4) to produce sulfated licaprazamycin(s) [41]]. Another intriguing question is halogenation of ACM/DACM, which probably involved AcmX and/or AcmY. Though they are 1Mb remote from AcmA-AcmW gene cluster, they can be regarded as candidate for adenine C-2 halogenase because they showed considerable homology to all FAD-type halogenase which are believed to be required for aromatic adenine ring chlorination. When JCM9888 genome was aligned with non-heme iron halogenases, SAM-dependent halogenases or vanadium and heme-dependent halogenses, no counterpart genes or homologues were identified [25] [26] [27]. We also checked the mycelium and spore formation of Streptomyces mutant DacmE, DacmG, DacmK but did not observed any difference between them and wild type JCM9888, suggesting that these genes or secondary metabolites ACM/DACM are not absolutely required for morphological differentiation.
It is an important aim of our project to provide more information to investigate the enigmatic fluorinated natural compound nucleocidin, which was discovered in 1969 [3], while all efforts to identify this famous fluorinated natural products from Streptomyces deposited in public collections failed in the past twenty years (Prof. David O9Hagan, University of St Andrews, UK, personal communication). Nucleocidin possess a very similar chemical structure with dealanylascamycin except that nucleocidin has a fluorine atom at C49 of the ribose while the later is chlorinated at C2 position of the adenine ring. We suspect that these two metabolic pathways use a set of very similar genes and proteins. Given access to the recently published genome of the putative nucleocidin producer Streptomyces calvus DSM40010 [42], we aligned the ascamycin/dealanylascamycin gene cluster against the S. calvus genome database, but no homologous counterpart was identified, suggesting that commercially obtained cultures of S. calvus DSM40010 (or ATCC13382) may not be the correct nucleocidin-producing strain. We suspect that a true nucleocidin-producing Streptomyces would also carry the related genes of ACM/DACM biosynthesis. Thus we envisage that with certain effort, it would be possible to re-isolate a nucleocidin-producing Streptomyces strain from such colonies by soil screening approach. Knowledge of the genes of related ACM/DACM biosynthesis would be necessary in order to devise PCR screening primers. This target-narrowing using gene PCR provides obvious advantages over verification by fermentation, extraction and metabolite analyses of candidate Streptomyces colonies individually.

Bacterial strains, plasmids and reagents
Bacterial strains and plasmids used and constructed in this study are listed in Table S1. Primers were chemically synthesized by commercial resources and listed in Table S2. strains JCM9888 were grown at 28˚C on ISP4 medium for growth and sporulation [43]. MS medium (soy bean flour 20 g, mannitol 20 g, agar 20 g, 2.5 mM MgSO 4 , 2.5 mM KCl in 1 L of tap water, pH 8.0) is used for petri dish fermentation of antibiotics ascamycin/dealanylascamycin [20,21]. ISP4 medium supplemented with 0.2% yeast extract is used for E. coli-Streptomyces bi-parental conjugation.

DNA extraction, manipulation and sequencing
Total DNA isolations, plasmid DNA preparations, restriction endonuclease digestions, ligations, and other DNA manipulations were performed according to standard procedures for E. coli and Streptomyces [43] [44]. The sequencing service at the Chinese National Genome Centre (Shanghai) was used for DNA sequencing. Computer-assisted database searching and sequence analysis were carried out online using the frame-plot 4.0 sequence analysis software (http:// nocardia.nih.go.jp/fp4/) and the BLAST program (http://blast.ncbi.nlm.nih.gov/ Blast.cgi).

Gene Inactivation Experiments
The lambda-RED-mediated gene replacements were performed as standard procedures [45]. In all cases, the aac(3)IV gene from pIJ773 was inserted within the coding region of the gene to be disrupted and in the same direction of transcription. The introduction of DNA into JCM9888 was achieved through intergeneric conjugation from E. coli ET12567/pUZ8002. After the introduction of the constructs into Streptomyces by intergeneric conjugation, exconjugants, in which a double crossover occurred, were identified by their resistance to apramycin and susceptibility to kanamycin.
To generate a mutant containing a deletion or an insertion mutation, the following primers were designed: (i) Mutagenesis of AcmE (esterase): AcmE-FP:  CCC GGC CAA CCG CGC TTC ATC TGC TCG ACG TTG ACC GTC att ccg ggg  atc cgt cga cc, AcmE-RP: CCG CAA CGA CCC GCG TGC GCC GCC CCG ATG  GTC AAG CGG tgt agg ctg gag ctg  The apramycin resistance cassette pIJ773 was used as a template throughout the experiments. The PCR products were transferred into competent cells of BW25113 (pIJ790, cosmid 15H6) to isolate apramycin-resistant and amplicillinresistant E. coli colonies. These mutated cosmids were introduced into ET12567 (pUZ8002) and then transferred by conjugation to JCM9888 to screen for apramycin resistant and kanamycin-sensitive colonies. This resulted in DacmE, DacmG and DacmK mutants respectively, which were validated by PCR and subsequent sequencing. (Fig. S1) Production, purification and analysis of ascamycin/ dealanylascamycin Ascamycin/dealanylascamycin isolation were performed according to the literature [20] [46]. Fermentation agar from petri dishes (weighing approx. 200 g) was extracted with 60% aqueous acetone. After removal of acetone by rotatory evaporation, the residual extract was passed through a Dowex 80WX8(H) resin column (20 ml), which was then eluted by 0.5 M ammonia solution.
The HPLC-MS analysis of the antibiotics obtained from above was performed on a C-18 reversed-phase column (150 by 4.6 mm; particle size, 5 mm [Phenomenex]) with UV detection at 258 nm. The solvent system consisted of solvent A (0.1% formic acid in water), and solvent B (CH 3 CN). The HPLC gradient was as follows: 10 % to 55 % solvent B (0 to 15 min), 55 % solvent B (15 to 18 min), 55 % to 10 % solvent B (18 to 20 min), and 10 % solvent B (20 to 25 min) at a flow rate of 0.5 ml/min.

Nucleotide sequence accession number
The complete sequence of the ascamycin/dealanylascamycin biosynthetic gene cluster and annotated proteins have been deposited in GenBank under accession number (GenBank accession number KJ817374 and KJ817375).