The RNA-guided DNA editing technology CRISPRs (clustered regularly interspaced short palindromic repeats)/Cas9 had been used to introduce double-stranded breaks into genomes and to direct subsequent site-specific insertions/deletions or the replacement of genetic material in bacteria, such as Escherichia coli, Streptococcus pneumonia, and Lactobacillus reuteri. In this study, we established a high-efficiency CRISPR/Cas9 genome editing plasmid pKCcas9dO for use in Streptomyces genetic manipulation, which comprises a target-specific guide RNA, a codon-optimized cas9, and two homology-directed repair templates. By delivering pKCcas9dO series editing plasmids into the model strain Streptomyces coelicolor M145, through one-step intergeneric transfer, we achieved the genome editing at different levels with high efficiencies of 60%-100%, including single gene deletion, such as actII-orf4, redD, and glnR, and single large-size gene cluster deletion, such as the antibiotic biosynthetic clusters of actinorhodin (ACT) (21.3 kb), undecylprodigiosin (RED) (31.6 kb), and Ca 2+ -dependent antibiotic (82.8 kb). Furthermore, we also realized simultaneous deletions of actII-orf4 and redD, and of the ACT and RED biosynthetic gene clusters with high efficiencies of 54% and 45%, respectively. Finally, we applied this system to introduce nucleotide point mutations into the rpsL gene, which conferred the mutants with resistance to streptomycin. Notably, using this system, the time required for one round of genome modification is reduced by one-third or one-half of those for conventional methods. These results clearly indicate that the established CRISPR/Cas9 genome editing system substantially improves the genome editing efficiency compared with the currently existing methods in Streptomyces, and it has promise for application to genome modification in other Actinomyces species.

We found that the biosynthesis of actinorhodin (Act), undecylprodigiosin (Red), and calcium-dependent antibiotic (CDA) are dramatically activated by introducing certain mutations into the rpoB gene that confer resistance to rifampin to Streptomyces lividans 66, which produces less or no antibiotics under normal growth conditions. Activation of Act and/or Red biosynthesis by inducing mutations in the rpoB gene was shown to be dependent on the mutation's position and the amino acid species substituted in the ␤-subunit of the RNA polymerase. Mutation analysis identified 15 different kinds of point mutations, which are located in region I, II, or III of the rpoB gene and, in addition, two novel mutations (deletion of nucleotides 1287 to 1289 and a double substitution at nucleotides 1309 and 1310) were also found. Western blot analyses and S1 mapping analyses demonstrated that the expression of actII-ORF4 and redD, which are pathway-specific regulatory genes for Act and Red, respectively, was activated in the mutants able to produce Act and Red. The ActIV-ORF1 protein (an enzyme for Act biosynthesis) and the RedD protein were produced just after the upregulation of ActII-ORF4 and RedZ, respectively. These results indicate that the mutation in the rpoB gene of S. lividans, resulting in the activation of Act and/or Red biosynthesis, functions at the transcription level by activating directly or indirectly the key regulatory genes, actII-ORF4 and redD. We propose that the mutated RNA polymerase may function by mimicking the ppGpp-bound form in activating the onset of secondary metabolism in Streptomyces.

We developed a novel approach for improving the production of antibiotic from Streptomyces coelicolor A3(2) by inducing combined drug-resistant mutations. Mutants with enhanced (1.6-to 3-fold-higher) actinorhodin production were detected at a high frequency (5 to 10%) among isolates resistant to streptomycin (Str r ), gentamicin (Gen r ), or rifampin (Rif r ), which developed spontaneously on agar plates which contained one of the three drugs. Construction of double mutants (str gen and str rif) by introducing gentamicin or rifampin resistance into an str mutant resulted in further increased (1.7-to 2.5-fold-higher) actinorhodin productivity. Likewise, triple mutants (str gen rif) thus constructed were found to have an even greater ability for producing the antibiotic, eventually generating a mutant able to produce 48 times more actinorhodin than the wild-type strain. Analysis of str mutants revealed that a point mutation occurred within the rpsL gene, which encodes the ribosomal protein S12. rif mutants were found to have a point mutation in the rpoB gene, which encodes the ␤-subunit of RNA polymerase. Mutation points in gen mutants still remain unknown. These single, double, and triple mutants displayed in hierarchical order a remarkable increase in the production of ActII-ORF4, a pathway-specific regulatory protein, as determined by Western blotting analysis. This reflects the same hierarchical order observed for the increase in actinorhodin production. The superior ability of the triple mutants was demonstrated by physiological analyses under various cultural conditions. We conclude that by inducing combined drug-resistant mutations we can continuously increase the production of antibiotic in a stepwise manner. This new breeding approach could be especially effective for initially improving the production of antibiotics from wild-type strains.

Learning joint embedding space for various modalities is of vital importance for multimodal fusion. Mainstream modality fusion approaches fail to achieve this goal, leaving a modality gap which heavily affects cross-modal fusion. In this paper, we propose a novel adversarial encoder-decoder-classifier framework to learn a modality-invariant embedding space. Since the distributions of various modalities vary in nature, to reduce the modality gap, we translate the distributions of source modalities into that of target modality via their respective encoders using adversarial training. Furthermore, we exert additional constraints on embedding space by introducing reconstruction loss and classification loss. Then we fuse the encoded representations using hierarchical graph neural network which explicitly explores unimodal, bimodal and trimodal interactions in multi-stage. Our method achieves state-of-the-art performance on multiple datasets. Visualization of the learned embeddings suggests that the joint embedding space learned by our method is discriminative.

Working with a Streptomyces albus strain that had previously been bred to produce industrial amounts (10 mg/ml) of salinomycin, we demonstrated the efficacy of introducing drug resistance-producing mutations for further strain improvement. Mutants with enhanced salinomycin production were detected at a high incidence (7 to 12%) among spontaneous isolates resistant to streptomycin (Str r ), gentamicin, or rifampin (Rif r ). Finally, we successfully demonstrated improvement of the salinomycin productivity of the industrial strain by 2.3-fold by introducing a triple mutation. The Str r mutant was shown to have a point mutation within the rpsL gene (encoding ribosomal protein S12). Likewise, the Rif r mutant possessed a mutation in the rpoB gene (encoding the RNA polymerase ␤ subunit). Increased productivity of salinomycin in the Str r mutant (containing the K88R mutation in the S12 protein) may be a result of an aberrant protein synthesis mechanism. This aberration may manifest itself as enhanced translation activity in stationary-phase cells, as we have observed with the poly(U)-directed cell-free translation system. The K88R mutant ribosome was characterized by increased 70S complex stability in low Mg 2؉ concentrations. We conclude that this aberrant protein synthesis ability in the Str r mutant, which is a result of increased stability of the 70S complex, is responsible for the remarkable salinomycin production enhancement obtained.Improvement of the productivity of commercially viable microbiotic strains is an important field in microbiology, especially since wild-type strains isolated from nature usually produce only a low level (1ϳ100 g/ml) of antibiotics. A great deal of effort and resources is therefore committed to improving antibiotic-producing strains to meet commercial requirements. Current methods of improving the productivity of industrial microorganisms range from the classical random approach to using highly rational methods, for example metabolic engineering. Although classical methods are still effective even without using genomic information or genetic tools to obtain highly productive strains, these methods are always time-and resource-intensive (27,31).Members of the genus Streptomyces produce a wide variety of secondary metabolites that include about half of the known microbial antibiotics. We reported previously that a certain str mutation that confers streptomycin resistance is able to give rise to secondary metabolite production in Streptomyces lividans and Streptomyces coelicolor A3(2) (3,20,24). Later, we demonstrated that the introduction of a specific str mutation, as well as a gentamicin resistance-producing mutation (gen), into other bacterial genera gave rise to a marked antibiotic productivity increase, thus further elucidating the antibiotic production mechanism in S. coelicolor A3(2) (6). Furthermore, by inducing a mutation that confers resistance to rifampin, we were able to restore the impaired production of antibiotics in the relA and relC mutants of S. coelicolor A3(2) (8,11,...