Specialized Metabolites from Ribosome Engineered Strains of Streptomyces clavuligerus

Abstract

Bacterial specialized metabolites are of immense importance because of their medicinal, industrial, and agricultural applications. Streptomyces clavuligerus is a known producer of such compounds; however, much of its metabolic potential remains unknown, as many associated biosynthetic gene clusters are silent or expressed at low levels. The overexpression of ribosome recycling factor (frr) and ribosome engineering (induced rpsL mutations) in other Streptomyces spp. has been reported to increase the production of known specialized metabolites. Therefore, we used an overexpression strategy in combination with untargeted metabolomics, molecular networking, and in silico analysis to annotate 28 metabolites in the current study, which have not been reported previously in S. clavuligerus. Many of the newly described metabolites are commonly found in plants, further alluding to the ability of S. clavuligerus to produce such compounds under specific conditions. In addition, the manipulation of frr and rpsL led to different metabolite production profiles in most cases. Known and putative gene clusters associated with the production of the observed compounds are also discussed. This work suggests that the combination of traditional strain engineering and recently developed metabolomics technologies together can provide rapid and cost-effective strategies to further speed up the discovery of novel natural products.

Keywords: Streptomyces; global molecular networking; metabolomics; ribosome engineering; ribosome recycling factor; specialized metabolites.

Figure 1

Overview of the numbers of spectral features detected in wild-type S. clavuligerus (wt) and strains overexpressing frr, rpsL, or rpsL variants (K88E, L90K, and R94G), respectively. The total number of spectral features detected in each ionization mode is indicated in parentheses.

Figure 2

Triterpenoid molecular networks and related biosynthetic gene clusters (BGCs) corresponding to metabolites detected only in S. clavuligerus strains overexpressing different rpsL variants (K88E, L90K, and R94G). (A) Negative-mode molecular network comprising unknown and predicted triterpenoids from the current study. (B) Predicted structures of saponin triterpenoids annotated with network annotation propagation (NAP). (C) Structural prediction of another triterpenoid molecular network detected using positive-mode ionization. (A–C) Each node depicts a mass spectrum (labeled with m/z of the respective precursor mass) and edges represent the relationship between different nodes. (B,C) Top-ranked NAP-consensus structural predictions (red boundary) and those annotated by spectral library matching (green boundary) present in S. clavuligerus cultures overexpressing different rpsL variants (K88E, L90K, and R94G) (yellow fill) are shown. (D) BGCs proposed to be associated with the production of such metabolites in S. clavuligerus. TT1 is responsible for the production of hopanoids, whereas the products of the other three BGCs are not known.

Figure 2

Triterpenoid molecular networks and related biosynthetic gene clusters (BGCs) corresponding to metabolites detected only in S. clavuligerus strains overexpressing different rpsL variants (K88E, L90K, and R94G). (A) Negative-mode molecular network comprising unknown and predicted triterpenoids from the current study. (B) Predicted structures of saponin triterpenoids annotated with network annotation propagation (NAP). (C) Structural prediction of another triterpenoid molecular network detected using positive-mode ionization. (A–C) Each node depicts a mass spectrum (labeled with m/z of the respective precursor mass) and edges represent the relationship between different nodes. (B,C) Top-ranked NAP-consensus structural predictions (red boundary) and those annotated by spectral library matching (green boundary) present in S. clavuligerus cultures overexpressing different rpsL variants (K88E, L90K, and R94G) (yellow fill) are shown. (D) BGCs proposed to be associated with the production of such metabolites in S. clavuligerus. TT1 is responsible for the production of hopanoids, whereas the products of the other three BGCs are not known.

Figure 3

Molecular network (negative ionization mode) and associated biosynthetic gene cluster (BGC) for flavonoid glycosides detected in S. clavuligerus strains overexpressing the K88E variant of rpsL (A) In silico structure prediction of metabolites using GNPS-based molecular networking and network annotation propagation (NAP). Each node depicts a mass spectrum (labeled with m/z of the respective precursor mass) and edges represent the relationship between different nodes. The top-ranked NAP-consensus structural predictions (red boundary) present in the S. clavuligerus strain overexpressing the K88E variant of rpsL (yellow fill) are shown. (B) The proposed BGC in S. clavuligerus associated with the biosynthesis of flavonoid glycosides, including the previously reported genes required for naringenin (flavonoid) production.

Figure 4

Molecular network (positive ionization mode) and associated biosynthetic gene cluster (BGC) for the tunicamycins and streptovirudins detected in S. clavuligerus strains. (A) Predicted structures of some metabolites using GNPS-based molecular networking and network annotation propagation (NAP). Each node depicts a mass spectrum (labeled with m/z of the respective precursor mass) and edges represent the relationship between different nodes. The top-ranked consensus structural predictions from NAP (red boundary) and GNPS (green boundary), or not predicted by both GNPS and NAP (black boundary) are shown. The pink nodes indicate the presence of the metabolites detected in both wild-type S. clavuligerus and the overexpression strains, whereas yellow nodes represent those detected only in strains overexpressing frr. Tunicamycins are included for comparison with the predicted structures. (B) The BGC is associated with the biosynthesis of tunicamycins and possibly that of the streptovirudins from S. clavuligerus.

Figure 5

Molecular network (positive ionization mode) for polycyclic tetramate macrolactams (PTMs) present in different cultures from the current study and the associated biosynthetic gene cluster (BGC) in S. clavuligerus. (A) In silico structure prediction of the metabolites by GNPS-based molecular networking and network annotation propagation (NAP). Each node depicts a mass spectrum (labeled with m/z of the respective precursor mass) and edges represent the relationship between different nodes. The top-ranked NAP-consensus structural predictions (red boundary) for the metabolites present in all S. clavuligerus cultures, including wild-type (pink fill), and only in the different overexpression strains (yellow fill), are shown. (B) The proposed BGC is associated with the biosynthesis of PTMs detected in S. clavuligerus.

Figure 6

Molecular network (negative ionization mode) and associated biosynthetic gene cluster (BGC) for bafilomycin in S. clavuligerus. (A) Predicted structure of bafilomycin in a GNPS-based molecular network obtained using network annotation propagation (NAP). Each node depicts a mass spectrum (labeled with m/z of the respective precursor mass) and the edges represent the relationship between different nodes. The top-ranked NAP-consensus structural predictions (red boundary) present in all S. clavuligerus strains (pink fill) and only in the rpsL-K88E overexpression strain (yellow fill) are shown. Bafilomycin A1 is included for comparison with the predicted bafilomycin from the current study. (B) The bafilomycin-like BGC present in S. clavuligerus is predicted to be involved in producing the metabolite.

Figure 7

Examples of other molecular networks detected in S. clavuligerus strains engineered from the current study. (A) A positive ionization mode molecular network of cembrane diterpenoids present in S. clavuligerus strains overexpressing different rpsL variants (K88E, L90K, R94G). (B) A negative ionization mode molecular network of organooxygen and organonitrogen compounds detected in S. clavuligerus overexpressing rpsL only. The structures were predicted by GNPS-based molecular networking and network annotation propagation (NAP). Each node depicts a mass spectrum (labeled with m/z of the respective precursor mass) and the edges represent the relationship between different nodes. The top-ranked NAP-consensus structural predictions (red boundary) present in the overexpression strains (yellow fill) are shown.