Graduate School of Agricultural and Life Sciences, The University of Tokyo 1-1-1 Yayoi, Bunkyo, Tokyo 113-8657, Japan
Actinomycetes are a major source of natural bioactive products with important chemical and biological properties.
Recent genome analysis has revealed the previously unrecognized huge potential of biosynthesis of natural products by actinomycetes. It is now generally accepted that more microbial chemical and biosynthetic diversities remain undiscovered. Increased knowledge of microbial production of bioactive compounds would increase the repertoire of useful agents. Moreover, bioengineering involving genes and enzymes would generate new useful compounds. However, this potential remains challenging.
Methods have been developed to activate biosynthetic gene clusters that are normally silent or poorly expressed under laboratory conditions. Section I highlights research aimed at the discovery of novel compounds by using co-culture, especially the interaction between intergeneric actinobacteria. Additionally, we discuss the importance of understanding the natural enzymatic assembly of complex small molecules in order to exploit new resources for biocatalysis, genes, and chemistry, which can lead to the creation of new antibiotics. This knowledge could enable the rational design of metabolic pathways to produce “artificial”
natural products in engineered bacteria. Section II details current research on the biosynthetic mechanisms of C7N aminocyclitol natural products having a unique chemical structure and important biological activities.
Section I
Bacterial interaction-driven natural product discovery
Actinomycetes are an important source of natural products with significant chemical and biological properties. Bioactive natural products isolated from actinomycetes have been used widely and include antibacterial, antifungal, and antiparasitic agents for treatment of infectious diseases; insecticides and herbicides for agricultural purposes; and anticancer and immunosuppressive drugs for clinical chemotherapy (Demain and Sanchez, 2009). However, the discovery rate of new antibiotics has been declining in recent decades, despite these successes and the more contemporary emerging/rising threats
to human health that include global expansion of multi-drug resistance bacteria (Martens and Demain, 2017) and neglected tropical diseases in developing countries (Buscaglia et al., 2015). New strategies and technologies are becoming indispensable for the effective discovery and/or generation of novel bioactive compounds (Katz and Baltz, 2016).
After the publication of genome sequences for the model actinomycetes Streptomyces coelicolor A3(2) (Bentley et al., 2002) and the avermectin-producer S. avermitilis MA-4680 (Ikeda et al., 2003) in the early 2000s, we quickly recognized that actinomycetes possess more potential to produce secondary metabolites than previously thought (Nett et al., 2009). The database maintained by the National Center for Biotechnology Information (NCBI) now contains several hundred actinomycete genome sequences (including draft genome sequences). Scrutiny of these sequences using bioinformatics tools like antiSMASH and PRISM has readily revealed putative secondary metabolite gene clusters (Blin et al., 2017; Skinnider et al., 2017). Twenty to forty putative secondary metabolite biosynthetic gene clusters have been identified in the genomes of individual strains. Most remain uncharacterized. The accepted view is that their remains vast chemical and biosynthetic diversities in the microbial world.
Understanding and exploiting the uncharacterized microbial chemistry would drive the discovery of new chemical agents to control bioactivity. Furthermore, the use of bioengineering tools including those directed at genes and enzymes would allow the creation of new useful compounds (Katz and Baltz, 2016). However, these goals remain challenged by the difficulty activating the relevant gene clusters and identifying their products. Methodologies to activate biosynthetic gene clusters that are silent or poorly expressed in laboratory conditions have been developed (Ochi, 2017). This review will focus on research that aims to achieve effective discovery of novel compounds by using a co-culture strategy, especially using an interaction of intergeneric actinobacteria involving Streptomyces species and mycolic acid-containing bacteria.
Bacterial co-culture as a means of discovering natural products
Isolated actinomycetes are traditionally cultured alone as a mono-culture to search for new natural products. However, the natural environment where actinobacteria live involve complex interactions at the intra and interspecies, genetic, and -kingdom levels (van der Meij et al., 2017). Yet, little is known about how the specialized metabolites encoded by cryptic gene clusters is used for the actinomycete life cycle in the complex, real-world environment (Traxler and Kolter, 2015). Discovery of useful bioactive natural products based on mono-culture has been successful, although this strategy is laborious. With the increasing evidence that bacterial interaction can drive the activation of previously quiescent secondary metabolite gene clusters (Bertrand et al., 2014), development and understanding of the induction of specialized metabolites during co-culture has become recognized as a research priority.
Interaction between Streptomyces lividans and mycolic acid-containing bacteria
Using the pigment production by Streptomyces lividans TK23 as indication of specialized metabolites activation, Onaka et al. (2011) discovered Tsukamurella pulmonis TP-B0596 from a laboratory bacterial culture collection. T. pulmonis phylogenetically belongs to the order Actinomycetales, the same order as Streptomyces species. T. pulmonis phylogenetically diverges to the suborder Corynebacterineae.
Most species in this suborder possess specific long chain fatty acids, mycolic acids, on the cell outer membrane (Jackson, 2014).
Examination of the interaction between S.
lividans and T. pulmonis in a dual culture agar plate experiment revealed a response by S. lividans featuring production of the red pigmented compound undecylprodigiosin upon contact with T. pulmonis colonies (Asamizu et al., 2015; Onaka et al., 2011). (Fig. 1) The production of the red pigment required the physical contact between the strains, since when the strains were physically separated during liquid culture using a dialysis membrane, the red pigment was not produced (Onaka et al., 2011). The mycolic acids on the T. pulmonis outer membrane
was implicated, given the similar effects on production of pigments in liquid culture by Corynebacterineae, which also possess mycolic acids (Onaka et al., 2011). To test the idea that contact with the mycolic acid-containing cell membrane was necessary to induce production of undecylprodigiosin, dead cells of T. pulmonis, which were intact and still contained mycolic acids, were prepared by formaldehyde fixation and gamma-irradiation (Asamizu et al., 2015). The dead cells did not induce the pigment formation by S. lividans, suggesting the involvement of another factor (Asamizu et al., 2015). (Fig. 1) When co-cultures of S. lividans and T. pulmonis or Rhodococcus opacus B4 were closely observed by scanning electron microscopy, co-aggregation was evident (Asamizu et al., 2015). (Fig. 1) The presence of an intimate relationship between microbes to alter the specialized pattern of metabolites has been described in the co-culture of the fungi Aspergillus nidulans or A. fumigatus with Streptomyces rapamycinicus (Netzker et al., 2015). Although contact-mediated interaction in microbes has not been well characterized yet (Stubbendieck and Straight, 2016; Westhoff et al., 2017), close distance recognition may well be beneficial in ecosystems such as soil.
Physicochemical-based discovery of specialized metabolites from combined-culture
Comparison of high performance liquid chromatography (HPLC) patterns between culture extracts from mono-cultures and combined-cultures has shown that T. pulmonis can markedly change its production of secondary metabolites.
Examination of 112 strains of actinomycetes isolated from soil samples collected in the Hokuriku district of Japan revealed new metabolite peaks in 41 strains, with increased production of metabolites in 61 strains. In total, 99 strains showed variation in the HPLC traces (Onaka et al., 2011). The same study documented that some of the soil-isolated actinomycetes showed the induced antibiotic activity in combined-culture.
Among them, the antibiotic alchivemycin A was isolated from a co-culture of Streptomyces sp. S522 (NBRC109436) and T.
pulmonis (Igarashi et al., 2010; Onaka et al., 2011). (Fig. 2)
Figure 1. Interaction between S. lividans TK23 and T. pulmonis TP-B0596 or R. opacus B4. Growing colony of T. pulmonis or R. opacus induced production of red pigments by S. lividans upon contact (A). R. opacus was observed to adhere on the mycelium of S. lividans during the liquid culture (B).
More recently, HPLC trace comparison-based screening of the new compounds from combined-cultures enabled the identification of new metabolites using Streptomyces species isolated from soil or obtained from culture collection that were co-cultured with T. pulmonis TP-B0596. These include indolocarbazole arcyriaflavin E production by S. cinnamoneus NBRC13823 (Hoshino et al., 2015c), cytotoxic butanolides chojalactone A–C from Streptomyces sp. CJ-5 (Hoshino et al., 2015b), and macrolactams niizalactam A–C from Streptomyces sp. NZ-6 (Hoshino et al., 2015a). (Fig. 2)
Similar co-culturing methods were reported by Bachmann and co-workers, in which comparative metabolomics enabled visualization of differentially expressed metabolites produced by S. coelicolor A3(2) with several known secondary metabolites inducing factors, such as rare earth elements, streptomycin/rifampicin resistance, and co-cultures (Goodwin
et al., 2015). Subtraction of a self-organizing heat map revealed differentially expressed metabolites; using several co-culture challengers, the authors found the mycolic acid-containing bacterial strain, Rhodococcus wratislaviensis, induced Nocardiopsis sp. FU40 ΔapoS strain to produce cytotoxic ciromicin A and B (Derewacz et al., 2015). (Fig. 2)
Traxler et al. (2013) used imaging mass spectrometry to visualize the secreted metabolome of S. coelicolor A3(2) and Amycolatopsis sp. AA4. They found that in consequence of amychelin production by Amycolatopsis sp. AA4, S. coelicolor A3(2) react to produce several new acyl-desferrioxamines, which are different from regular siderophores found to produce by S. coelicolor A3(2). The study highlight competition of bacteria using siderophores for Fe uptake (Traxler et al., 2013).
(Fig. 2)
Figure 2. Structure of induced specialized metabolites found in combined-culture, and other co-culture between intergeneric actinobacteria. Undecylprodigiosins and actinorhodins from S. lividans and T. pulmonis, alchivemycin A from Streptomyces sp. S522 and T.
pulmonis, 5aTHQs and streptoaminals from Streptomyces sp. HEK616 and T. pulmonis, arcyliaflavin E from S. cinnamoneus NRBC13823 and T.
pulmonis, Cyojalactone A-C from Streptomyces sp. CJ-5 and T. pulmonis, Niizalactame A-C from Streptomyces sp. NZ-6 and T. pulmonis, Ciromicin A and B from Nocardiopsis sp. FU40ΔapoS and R. wratislaviensis (study by Derewacz DK, et al. 2015), acyl-desferrioxamines from S.
coelicolor A3(2) and Amycolatopsis sp. AA4 (study by Traxler MF, et al. 2013)
Combined-culture with S. lividans harboring exogenous gene cluster
As production of several endogenous secondary metabolites from S. lividans TK23 (RED and ACT) were effectively induced by T. pulmonis, effects for production of exogenous gene cluster coding metabolites were examined (Onaka et al., 2015). Interestingly, when S. lividans mutant strains harboring exogenous gene clusters were cultured with T.
pulmonis, production of the exogenous secondary metabolites goadsporin (Onaka et al., 2001), staurosporine (Onaka et al., 2002), and rebeccamycin (Onaka et al., 2003) were significantly increased in mixed cultures compared to mono-culture (Onaka et al., 2015). The method was applied for gene disruptants; significantly improved accumulation of goadsporin C (a glutamylated-Ser4 variant of goadsporin B) was observed (Ozaki et al., 2016). This improved production of shunt intermediates contributed to the elucidation of important biosynthetic steps in the thiopeptide family of ribosomally synthesized peptide natural products (Ozaki et al., 2016).
Bioactivity-guided discovery of natural products from combined-cultures
Sugiyama et al. (2015) searched for the yeast membrane interacting small molecules from combined-culture induced bacterial metabolites. The extracts from combined-cultures of actinomycetes isolated from Hegura Island, Ishikawa, Japan, and T. pulmonis were tested against wild-type fission yeast and ergosterol premature mutants. This bioactivity-guided screening successfully led to the isolation of eight 5-alkyl-1,2,3,4-tetrahydroquinolines (5aTHQs) with diversity in the alkyl side chains (Sugiyama et al., 2015). (Fig. 2) 5aTHQ-7n was shown to be the most potent antifungal agent of the eight congeners. Moreover, 5aTHQ-9i showed selective antifungal activity to the wild-type, but not against ergosterol premature mutants (Sugiyama et al., 2015). The results suggested that 5aTHQs bioactivity may involve targeting of the yeast cell membrane. Sugiyama et al. (2016) also isolated broad-spectrum antibiotic streptoaminals from the combined-culture extracts containing a similar alkyl chain pattern to that of 5aTHQs (Sugiyama et al., 2016). (Fig. 2) Production of streptoaminals was enhanced by combined-culture. The structural similarity between 5aTHQs and streptoaminals implies that both compounds share their biosynthetic routes. Interestingly, 5aTHQs were only detected in the combined-culture of Streptomyces sp. HEK616 and T. pulmonis. However, 5aTHQs did not show antibacterial activities. Further biosynthesis studies may provide insight into the molecular mechanism of the specific production of 5aTHQs by Streptomyces sp.
HEK616 during co-culture with T. pulmonis.
Future perspectives
We have observed a variety of specialized metabolites induced during co-culture. However, the link between the induced small molecules and the function within the co-cultured bacteria remains unclear. One bacterium can cause significant changes in the culture (living) environment during the growth process, which can incidentally trigger the production of irrelative compounds. Interestingly, some of the induced compounds have antibiotic activity, which could reflect the
competition for survival between the two bacteria. However, knowledge is limited and more studies are needed to address a number of questions. What are the stimuli? How do bacteria sense the stimuli? How do the stimuli lead to the production of specialized metabolites? Are the interactions observed in laboratory co-culture relevant to real-world ecosystems?
Section II:
C7N aminocyclitol natural products
The C7N aminocyclitol family of natural products has clinically important biological activities; therefore, C7N aminocyclitol natural products and their derivatives have been used in agricultural and pharmaceutical fields (Mahmud, 2003).
The antifungal agent validamycin A (Iwasa et al., 1970) and α-glucosidase inhibitor acarbose (Schmidt et al., 1977) are prominent examples of C7N aminocyclitols, and these bacterial secondary metabolites are associated with pseudo-oligosaccharides (or simply pseudosugars), which function as sugar hydrolase inhibitors (Gloster and Davies, 2010; Mahmud, 2003). One unique structural feature in this family is their C7N carbasugar scaffold, primarily valienamine moieties. In addition to validamycin A and acarbose, typical compounds that contain valienamine moieties also include the trehalase inhibitor salbostatin (Vertesy et al., 1994), α-amylase inhibitor trestatins (Yokose et al., 1983), and antibiotic pyralomicin (Kawamura et al., 1995) (Fig. 3). Along with the recent discovery of novel cyclitol natural products and an understanding of their origins, biosynthesis, biological activities, and ecological functions, the structurally more diverse family of C7N aminocyclitols, which includes the cytotoxic carbasugar cetoniacytone A (Schlorke et al., 2002), antibiotic epoxyquinomicin (Tsuchida et al., 1996), and kirkamide (Pinto-Carbo et al., 2016; Sieber et al., 2015), has been identified. Mycosporine-like amino acids (e.g., shinorine) are natural sunscreen compounds that have the same precursor and share homologous biosynthetic enzymes in the initial step (Asamizu et al., 2012; Balskus and Walsh, 2010; Mahmud, 2003; Miyamoto et al., 2014; Wu et al., 2007) (Fig. 3). In this review, we discuss the recently investigated biosynthetic steps of C7N aminocyclitol natural products. In particular, we discuss the pseudoglycosyltranferase-catalyzed C-N bond formation process during validamycin A biosynthesis and the catalytic divergence of sugar phosphate cyclases leading to the generation of various C7N cyclitol natural products.
Biosynthesis of the antifungal trehalase inhibitor validamycin A
Validamycin A was originally isolated from Streptomyces hygroscopicus subsp. limoneus by a group from Takeda Pharmaceutical Co. in the early 1970s (Iwasa et al., 1970). The compound inhibits the growth of the fungus Rhizoctonia solani, which causes sheath blight disease in rice (Iwasa et al., 1970) by inhibiting the activity of trehalase (Asano et al., 1987).
Therefore, the antifungal agent validamycin A has been used as a crop protectant in East/Southeast Asia. Later, the α-glucosidase inhibitor voglibose (Fig. 3) was synthesized from validamycin A and used to treat type-II insulin-independent diabetes.
The biosynthetic gene cluster for validamycin A was first cloned from Streptomyces hygroscopicus subsp. jingangensis 5008 (val cluster) (Yu et al., 2005). The first step in the secondary metabolism of validamycin is catalyzed by 2-epi-5-epi-valiolone synthase (EEVS), which converts D -sedoheptulose 7-phosphate (SH7P), a pentose phosphate pathway intermediate, to 2-epi-5-epi-valiolone (EEV; Fig. 4A).
The first EEVS (AcbC) was found and characterized in the acarbose biosynthetic pathway from Actinoplanes sp. SE 50/110 by precursor feeding studies (Mahmud et al., 1999) and biochemical studies (Stratmann et al., 1999). Later, ValA was found to be EEVS in the validamycin A biosynthetic pathway (Yu et al., 2005).
Interestingly, the biosynthesis of C7N aminocyclitols was suggested to diverge into several assembly lines, including those for validamycin A, acarbose (Rockser and Wehmeier, 2009), salbostatin (Choi et al., 2008), pyralomicin 1a (Flatt et al., 2013), and cetoniacytone A (Wu et al., 2009), after the formation of EEV. In this review, the validamycin A biosynthetic pathway will be described in detail. A summary of the biosynthetic pathways for other cyclitols can be found in previous reviews (Flatt and Mahmud, 2007; Mahmud, 2009).
The second step in validamycin A biosynthesis, epimerization of EEV to generate 5-epi-valiolone, was found to be catalyzed by cyclitol epimerase ValD in vitro (Xu et al., 2009b) (Fig. 4A). Next, dehydration of 5-epi-valiolone to produce valienone is thought to be catalyzed by ValK, a putative dehydratase (Cui et al., 2016). Then, ATP-dependent phosphorylation of valienone to produce valienone 7-phosphate is catalyzed by cyclitol kinase ValC in vitro (Minagawa et al., 2007). After the formation of valienone 7-phosphate, the pathway was predicted to branch into two pathways to generate two different C7 cyclitol units, GDP-valienol and validamine
7-phosphate.
To produce GDP-valienol, the first ketoreduction of valienone 7-phosphate yields valienol 7-phosphate through the function of ValN, a putative bifunctional oxidoreductase (Fig.
4A). Next, phosphomutation of valienol 7-phosphate to give valienol 1-phosphate is thought to occur through the activity of ValO, a putative bifunctional phosphomutase/phosphatase.
Then, using valienol 1-phosphate and GTP, the nucleotidylation reaction produces GDP-valienol through catalysis by ValB in vitro (Yang et al., 2011).
To produce validamine 7-phosphate, transamination of valienone 7-phosphate yields valienamine 7-phosphate through the activity of ValM, a putative pyridoxal 5′-phosphate (PLP)-dependent aminotransferase (Fig. 4A). Then, reduction of valienamine 7-phosphate to give validamine 7-phosphate occurs through catalysis by ValN (Xu et al., 2009a). The coupling reaction of the two cyclitol units (GDP-valienol and validamine 7-phosphate) will be described in the next section.
After formation of validoxylamine A, ValG catalyzes the O-glucosyltransferase reaction to produce the final product validamycin A (Bai et al., 2006; Xu et al., 2008).
Several oxygenated validamycin derivatives have also been isolated from cultures of validamycin A-producing Streptomyces species (Mahmud, 2003). VldW, an α-ketoglutarate and Fe(II) dependent dioxygenase from Streptomyces hygroscopicus var. linoneus (vld cluster) (Singh et al., 2006), was characterized and found to catalyze the production of validamycin B from validamycin A by the regio-/stereo-selective oxygenation of the methylene carbon (Almabruk et al., 2012) (Fig. 4A).
Nonglycosidic C-N bond formation in validamycin A biosynthesis is catalyzed by pseudoglycosyltransferase Figure 3. Structure of C7N aminocyclitol natural products.
ValL/VldE shares 29% identity (41% similarity) with trehalose 6-phosphate synthase (OtsA) from Streptomyces coelicolor A3(2). OtsA is a retaining-type glycosyltransferase that synthesizes trehalose 6-phosphate with an α,α-1,1′-glycosidic bond using nucleotide diphosphate (NDP)-glucose and glucose 6-phosphate, and OtsB dephosphorylates trehalose 6-phosphate to give trehalose (Giaever et al., 1988). Yang et al.
showed that ValB catalyzes the formation of GDP-valienol from valienol 1-phosphate and GTP (Yang et al., 2011). Hence, researchers hypothesized that validoxylamine A 7′-phosphate (mimic of trehalose 6-phosphate) may be produced by the coupling of GDP-valienol (mimic of NDP-glucose) and validamine 7-phosphate (mimic of glucose 6-phosphate; Fig.
4A). However, for the retaining-type glycosyltransferase reaction, such as in OtsA, an internal return (SNi)-like reaction mechanism in which the donor sugar molecule is in the oxocarbenium transition state, may exist during the reaction (Errey et al., 2010; Lee et al., 2011). Since the cyclitol molecules (valienol moiety) cannot form the “oxocarbenium
transition state”, it remains uncertain whether this prediction is true.
VldB (cyclitol nucleotidyltransferase) (Yang et al., 2011), VldE (trehalose 6-phosphate synthase homolog), and VldH (putative phosphatase) from Streptomyces hygroscopicus subsp.
limoneus were expressed in Escherichia coli and purified as recombinant proteins to test the hypothesis (Asamizu et al., 2011). First, VldB was confirmed to be a nucleotidyltransferase that gave GDP-valienol from valienol 1-phosphate and GTP (Asamizu et al., 2011) (Fig 4A). Next, to examine the catalytic activity of VldE, the enzyme was incubated with the possible substrates validamine 7-phosphate and GDP-valienol. High-performance liquid chromatography (HPLC) and mass spectrometry (MS) analyses showed that VldE catalyzed the formation of validoxylamine 7′-phosphate with net retention of an anomeric-like configuration by accepting GDP-valienol and validamine 7-phosphate as substrates (Asamizu et al., 2011) (Fig. 4B). Interestingly, VldE did not accept GDP-glucose and glucose phosphate as substrates to produce trehalose 6-Figure 4. Proposed biosynthetic pathway of validamycin A from Streptomyces hygroscopicus (A), and proposed reaction mechanism of pseudoglycosyltransferase, VldE (B).
phosphate. This indicated the narrow substrate tolerance of the dedicated VldE enzyme in validamycin A biosynthesis (Abuelizz and Mahmud, 2015; Asamizu et al., 2011). Upon the addition of VldH, formation of validoxylamine A by consumption of validoxylamine A 7′-phosphate was observed by HPLC and MS analysis (Asamizu et al., 2011). These biochemical investigations could clearly demonstrate the interesting enzymatic conversion steps in validamycin A biosynthesis (Asamizu et al., 2011).
To investigate the reaction mechanism through which VldE catalyzes the coupling of the “nonsugar” donor molecule (GDP-valienol) and the acceptor molecule (validamine 7-phosphate) with the retention of stereochemistry, a series of VldE crystal structures cocrystallizing with different ligands were solved (Cavalier et al., 2012). The overall X-ray crystal structure of VldE showed a typical GT-B fold with two β/α/β Rossmann-like folding domains (Lairson et al., 2008). The products of VldE, i.e., GDP and validoxylamine A 7′-phosphate, were found to bind in the cleft formed by the two domains, indicating the position of the active center. Interestingly, the cocrystallized structure of VldE with GDP and validoxylamine A 7′-phosphate showed a ligand-binding conformation that was
similar to the cocrystallized X-ray structure of glycosyltransferase OtsA from E. coli with UDP and the mechanistic inhibitor validoxylamine 7′-phosphate (Cavalier et al., 2012; Errey et al., 2010; Lee et al., 2011). These structural comparisons indicated that OtsA, the true retaining-type glycosyltransferase OtsA, and the pseudoglycosyltransferase VldE exhibited similar reaction mechanisms. Thus, analogous to the proposed reaction mechanism for the retaining-type glycosyltransferase OtsA (Lee et al., 2011), hydrogen bonding interactions among the donor phosphate group and acceptor nucleophile were proposed to enable front side attack to promote the substitution reaction while retaining its configuration, and the double bond π-electron of the donor nonsugar was predicted to mimic the transition state in the PsGT reaction (Asamizu et al., 2011; Cavalier et al., 2012) (Fig.
4B).
Based on biochemical and structural studies, Abuelizz and Mahmud produced domain-swapped chimeric proteins between VldE and OtsA from Streptomyces coelicolor A3(2) and examined the catalytic activity of the “chimeras” to elucidate their substrate tolerances (Abuelizz and Mahmud, 2015). By swapping the substrates, they showed the potential for Figure 5. Catalytic divergence in sugar phosphate cyclase family enzymes. SH7P cyclases (EEVS, EVS, and DDGS) involved in biosynthesis of C7N aminocyclitol natural products share homologies with DHQS, aDHQS, and DOIS. These enzymes represent the family of sugar phosphate cyclase involved in primary and secondary metabolism. (Abbreviation: DAHP; Deoxy-D-arabinoheptulosonate 7-phosphate, DHQ; 3-dehydroquinic acid, aminoDAHP; 3,4-dideoxy-4-amino-D-arabinoheptulosonate 7-phosphate, 3,5-AHBA; 3-amino-5-hydroxybenzoic acid, SH7P; D-sedoheptulose 7-phosphate, EEV; 2-epi-5-epi-valiolone, 2EV; 2-epi-valiolone, DDG; desmethyl-4-deoxygadusol, MAA; mycosporine-like amino acid, G6P; glucose 6-phosphate, DOI; 2-deoxy-scyllo-inosose.)
biocatalysis of engineered proteins and demonstrated the importance of the amine group as a better nucleophile to promote the coupling reaction (Abuelizz and Mahmud, 2015).
Further characterization of other “PsGT” candidates found in acarbose, salbostatin, pyralomicins, and many other compounds in genome databases will expand our knowledge of the unique PsGT-catalyzed reaction. Since true glycosyltransferase enzymes are ubiquitous in both primary and secondary metabolism (Elshahawi et al., 2015), protein engineering of a glycosyltransferase to be a PsGT catalyst would allow the creation of useful tools to generate novel pseudoglycosylated conjugants.
A divergent pathway for production of C7N cyclitols from sedoheptulose 7-phosphate
During genome mining to search for structurally novel aminocyclitol natural products, PsGT-containing gene clusters with 3-dehydroquinate synthase (DHQS) homolog genes were unexpectedly found in the genomes of several bacteria (Asamizu et al., 2012). Previously characterized gene clusters for C7N aminocyclitol natural products were found to all contain EEVS genes (Bai et al., 2006; Choi et al., 2008; Flatt and Mahmud, 2007; Stratmann et al., 1999; Wu et al., 2007; Wu et al., 2009; Yu et al., 2005). Although EEVS genes share homology with 3-dehydroquinate synthase (Stratmann et al., 1999; Wu et al., 2007), the identified putative proteins (e.g., Amir_2000 from the actinomycete Actinosynnema mirum and Staur_3140 from the myxobacteria Stigmatella aurantiaca DW4/3-1) showed more similarity in their fingerprint amino acid residues to DHQS than to EEVS (Kean et al., 2014; Wu et al., 2007) and existed in different phylogenetic clades from known EEVSs (Asamizu et al., 2012; Osborn et al., 2015).
To characterize the catalytic function of the genes, Amir_2000 and Staur_3140 were examined for enzymatic activities using purified recombinant proteins expressed in E.
coli. First, as annotated in the NCBI database, the proteins were tested for their DHQS activity by incubating them with 3-deoxy-D-arabinoheptulosonate 7-phosphate (DAHP), a substrate of DHQS in the shikimate pathway; however, no consumption of DAHP was observed (Asamizu et al., 2012).
Then, SH7P was tested as a substrate and incubated with the proteins. Consumption of SH7P was observed; however, surprisingly, the converted products showed different chemical properties from EEV, the most likely product to be generated (Asamizu et al., 2012). Comparative analysis with synthetic C7
cyclitols which exhibited different stereochemistries, by gas chromatography (GC)/MS and in situ nuclear magnetic resonance (NMR) revealed that the true product was 2-epi-valiolone (2EV), a diastereomer of EEV (Asamizu et al., 2012) (Fig. 5). Interestingly, in situ NMR gave a two sets of 1H NMR signals for the products, which were confirmed by a quantum mechanics/molecular mechanics (QM/MM) study to be derived from two stable conformations of 2EV (Asamizu et al., 2012).
To further elucidate the metabolite(s) of the gene cluster that encodes the 2EV synthase (EVS) gene (amir_2000) and the PsGT homolog gene (amir_1997), both genes in A. mirum were disrupted individually, and the metabolites from culture of the wild-type gene, amir_1997 disruptant, and amir_2000 disruptant were analyzed by comparative metabolomics using
liquid chromatography-high-resolution MS (LC-HRMS) (Asamizu et al., 2013). The identified specific metabolite with m/z 314 was purified, and the chemical structure was determined to be validoxylamine A (Asamizu et al., 2013) (Fig.
4A). Thus, there are pathways with different steric courses in the assembly pathway for C7N aminocyclitol natural products.
Other sugar phosphate cyclase (SPC) members are also involved in the biosynthesis of mycosporine-like amino acids by several cyanobacteria (Wu et al., 2007). Balskus and Walsh identified a biosynthetic gene cluster for the mycosporine-like amino acid shinorine in cyanobacteria (Balskus and Walsh, 2010) (Fig. 5). They demonstrated the enzyme activities of Ava_3858 (desmethyl-4-deoxygadusol synthase [DDGS]) and Ava_3857 (S-adenosylmethionine [SAM]-dependent methyltransferase) from the cyanobacteria Anabaena variabilis ATCC 29413, which generated 4-deoxygadusol from SH7P (Balskus and Walsh, 2010) (Fig. 5). DDGS and EEVS share homology with each other (Wu et al., 2007); therefore, researchers tested whether there was a common intermediate during the DDGS reaction, in which additional dehydration was involved. The reaction of Ava_3858 and Npun_5600 (DDGSs from the cyanobacteria Nostoc punctiforme PCC 73102) (Balskus and Walsh, 2010) was traced by in situ NMR, and the
1H NMR signals showed only the chemical shifts for DDG, indicating that no detectable intermediate was generated during the entire reaction (Asamizu et al., 2012).
To gain insights into how three homologous enzymes (EEVS, EVS, and DDGS) catalyze different cyclization reactions using the same substrate SH7P, the crystal structures of ValA (EEVS: 4P53) (Kean et al., 2014) and Ava_3858 (DDGS: 5TPR) (Osborn et al., 2017a) were solved. The crystal structures of ValA and Ava_3858 were found to be cocrystalized with NAD+ and Zn2+ and showed folds that were similar to those of DHQS (Carpenter et al., 1998). A comparison of the amino acid residues forming the catalytic pocket among EEVS, DDGS, and DHQS provided some insights into the fingerprint amino acid residues used for accurate annotation of gene function for similar enzymes (Kean et al., 2014; Osborn et al., 2017a).
However, swapping the amino acid residues that are specifically conserved in each enzyme (L267E/D281A/H360T for ValA;
E254L/A268D/T347H for Ava_3858) did not convert the activity of the enzyme; thus, it remained unclear how the additional dehydration process could proceed in the DDGS reaction (Osborn et al., 2017a).
In addition, during genome mining using the EEVS sequence as a probe, unexpectedly, homologous genes were found in the genome of vertebrates, such as fish, birds, reptiles, and amphibians (Osborn et al., 2015). Interestingly, the putative EEVS genes from animals were flanked by a putative protein with an oxidoreductase (Ox) domain and a methyltransferase (MT) domain, and were also flanked by putative transcriptional regulators (Osborn et al., 2015). The genes for the EEVS homolog and the Ox-MT di-domain protein were synthesized based on the zebrafish (Danio rerio) sequences and expressed in E. coli. The purified recombinant DrEEVS was shown to synthesize EEV from SH7P. Furthermore, co-incubation with Ox-MT and SAM resulted in the formation of gadusol from EEV (Osborn et al., 2015) (Fig. 5). Gadusol is a natural sunscreen compound that possesses UV-resistance activity
(Shick and Dunlap, 2002). Gadusol was originally identified in fish eggs and was believed to accumulate from consumption in the diet (Shick and Dunlap, 2002). However, this study provided insights into the animal de novo synthetic pathway of a natural C7 cyclitol sunscreen compound that exists not only in prokaryotes but also in higher organisms, such as vertebrates (Osborn et al., 2015).
Until recently, the EEVS involved in the biosynthesis of C7N aminocyclitol natural products from actinomycetes was the only characterized SH7P cyclase, a family of enzymes that convert SH7P to carbocyclic molecules, such as EEV, 2EV, and DDG (Osborn et al., 2017b). However, genomic and biochemical investigations have revealed that SH7P cyclase is distributed in a wide range of species, including actinomycetes, cyanobacteria (Asamizu et al., 2012; Balskus and Walsh, 2010), myxobacteria (Asamizu et al., 2012), and vertebrates, such as zebrafish (D. rerio) (Osborn et al., 2015). Further bioinformatics analysis of homologous genes for SH7P cyclase revealed that this gene is widely distributed in a variety of organisms (Osborn et al., 2017a; Osborn et al., 2017b). These recently investigated SH7P cyclase genes will provide a template for easier access to gene clusters for new C7N aminocyclitols that are buried inside the growing genome databases.
Conclusion and perspective
In this section, I reviewed recent progress in research on the biosynthesis of C7N aminocyclitol validamycin A. These recent investigations have revealed the intriguing assembly pathways of these secondary metabolites by unique enzymes.
The recent expansion of genome databases has been a driving force in the discovery of unprecedented biological, chemical, and catalytic repertoires, which can provide challenges to create novel bioactive “artificial” natural products. Further accumulation of knowledge is indispensable and will facilitate the development of new technologies to achieve the aim of creating novel cell factories that can synthesize complex small molecules.
Acknowledgments
I am grateful to Prof. Hiroyasu Onaka at The University of Tokyo, and Prof. Taifo Mahmud at Oregon State University for their kind support. I would like to acknowledge Prof. Andrew Karplus at Oregon State University, Prof. Yong-Hwan Lee at Louisiana State University, Dr. Hideaki Kakeya at Kyoto University, Dr. Satoh Katsuya at QST, and Dr. Kanae Teramoto at JOEL Ltd. for research collaboration. I would like to thank Dr. Taro Ozaki, Dr. Shohei Hayashi, and Dr. Yoshinori Sugai for their kind support. I would also like to thank all present and past laboratory members at The University of Tokyo and Oregon State University for their kind support. Finally, I would like to thank the Society of Actinomycetes Japan for the Hamada Award.
References
Abuelizz, H.A., and Mahmud, T. (2015). Distinct substrate specificity and catalytic activity of the
pseudoglycosyltransferase VldE. Chem. Biol. 22, 724-733.
Almabruk, K.H., et al. (2012). The alpha-ketoglutarate/Fe(II)-dependent dioxygenase VldW is responsible for the formation of validamycin B. Chembiochem. 13, 2209-2211.
Asamizu, S., et al. (2011). Pseudoglycosyltransferase catalyzes nonglycosidic C-N coupling in validamycin A biosynthesis.
J. Am. Chem. Soc. 133, 12124-12135.
Asamizu, S., et al. (2012). Evolutionary divergence of sedoheptulose 7-phosphate cyclases leads to several distinct cyclic products. J. Am. Chem. Soc. 134, 12219-12229.
Asamizu, S., et al. (2013). Comparative metabolomic analysis of an alternative biosynthetic pathway to pseudosugars in Actinosynnema mirum DSM 43827. Chembiochem. 14, 1548-1551.
Asamizu, S., et al. (2015). Killing of Mycolic Acid-Containing Bacteria Aborted Induction of Antibiotic Production by Streptomyces in Combined-Culture. PLoS One. 10, e0142372.
Asano, N., et al. (1987). Effect of validamycins on glycohydrolases of Rhizoctonia solani. J. Antibiot. 40, 526-532.
Bai, L., et al. (2006). Functional analysis of the validamycin biosynthetic gene cluster and engineered production of validoxylamine A. Chem. Biol. 13, 387-397.
Balskus, E.P., and Walsh, C.T. (2010). The genetic and molecular basis for sunscreen biosynthesis in cyanobacteria.
Science. 329, 1653-1656.
Bentley, S.D., et al. (2002). Complete genome sequence of the model actinomycete Streptomyces coelicolor A3(2). Nature.
417, 141-147.
Bertrand, S., et al. (2014). Metabolite induction via microorganism co-culture: a potential way to enhance chemical diversity for drug discovery. Biotechnol Adv. 32, 1180-1204.
Blin, K., et al. (2017). antiSMASH 4.0-improvements in chemistry prediction and gene cluster boundary identification. Nucleic Acids Res.
Buscaglia, C.A., et al. (2015). Neglected Tropical Diseases in the Post-Genomic Era. Trends Genet. 31, 539-555.
Carpenter, E.P., et al. (1998). Structure of dehydroquinate synthase reveals an active site capable of multistep catalysis.
Nature. 394, 299-302.
Cavalier, M.C., et al. (2012). Mechanistic insights into validoxylamine A 7'-phosphate synthesis by VldE using the structure of the entire product complex. PLoS One. 7, e44934.
Choi, W.S., et al. (2008). Genetic organization of the putative salbostatin biosynthetic gene cluster including the 2-epi-5-epi-valiolone synthase gene in Streptomyces albus ATCC 21838. Appl. Microbiol. Biotechnol. 80, 637-645.
Cui, L., et al. (2016). De Novo biosynthesis of beta-valienamine in engineered Streptomyces hygroscopicus 5008.
ACS Synth. Biol. 5, 15-20.
Demain, A.L., and Sanchez, S. (2009). Microbial drug discovery: 80 years of progress. J. Antibiot. 62, 5-16.
Derewacz, D.K., et al. (2015). Mapping Microbial Response Metabolomes for Induced Natural Product Discovery. ACS Chem. Biol. 10, 1998-2006.
Elshahawi, S.I., et al. (2015). A comprehensive review of glycosylated bacterial natural products. Chem. Soc. Rev. 44,
7591-7697.
Errey, J.C., et al. (2010). Mechanistic insight into enzymatic glycosyl transfer with retention of configuration through analysis of glycomimetic inhibitors. Angew. Chem. Int. Ed.
49, 1234-1237.
Flatt, P.M., and Mahmud, T. (2007). Biosynthesis of aminocyclitol-aminoglycoside antibiotics and related compounds. Nat. Prod. Rep. 24, 358-392.
Flatt, P.M., et al. (2013). Genetic insights into pyralomicin biosynthesis in Nonomuraea spiralis IMC A-0156. J. Nat.
Prod. 76, 939-946.
Giaever, H.M., et al. (1988). Biochemical and genetic characterization of osmoregulatory trehalose synthesis in Escherichia coli. J. Bacteriol. 170, 2841-2849.
Gloster, T.M., and Davies, G.J. (2010). Glycosidase inhibition: assessing mimicry of the transition state. Org.
Biomol. Chem. 8, 305-320.
Goodwin, C.R., et al. (2015). Structuring Microbial Metabolic Responses to Multiplexed Stimuli via Self-Organizing Metabolomics Maps. Chem. Biol. 22, 661-670.
Hoshino, S., et al. (2015a). Niizalactams A-C, Multicyclic Macrolactams Isolated from Combined Culture of Streptomyces with Mycolic Acid-Containing Bacterium. J.
Nat. Prod. 78, 3011-3017.
Hoshino, S., et al. (2015b). Chojalactones A-C, cytotoxic butanolides isolated from Streptomyces sp. cultivated with mycolic acid containing bacterium. Org. Lett. 17, 1501-1504.
Hoshino, S., et al. (2015c). Arcyriaflavin E, a new cytotoxic indolocarbazole alkaloid isolated by combined-culture of mycolic acid-containing bacteria and Streptomyces cinnamoneus NBRC 13823. J. Antibiot. 68, 342-344.
Igarashi, Y., et al. (2010). Alchivemycin A, a bioactive polycyclic polyketide with an unprecedented skeleton from Streptomyces sp. Org. Lett. 12, 3402-3405.
Ikeda, H., et al. (2003). Complete genome sequence and comparative analysis of the industrial microorganism Streptomyces avermitilis. Nat Biotechnol. 21, 526-531.
Iwasa, T., et al. (1970). Studies on validamycins, new antibiotics. I. Streptomyces hygroscopicus var. limoneus nov.
var., validamycin-producing organism. J. Antibiot. 23, 595-602.
Jackson, M. (2014). The mycobacterial cell envelope-lipids.
Cold Spring Harb Perspect Med. 4.
Katz, L., and Baltz, R.H. (2016). Natural product discovery:
past, present, and future. J Ind Microbiol Biotechnol. 43, 155-176.
Kawamura, N., et al. (1995). Pyralomicins, new antibiotics from Actinomadura spiralis. J. Antibiot. 48, 435-437.
Kean, K.M., et al. (2014). Structure of a sedoheptulose 7-phosphate cyclase: ValA from Streptomyces hygroscopicus.
Biochemistry. 53, 4250-4260.
Lairson, L.L., et al. (2008). Glycosyltransferases: structures, functions, and mechanisms. Annu. Rev. Biochem. 77, 521-555.
Lee, S.S., et al. (2011). Mechanistic evidence for a front-side, SNi-type reaction in a retaining glycosyltransferase. Nat.
Chem. Biol. 7, 631-638.
Mahmud, T., et al. (1999). Biosynthetic studies on the alpha-glucosidase inhibitor acarbose in Actinoplanes sp.:
2-epi-5-epi-valiolone is the direct precursor of the valienamine moiety. J. Am. Chem. Soc. 121, 6973-6983.
Mahmud, T. (2003). The C7N aminocyclitol family of natural products. Nat. Prod. Rep. 20, 137-166.
(2009). Progress in aminocyclitol biosynthesis. Curr. Opin.
Chem. Biol. 13, 161-170.
Martens, E., and Demain, A.L. (2017). The antibiotic resistance crisis, with a focus on the United States. J Antibiot (Tokyo). 70, 520-526.
Minagawa, K., et al. (2007). ValC, a new type of C7-cyclitol kinase involved in the biosynthesis of the antifungal agent validamycin A. Chembiochem. 8, 632-641.
Miyamoto, K.T., et al. (2014). Discovery of gene cluster for mycosporine-like amino acid biosynthesis from Actinomycetales microorganisms and production of a novel mycosporine-like amino acid by heterologous expression.
Appl. Environ. Microbiol. 80, 5028-5036.
Nett, M., et al. (2009). Genomic basis for natural product biosynthetic diversity in the actinomycetes. Nat Prod Rep.
26, 1362-1384.
Netzker, T., et al. (2015). Microbial communication leading to the activation of silent fungal secondary metabolite gene clusters. Front Microbiol. 6, 299.
Ochi, K. (2017). Insights into microbial cryptic gene activation and strain improvement: principle, application and technical aspects. Journal of Antibiotics. 70, 25-40.
Onaka, H., et al. (2001). Goadsporin, a chemical substance which promotes secondary metabolism and morphogenesis in streptomycetes. I. Purification and characterization. J.
Antibiot. 54, 1036-1044.
Onaka, H., et al. (2002). Cloning of the staurosporine biosynthetic gene cluster from Streptomyces sp. TP-A0274 and its heterologous expression in Streptomyces lividans. J.
Antibiot. 55, 1063-1071.
(2003). Characterization of the biosynthetic gene cluster of rebeccamycin from Lechevalieria aerocolonigenes ATCC 39243. Biosci. Biotechnol. Biochem. 67, 127-138.
Onaka, H., et al. (2011). Mycolic acid-containing bacteria induce natural-product biosynthesis in Streptomyces species.
Appl. Environ. Microbiol. 77, 400-406.
Onaka, H., et al. (2015). Mycolic acid-containing bacteria activate heterologous secondary metabolite expression in Streptomyces lividans. J. Antibiot. 68, 594-597.
Osborn, A.R., et al. (2015). De novo synthesis of a sunscreen compound in vertebrates. Elife. 4.
Osborn, A.R., et al. (2017a). Evolution and Distribution of C7-Cyclitol Synthases in Prokaryotes and Eukaryotes. ACS Chem Biol. 12, 979-988.
Osborn, A.R., et al. (2017b). The sedoheptulose 7-phosphate cyclases and their emerging roles in biology and ecology. Nat Prod Rep.
Ozaki, T., et al. (2016). Insights into the Biosynthesis of Dehydroalanines in Goadsporin. Chembiochem. 17, 218-223.
Pinto-Carbo, M., et al. (2016). Evidence of horizontal gene transfer between obligate leaf nodule symbionts. Isme Journal. 10, 2092-2105.
Rockser, Y., and Wehmeier, U.F. (2009). The gac-gene cluster for the production of acarbose from Streptomyces glaucescens GLA.O: identification, isolation and