Background & Summary

With the rapid emergence of antibiotic microbial resistance (AMR) to all major classes of antibiotics and the decline in number of potential candidates for new antibiotics, there is a pressing need for the discovery of novel antibacterial compounds1. Streptomyces, soil dwelling gram-positive bacteria, continue to be promising microorganisms for the production of clinically important secondary metabolites, including not only antibiotics, but also antiviral, antifungal, and antiparasitic agents, and antitumorals and immunosuppressant compounds2. Streptomyces are distinguished by their complex life cycle and high G + C content (often over 70%) in their linear genomes3,4. Traditionally, drug discovery from Streptomyces has been based on bioactivity screening followed by mass spectrometry and NMR-based molecular identification5. However, recent advances in genomics-based approaches revealed that most of the secondary metabolite biosynthetic gene clusters (smBGCs) of streptomycetes are inactive under laboratory conditions, suggesting that the ability of streptomycetes to produce secondary metabolites has been under-estimated5,6. Each Streptomyces species has the genetic potential to produce more than 30 secondary metabolites on average, which are diverse and differ between species7,8. Considering Streptomyces is the largest genus of actinobacteria with approximately 900 species characterized so far, streptomycetes are a valuable resource for the discovery of novel secondary metabolites9.

SmBGCs, especially polyketide and non-ribosomal peptide synthetase types, are often composed of extraordinarily long genes (>5 kb) encoding multi-modular enzymes with repetitive domain structures. Therefore, accurate gene annotations based on high quality genome sequences are essential for the precise identification of smBGCs10. Gene annotation with the high quality genome of S. clavuligerus revealed that 30% out of a total of 7,163 protein coding genes were incorrectly annotated in the previous draft genome of S. clavuligerus containing ambiguous and inaccurate nucleotides, indicating the importance of high quality genome sequences11. In addition, high quality genome sequences are essential for multi-omics analysis, which facilitates the understanding of the complex regulation on smBGCs and rational engineering for increasing secondary metabolites production11,12.

Among the 1,614 streptomycetes genomes that have been deposited in the NCBI Assembly database to date (as of 9th December 2019), only 189 and 35 assemblies were designated as complete genome level and chromosome level, respectively. More than 86% of assemblies were draft-quality genome sequences, which contain fragmented multiple contigs or ambiguous sequences4,13,14,15. One of the main obstacles to obtaining high quality genomic information of streptomycetes is the low fidelity of sequencing techniques when dealing with high G w C genomes and frequently repetitive sequences such as terminal inverted repeats13. In addition, since streptomycetes have linear chromosome, it is difficult to confirm the completeness of the assembled chromosome.

In this study, we present the high-quality genome sequences of 30 streptomycetes, increasing the total number of reported complete Streptomyces genome by about 10%. The target streptomycetes were 22 Streptomyces type strains and eight different Streptomyces venezuelae strains, most of which are currently used as industrial strains for producing various bioactive compounds. We applied hybrid assembly strategy with long-read (PacBio) and short-read (Illumina) sequencing techniques to obtain complete genome sequences. PacBio sequencing provides long reads of several kb in length which allows the readthrough of regions with low complexity, enabling the assembly of repetitive regions, which are difficult to assemble by using Illumina sequencing reads, even with the high coverage data16. However, Illumina sequencing provides reads with a lower error rate compared to the PacBio sequencing, and assembled contigs based on the Illumina sequencing reads are not simply a subset of the contigs from PacBio sequencing reads13,17. Therefore, reconciling PacBio and Illumina sequencing methods enables one to generate more complete genomes by overcoming the shortcomings of each method. During the genome assembly using reads from PacBio (0.46~5.18 Gbp) and Illumina (0.5~3.0 Gbp) sequencing, we constructed 6.7 to 10.1 Mbp of streptomycetes genomes, most of which consist of single chromosomes with 72% G + C contents on average. Inaccurate sequences in the assembled genome were corrected using Illumina sequencing reads. The complete streptomycetes genomes have more than 97.4% gene space completeness and on average 7,000 protein coding genes, 20 rRNAs, and 68 tRNAs were annotated. Finally, based on the complete genome sequences and annotations, we predicted a total of 922 smBGCs. The complete genome sequences and newly determined smBGCs in this study should prove to be a fundamental resource for understanding the genetic basis of streptomycetes and for discovering novel secondary metabolites.

Methods

Genomic DNA (gDNA) extraction

Total 30 streptomycetes were purchased from Korean Collection for Type Cultures (KCTC, Korea). A stock of streptomycetes were inoculated to 50 mL of liquid culture medium with 0.16 g mL−1 of glass beads (3 ± 0.3 mm diameter) in 250 mL baffled flask and grown at 30 °C in a 200 rpm orbital shaker. Each streptomycetes was grown in one of four different culture medium, R5(–) medium (25 mM TES (pH 7.2), 103 g L−1 sucrose, 1% (w/v) glucose, 5 g L−1 yeast extract, 10.12 g L−1 MgCl2∙6H2O, 0.25 g L−1 K2SO4, 0.1 g L−1 casamino acids, 0.08 g L−1 ZnCl2, 0.4 mg L−1 FeCl3, 0.02 mg L−1 CuCl2∙2H2O, 0.02 mg L−1 MnCl2∙4H2O, 0.02 mg L−1 Na2B4O7∙10H2O, and 0.02 mg L−1 (NH4)6Mo7O24∙4H2O), 1 × sporulation medium (3.33 g L−1 glucose, 1 g L−1 yeast extract, 1 g L−1 beef extract, 2 g L−1 tryptose, and 0.006 g L−1 FeSO4∙7H2O), YEME medium (340 g L−1 sucrose, 10 g L−1 glucose, 3 g L−1 yeast extract, 5 g L−1 bacto peptone, and 3 g L−1 oxoid malt extract), and MYM medium (4 g L−1 maltose, 4 g L−1 yeast extract, 10 g L−1 malt extract). For gDNA extraction, 25 mL cultured cells were harvested at the exponential growth phase and washed twice with same volume of 10 mM EDTA, followed by the lysozyme (10 mg mL−1) treatment at 37 °C for 45 min. gDNA was extracted using a Wizard Genomic DNA Purification Kit (Promega, Madison, WI, USA) according to the manufacturer’s instruction. Quality and quantity of extracted gDNA samples were evaluated using 1% agarose gel electrophoresis and Nanodrop (Thermo Fisher Scientific, Waltham, MA, USA), respectively.

Short-read (Illumina) genome sequencing

For construction of short-read genome sequencing library, 2.5 μg of gDNA was sheared to approximately 350 bp by a Covaris instrument (Covaris Inc., Woburn, MA, USA) with the following conditions; Power 175, Duty factor 20%, C. burst 200, Time 23 s, 8 times. The library was constructed using a TruSeq DNA PCR-Free LT kit (Illumina Inc., San Diego, CA, USA) following manufacturer’s instruction. Briefly, the fragmented DNA samples were cleaned and end-repaired, followed by the adaptor ligation and bead-based size selection ranging from 400 to 500 bp. Quantity of final libraries was measured using Qubit® dsDNA HS Assay Kit (Thermo Fisher Scientific) and the library size was determined using Agilent 2200 Bioanalyzer (Agilent Technologies, Santa Clara, CA, USA). Among the constructed sequencing libraries, 29 libraries were sequenced with the HiSeq. 2500 (Illumina Inc.) as 100 bp single-end reads and remaining one library for S. tsukubaensis was sequenced with the Miseq v.2 (Illumina Inc.) with 50 bp single-read recipe. Finally, 0.46 to 5.18 Gbp of raw sequence data were obtained and the read qualities were examined by creating sequencing QC reports function of CLC genomic workbench version 6.5.1 (CLC bio, Denmark) (Online-only Table 1 and Fig. 1a).

Fig. 1
figure 1

Quality of the genome sequencing data. (a) Distribution of Illumina reads quality based on Phred score. (b) Read quality distribution of PacBio reads. Black line indicates total number of bases in the reads which have greater read quality than the corresponding read quality value on x-axis.

Full size image

Long-read (PacBio) genome sequencing

A total of 5 μg gDNA was used as input for PacBio genome sequencing library preparation. The sequencing library was constructed with the PacBio SMRTbellTM Template Prep Kit (Pacific Biosciences, Menlo Park, CA, USA) following manufacturer’s instructions. Fragments smaller than 20 kbp were removed using the Blue Pippin Size selection system (Sage Science, Beverly, MA, USA) and the constructed libraries were validated using Agilent 2100 Bioanalyzer (Agilent Technologies). Final SMRTbell libraries were sequenced using one or two SMRT cells with P6-C4-chemistry (DNA Sequencing Reagent 4.0) on the PacBio RS II sequencing platform (Pacific Biosciences). Approximately, 0.5 to 3.0 Gbp of raw sequence data were generated (Online-only Table 1).

Genome assembly

Among the raw PacBio sequencing reads, only the reads with a read quality value greater than 0.75 and a length longer than 50 bp were filtered (Fig. 1b). Post filtered reads were assembled by the hierarchical genome assembly process workflow (HGAP, Version 2.3), including consensus polishing with Quiver18. For each assembled contig, error correction was performed based on their estimated genome size and average coverage. Raw reads from the Illumina sequencing were quality trimmed using CLC genomic workbench version 6.5.1 (ambiguous limit 2 and quality limit 0.05) and assembled using de novo assembly function of CLC genomic workbench version 6.5.1 with default parameters. To expand the assembled contigs, all of assembled PacBio and Illumina contigs were aligned using MAUVE 2.4.019 and linked using GAP5 program (Staden package)20.

Genome correction

Quality trimmed Illumina sequencing reads were mapped to the assembled genome using CLC genomic workbench version 6.5.1 (mismatch cost 2, insertion cost 3, deletion cost 3, length fraction 0.9, and similarity fraction 0.9). Conflicts showing more than 80% frequency for Illumina reads were corrected as Illumina sequence (Table 1). In addition, percentage of mapped Illumina reads on to the assembled genome represents degree of completeness (Table 1 and Fig. 2b). Completeness of gene space was estimated using the BUSCO v3 (Table 2)21.

Table 1 The statistics of genome assembly and correction.
Full size table
Fig. 2
figure 2

Genome assembly of 30 streptomycetes. (a) Strategy for genome assembly and corrections. (b) Profile of Illumina reads mapped on assembled genomes. Data were visualized using SignalMap (Roche NimbleGen, Inc.). Red line indicates the average Illumina read coverage of all genomic positions.

Full size image
Table 2 Gene space completeness of completed genomes.
Full size table

Genome annotation and secondary metabolite biosynthetic gene cluster prediction

The complete genome sequences of streptomycetes were submitted to the NCBI GenBank database and annotated by the latest updated version of NCBI Prokaryotic Genome Annotation Pipeline (PGAP)22. Using the GenBank formatted files of each genomes as input, secondary metabolite biosynthetic gene clusters were predicted by antiSMASH 4.023.

Data Records

Raw reads from short-read (Illumina) and long-read (PacBio) sequencing were deposited in the NCBI Sequence Read Archive (SRA) (Online-only Table 1)24,25. 30 complete genome sequences were deposited in GenBank via the NCBI’s submission portal (Table 3)26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55. Detailed information on the predicted 922 smBGCs in 30 streptomycetes genomes has been deposited in FigShare56.

Table 3 Summary of genome annotation.
Full size table

Technical Validation

Streptomyces have drawn considerable attention because of their ability to produce various clinically important secondary metabolites. Total 30 streptomycetes genomes were sequenced by using both PacBio and Illumina sequencing methods to elucidate their biosynthetic potential. After cleaning the reads, on average 98,380 PacBio reads with 11,725 bp length and 18,223,235 Illumina reads with 100 bp length (50 bp for S. tsukubaensis) were generated (Fig. 1a,b and Online-only Table 1). Through the assembly of reads from two sequencing platforms using HGAP, CLC workbench, MAUVE, and GAP5 programs, single linear scaffolds ranging from 6.7 to 10.1 Mbp in length with 72% G + C contents were obtained for 27 streptomycetes, whereas two scaffolds were finally constructed for three remaining streptomycetes, S. clavuligerus (6.7 and 1.8 Mbp), S. albofaciens (4.8 and 4.5 Mbp), and S. filamentosus (5.7 and 2.1 Mbp) (Table 1). S. clavuligerus has been reported to have a large linear plasmid with a length of 1.8 Mbp, so the genome was correctly assembled into a single chromosome, while the S. albofaciens and S. filamentosus genomes appear to be assembled into two divided scaffolds11,57. To increase the accuracy of the assembled genome sequences, Illumina sequences showing more than 80% coverage at the conflict sites were taken as the corrected ones (Table 1). Approximately, 96.32% of Illumina sequencing reads were successfully mapped to the corresponding genomes (Table 1 and Fig. 2b). The completeness of the genomes were assessed using the BUSCO approach with a total of 352 orthologue groups from the Actinobacteria Dataset21. Results showed that 29 genomes have more than 99.1% gene space completeness and the S. clavuligerus genome has 97.4% gene space completeness (Table 2). Following NCBI PGAP, 30 genomes were annotated with 7,000 protein coding genes, 20 rRNAs, and 68 tRNAs on average (Table 3). Finally, based on the annotation, a total of 922 smBGCs were predicted in 30 streptomycetes genomes (Fig. 3). Detailed information, such as genomic positions, types, and putative products of each smBGC are publicly available in Figshare56.

Fig. 3
figure 3

Secondary metabolite biosynthetic gene clusters in 30 complete streptomycetes genomes.

Full size image