Genome mining of Streptomyces sp. BRB081 reveals the production of the antitumor pyrrolobenzodiazepine sibiromycin

Abstract

Employing a genome mining approach, this work aimed to further explore the secondary metabolism associated genes of Streptomyces sp. BRB081, a marine isolate. The genomic DNA of BRB081 was sequenced and assembled in a synteny-based pipeline for biosynthetic gene clusters (BGCs) annotation. A total of 27 BGCs were annotated, including a sibiromycin complete cluster, a bioactive compound with potent antitumor activity. The production of sibiromycin, a pyrrolobenzodiazepine, was confirmed by the analysis of obtained BRB081 extract by HPLC–MS/MS, which showed the presence of the sibiromycin ions themselves, as well as its imine and methoxylated forms. To verify the presence of this cluster in other genomes available in public databases, a genome neighborhood network (GNN) was constructed with the non-ribosomal peptide synthetase (NRPS) gene from Streptomyces sp. BRB081. Although the literature does not report the occurrence of the sibiromycin BGC in any other microorganism than Streptosporangium sibiricum, we have located this BGC in 10 other genomes besides the BRB081 isolate, all of them belonging to the Actinomycetia class. These findings strengthen the importance of uninterrupted research for new producer strains of secondary metabolites with uncommon biological activities. These results reinforced the accuracy and robustness of genomics in the screening of natural products. Furthermore, the unprecedented nature of this discovery confirms the unknown metabolic potential of the Actinobacteria phylum and the importance of continuing screening studies in this taxon.

Supplementary Information

The online version contains supplementary material available at 10.1007/s13205-022-03305-0.

Introduction

For almost a century, the bacterial genus Streptomyces has been known as a prolific source of metabolites with high biotechnological value. Alone, this taxon is responsible for more than half of all antibiotics in clinical use and for other relevant bioactivities, such as immunomodulators, antitumors, antifungals, antiparasitics, bioinsecticides, and bioherbicides (Zhao et al. 2019; Pham et al. 2019).

Although the study of terrestrial streptomycetes became widespread during the Golden Age of antibiotic discovery (Lewis 2013), many new natural products continue to be described from this genus, especially from marine isolates. As the sampling of marine microbiomes was only better established around the 1960s (Cragg and Newman 2000), this niche is still considered under-explored and has attracted numerous screening investigations.

Recently, Tangerina et al. (2020) reported the cytotoxic potential of the crude extract of Streptomyces sp. BRB081, a marine isolate, against the colon adenocarcinoma cell line (HCT-116 ATCC CCL-247). The observed cytotoxicity was attributed to the presence of surugamides, whose antitumor activity was described (Takada et al. 2013), but also to a profusion of other molecules that have not been identified through metabolomics.

This question led us to further study the metabolic potential of Streptomyces sp. BRB081 in the spotlight of genomics. Interestingly, this approach identified a biosynthetic gene cluster (BGC) of sibiromycin in this genome. Sibiromycin is a potent antitumor antibiotic isolated from Streptosporangium sibiricum (Gause et al. 1969), and its production had never been reported from the genus Streptomyces or from another taxon.

Here, the BGC of sibiromycin from Streptomyces sp. BRB081 is described, and the biosynthesis of this metabolite is chemically confirmed. In addition, we present the draft genome of this strain and corroborate its capability to produce several unknown molecules. Finally, our data reinforce the prominence of genomics in screening natural products, since its accuracy is independent of gene expression, cultivation, and extraction conditions.

Methods

Bacterial strain and culture condition

The bacterial strain Streptomyces sp. BRB081 was recovered from sediments collected at Araçá Beach, São Sebastião, SP, Brazil (23°48′53.83″S; 45°24′26.83″W). The protocols used for their isolation and identification were described by Tangerina et al. (2020). To increase cell mass, 2.5 mL of mycelia was inoculated in 250 mL Erlenmeyer flasks containing 22.5 mL of A1 liquid medium (10 g soluble starch, 4 g yeast extract, 2 g peptone, 1L Red Sea® artificial seawater) and 25 glass beads, used as an inert support. The culture was cultivated at 28 °C, with automatic stirring at 180 rpm for 48 h.

Genome sequencing, assembly, and annotation

The Streptomyces sp BRB081 genomic DNA was extracted with Wizard® Genomic DNA Purification Kit (Promega Corporation, Fitchburg, USA), and further was sequenced with the HiSeq System platform (Illumina Inc., San Diego, USA) at Macrogen laboratory (Seoul, South Korea).

The assembly of genomic DNA sequence was made with Geneious R11 software (Biomatters, New Zealand), and subsequently the contigs were merged on scaffolds based on synteny conservation using MeDuSa server v. 1.6 (Bosi et al. 2015). Finally, the completeness of this assembly was checked with gVolante 2.0.0 software (Nishimura et al. 2017), employing the BUSCO V5 pipeline (Simão et al. 2015). The annotation of Streptomyces sp. BRB081 genome was performed by the NCBI Prokaryotic Genome Annotation Pipeline (PGAP) (Tatusova et al. 2016; Haft et al. 2018).

The software antiSMASH 5.0 bacterial version was utilized to predict biosynthetic gene clusters of secondary metabolites (BGCs). The chosen parameters were “detection strictness” and threshold set to “strict” (detects well-defined clusters containing all required parts) (Blin et al. 2019).

Phenetic analysis

To identify genomes closely related to Streptomyces sp. BRB081, a Multilocus Sequence Analysis (MLSA) was performed based on the 16S rRNA, rpoB, gyrB, atpD, and recA concatenated genes. Through the software MEGA version X (Kumar et al. 2018), the MUSCLE (Edgar 2004) alignment algorithm was executed, and a phenetic tree was created using the Maximum Likelihood method and Jukes–Cantor model (Jukes and Cantor 1969) with 1000 bootstrap replications. A matrix of genetic distance pairwise with the Kimura 2-parameter model (Kimura 1980), was constructed using MEGA.

Generation of SSN and GNN

The generation of the sequence similarity network (SSN) and the genome neighborhood network (GNN) were achieved using the tools available on EFI—Enzyme Function Initiative Website (Gerlt et al. 2015; Gerit 2017; Zallot et al. 2018, 2019).

The SSNs were constructed using the Sequence BLAST Tab from EFI, in this option a BLAST of the MBL3808355.1 FASTA sequence was made against the UniProt database (https://www.uniprot.org/) to select a database of 1000 homolog proteins with Blast e-value lower than 1 × 10⁻⁵. In the SSN, each member of a family or group is called a node (rectangles) that can relate to other nodes by edges (lines). What defines the presence of an edge between two nodes is a threshold similarity measure (BLAST bit Score). The threshold for the first SSN was about 30% of protein identity (Score = 205), and for the second SSN, the threshold score was 380.

To generate a GNN employing the Genome Neighbor Tools from EFI, the input data was a SSN file that contains the set of studied proteins clustered. One output of GNT is the GNDs or Genomic neighborhood Diagrams (diagrams where the genomic neighborhood of each protein is plotted). In this analysis, it was opted to retrieve 20 adjoining genes (Neighborhood Size = 20) by SSN node.

LC–MS/MS analysis

Streptomyces sp. BRB081 was cultivated for 7 days in Erlenmeyer flasks (125 mL) containing 40 mL of liquid starch casein medium (soluble starch, 10.0 g; sodium chloride, 2.0 g; casein, 0.3 g; potassium nitrate, 2.0 g; dipotassium hydrogen phosphate, 2.0 g; magnesium sulfate heptahydrate, 0.05 g; calcium carbonate, 0.02 g; iron (II) sulfate heptahydrate, 0.01 g; synthetic sea salt (Red Sea®), 36.0 g; deionized water, 1.0 q.s./L.) at 28 °C and 180 rpm (Tangerina et al. 2020). The final pH was adjusted between 7.8 and 8.2 using hydrochloric acid 1.0 mol L⁻¹ or sodium hydroxide 1.0 mol L⁻¹. The final broth was extracted twice using 20 mL of ethyl acetate at 190 rpm for 30 min. Then, the extract was dried in an Eppendorf® Concentrator Plus system under vacuum, and the resulting material was diluted to 1.0 mg L⁻¹ using methanol HPLC grade.

Liquid chromatography–mass spectrometry analysis was carried out on an HPLC (Shimadzu®) coupled to a mass spectrometer ESI IT (Amazon SL, Bruker Daltonics), fitted with an electrospray ionization source and ion trap analyzer. The chromatographic method consisted of solvent A (0.1% formic acid in water) and B (0.1% formic acid in methanol), starting at 5% up to 100% of B in 30 min followed by a hold of 100% of B for 5 min, using a C18 column (Phenomenex® Luna, 5 μm, 4.6 × 250 mm). The method employed a flow of 1.0 mL min⁻¹, 15 μL injection volume, and column temperature of 40 °C. The mass spectrometer was operated in positive mode, monitoring a mass range from 100 to 1500 atomic mass units (amu), a capillary voltage of 3500 V, end plate offset of 500 V, nebulizer at 60 psi, dry gas 10 L min⁻¹ and dry temperature at 320 °C, using the untargeted mode (fragmentation at MS² level using a ramp of collision energy from 50 up to 75 eV).

Results

Genome draft of Streptomyces sp BRB081

The Illumina sequencing system was chosen to perform the whole genome sequencing (WGS) of Streptomyces sp. BRB081, the assembly of the obtained data was made first with Geneious R11 generating 70 contigs, these contigs were organized in scaffolds via MeDuSa server V1.6 using the principle of synteny conservation. Table 1 summarizes information about the Streptomyces sp. BRB081 WGS project.

Table 1.

General features of Streptomyces sp. BRB081, according to Minimal Information about any (x) Sequence (MIxS) standard checklist (Yilmaz et al. 2011)

Item	Description
Taxonomy	Domain bacteria
	Phylum Actinobacteria
	Class Actinomycetia
	Order Streptomycetales
	Family Streptomycetaceae
	Genus Streptomyces
	Specie unclassified
	Strain BRB081
Submitted to insdc	JACVQE000000000.2 (GenBank)
Investigation type	Bacteria
Project name	Streptomyces sp. BRB081, whole genome shotgun sequencing project
Bioproject	PRJNA659729
Biosample	SAMN15933637
Geographic location	Araçá Beach, São Sebastião, SP, Brazil
Latitude and longitude	23°48′53.83″S; 45°24′26.83″O
Collection date	2014-11-15
Isolation source	Sea
Isolation material	Sediments
Relationship to oxygen	Aerobic
Ploidy	Haploid
Growth condition	A1 liquid medium
Sequencing method	Illumina HiSeq
Assembly	Geneious v. R11.1; MeDuSa v. 1.6

Since the synteny-guided assembly (Bosi et al. 2015) needs related genomes, as “Comparison genomes” we provided the complete sequences of Streptomyces rutgersensis NBH77 (GCF_014216335.1), Streptomyces albidoflavus J1074 (GCF_000359525.1), and Streptomyces koyangensis VK-A60T (GCF_003428925.1), available at National Center for Biotechnology Information (NCBI, U.S. National Library of Medicine). The method used to select these three sequences as a reference for our assembly is described in section “Phenetic analysis”.

The use of these genomes as a reference for mapping the 70 contigs of Streptomyces sp. BRB081 resulted in the formation of just 5 scaffolds. The completeness of the genome assembly (Table 2) was verified with gVolante 2.0.0 software (Nishimura et al. 2017), employing the BUSCO V5 pipeline with Streptomycetales ortholog set (Simão et al. 2015). In this analysis, besides metrics based on sequence length, mainly the coverage of pre-selected genes is important to consider the proper assembly of sequencing reads. In this case, 99.18% of core genes from Streptomycetales order are present in both assembly levels carried out.

Table 2.

Completeness test of Streptomyces sp. BRB081 assemblies

Metric	Contig assembly level	Scaffold assembly level
Number of sequences	70	5
Total length (nt)	6,934,851	6,938,651
Longest sequence	609,359	6,853,601
Total # of core genes queried	1579	1579
Complete orthologs	1566 (99.18%)	1566 (99.18%)
Missing orthologs	12 (0.76%)	12 (0.76%)
Partial orthologs	1	1
Duplicated orthologs	5	5
N50 sequence length (nt)	307,015	6,853,601
L50 sequence count	8	1

The draft genome of this marine isolate has a linear chromosome of approximately 7 Mb, like the small genomes of the Streptomyces genus, which comprise species ranging from 6 to 12 Mb genome size (Hoff et al. 2018). Its 73.43% GC content is one of the highest known in Streptomyces, like other strains such as S. albidoflavus NRRL-1271 (Zaburannyi et al. 2014) and S. sampsonii KJ40 (Labeda et al. 2014). The functional annotation discovered 5605 CDSs and 8 copies of rRNAs (Table 3), as comparison in related strain of similar genome size S. albidoflavus NRRL-1271 which has 5832 CDS and 7 rRNAs copies (Zaburannyi et al. 2014). Information of Streptomyces sp. BRB081 whole genome shotgun sequencing project is available in the GenBank under the accession JACVQE000000000.2.

Table 3.

Functional features annotated in genome assembly of Streptomyces sp. BRB081

Genes	CDSs (coding protein)	Pseudogenes	rRNAs	tRNAs	ncRNAs
5940	5605	257	8	67	3

Phenetic analysis

To identify genomes closely related to Streptomyces sp. BRB081, a Multilocus Sequence Analysis (MLSA) was performed based on the 16S rRNA, rpoB, gyrB, atpD, and recA concatenated genes. Through the software MEGA version X (Kumar et al. 2018), the MUSCLE (Edgar 2004) alignment algorithm was executed, and a phenetic tree was created using the Maximum Likelihood method and Jukes–Cantor model (Jukes and Cantor 1969) with 1000 bootstrap replications. The similarity of the sequences compared is shown in the phenetic tree (Fig. 1).

Fig. 1.

Phenetic tree of BRB081. The accession number of the compared sequences is given in parentheses after the strain identification. The taxa were compared using MLSA. Results were inferred by the Maximum Likelihood method and Jukes–Cantor model (Jukes and Cantor 1969), with 1000 bootstrap replications

This analysis not only identified S. rutgersensis NBH77 as the most closely related sequence to isolate BRB081, but also provided strong evidence that these strains may belong to the same taxonomic specie. The matrix of genetic distance pairwise (Table 4) made with the Kimura 2-parameter model (Kimura 1980) shows values of evolutionary distance pairwise between BRB081 and NBH77 much lower than 0.007 (DNA–DNA hybridization of 70%), recognized as the cut-off point for strains classification as representatives of distinct Streptomyces species (Rong and Huang 2012).

Table 4.

MLSA evolutionary distance pairwise using Kimura 2-parameter model

	Strain	1	2	3	4	5	6	7
1	Streptomyces sp. BRB081
2	S. rutgersensis NBH77	0.000519
3	S. koyangensis VK-A60T	0.019801	0.019698
4	S. sampsonii KJ40	0.020761	0.020658	0.006360
5	S. coelicolor A3(2)	0.079927	0.080060	0.081447	0.080394
6	S. albidoflavus J1074	0.021082	0.020979	0.006255	0.001975	0.080389
7	M. aurantiaca ATCC 27029	0.198018	0.198206	0.200539	0.200542	0.204267	0.200675

The number of base substitutions per site from between sequences is shown. Analyses were conducted using the Kimura 2-parameter model (Kimura 1980). The value in italics corresponds to the pair Streptomyces sp. BRB081–S. rutgersensis NBH77

Other evidence is the taxonomy check data provided by NCBI about the genome draft of BRB081. The best-matching type strain is Streptomyces gougerotii (GCA_014648955.1), with an average nucleotide identity (ANI) of 99.53%. As suggested by Richter and Rosselló-Mora (2009), ANI values between strains of distinct species are lower than 95–96%. Along with S. diastaticus subsp. diastaticus, S. rutgersensis, and S. gougerotii form a single clade in the phylogenetic tree. Furthermore, the nucleotide sequences of its 16S rRNA are 100% identical, and its DNA–DNA relatedness ranges from 95.8 to 97.2%. Supported by phenotypic data, this indicates that the three genomes, described as independent species, are heterotypic synonyms (Komaki and Tamura 2020).

The NBH77 strain of S. rutgersensis was isolated from soil (Benaud et al. 2021) as well as many other strains of this specie; however, bacteria isolated from marine sediments are often related to terrestrial counterparts (Almeida et al. 2019), and they can present different metabolic and physiological capabilities (Almeida et al. 2019; Paderog et al. 2020).

Prediction of secondary metabolite biosynthetic gene clusters

The AntiSMASH 5.0 identified 27 biosynthetic gene clusters (BGCs) (Table S1, Supplementary Data). The most abundant types of BGC encodes for non-ribosomal peptide synthetases (NRPS) (8/27), polyketide synthases (PKS) (5/27) and hybrid NRPS/PKS (2/27). Their products often exhibit biological activity with clinical relevance and are biosynthesized in a wide variety by marine Streptomyces (Katz and Baltz 2016). Putative BGCs associated with the production of terpenes, siderophores, bacteriocins, lanthipeptides, ectoin, and aryl polyene were also found, in a manner compatible with that already reported for marine members of this genus (Williams 2013; Sung et al. 2017; Jackson et al. 2018; Xu et al. 2019).

An important piece of information contained in these data is the identity (ID) between the gene composition of putative BGCs with experimentally characterized BGCs. These values generally allow us to infer with some certainty the end products of the predicted routes. Almost half of BRB081 clusters were weakly related to known BGCs, with IDs < 16%. Corroborating this result, a profusion of unclassified molecules was identified in the culture extract of this bacteria (Tangerina et al. 2020).

In contrast, other putative BRB081 BGCs contained a high identity of gene composition with BGCs available in the MIBiG database—Minimum Information about a Biosynthetic Gene cluster (Kautsar et al. 2020). This indicates that BRB081 has the potential to biosynthesize already known molecules. This is the case of geosmin, a terpene that gives the typical wet earth odor to Streptomyces cultures (Jiang et al. 2007), and ectoin, a compatible solute accumulated by divers Streptomyces, which plays the role of protecting the cell against the effects of high salinity typical of marine environments (Bursy et al. 2008). Desferrioxamines (siderophore) and surugamides (antitumor), are also surely produced and were reported by Tangerina et al. (2020).

The unforeseen observation was the annotation of a putative sibiromycin BGC. Sibiromycin is a pyrrolo benzodiazepine (PBD) (Almeida et al. 2019a; Blin et al. 2019), and its cytotoxicity was first reported five decades ago (Gause et al. 1969; Gause and Dudnik 1971). It is the most potent antitumoral ever described in its chemical class, with IC₅₀ reaching 17 pM against the murine plasmacytoma cell line (Thurston et al. 1999). Given its intense biological activity, it could be one of the compounds responsible for the cytotoxicity that Tangerina et al. (2020) reported. However, sibiromycin was not detected in the cited work, with the use of the metabolomic approach. This is the major reason for the relevance of this result given that sibiromycin BGC was only identified in the terrestrial actinobacteria S. sibiricum (Li et al. 2009).

Genomic evidence for the sibiromycin biosynthesis

Sibiromycin is a secondary metabolite member of PBDs, selective DNA alkylating agents commonly biosynthesized by several Actinobacteria members (Hurley et al. 1977; Hurley and Thurston 1984). This compound is one of the two known glycosylated PBDs, containing sibirosamine, a rare amino sugar responsible for increasing significantly its DNA binding affinity (Brahnikova et al. 1972) (Fig. 2).

Fig. 2.

Chemical structure of sibiromycin consisting of a pyrrolo benzodiazepine ring containing a sibirosamine moiety (Almeida et al. 2019a; Blin et al. 2019)

Despite its intense biological activity, the use of this compound as a chemotherapy agent is limited by the cardiotoxic effects (Yonemoto et al. 2012). Nevertheless, in the last 40 years, several synthetic PBDs (monomers, dimers, and hybrids) have been produced and studied, with a more recent orientation towards the construction of antibody–drug conjugates (Kemp et al. 2017).

Near one of the chromosomal ends, the putative sibiromycin BGC in BRB081 has 92% of the gene content of the related cluster (Table 5), comprising the coding of all enzymatic functions necessary for the molecule biosynthesis.

Table 5.

Putative sibiromycin BGC of Streptomyces sp. BRB081

BRB081 ORFa	Locus tagb	Product length (AA)	Predicted function	S. sibiricum genec	Identitya
1	ICN76_RS28540	167	L-DOPA 2,3-dioxygenase	sibV	67%
2	ICN76_RS28545	335	Tyrosine hydroxylase	sibU	61%
3	ICN76_RS28550	295	F420-dependent reductase	sibT	69%
4	ICN76_RS28555	301	PhzF family phenazine biosynthesis isomerase	sibS	57%
5	ICN76_RS28560	304	Putative regulator	sibR	46%
6	ICN76_RS28565	418	Kinurenine hydrolase	sibQ	65%
7	ICN76_RS28570	278	Tryptophan 2,3-dioxygenase	sibP	73%
8	ICN76_RS28575	246	N-Methyltransferase	sibO	70%
9	ICN76_RS30520	398	dTDP-4-keto-6-deoxyglucose transaminase	sibN	74%
10	ICN76_RS28585	421	C-Methyltransferase	sibM	66%
11	ICN76_RS28590	355	C-Methyltransferase	sibL	78%
12	ICN76_RS28595	283	Esterase/arylformamidase	sibK	68%
13	ICN76_RS28600	194	dTDP-4-keto-6-deoxyglucose 3,5-epimerase	sibJ	57%
14	ICN76_RS28605	322	dTDP-glucose synthase	sibI	73%
15	ICN76_RS28610	393	Glycosyltransferase	sibH	74%
16	ICN76_RS28615	363	FAD-dependent monooxygenase	sibG	62%
17	ICN76_RS28620	774	UvrA excinuclease	sibF	80%
18	ICN76_RS28625	599	Non-Ribosomal Peptide Synthetase	sibE	68%
19	ICN76_RS28630	1528	Non-Ribosomal Peptide Synthetase	sibD	58%
20	ICN76_RS28635	450	Kinurenine 3-monooxygenase	sibC	61%
21	ICN76_RS28640	170	pyridoxamine 5′-phosphate oxidase	–	–
22	ICN76_RS28645	293	putative regulator	sibA	47%
23	ICN76_RS28650	334	dTDP-glucose-4,6-dehydratase	–	–
24	ICN76_RS28655	555	MFS transporter	–	–
25	ICN76_RS28660	356	Methyltransferase	sibZ	52%
26	ICN76_RS28665	576	y-Glutamyltransferase	sibY	55%
27	ICN76_RS28670	492	FAD-dependent oxidoreductase	sibW	67%

^aBlin et al. (2019)

^bBy the NCBI Prokaryotic Genome Annotation Pipeline (PGAP)

^cLi et al. (2009)

The molecule of sibiromycin is synthetized from L-tryptophan, which is converted to anthranilate moiety of the molecule; L-tyrosine, which is converted to dihydropyrrole moiety of molecule; and by L-methionine which contributes as a methyl donor group (Gerratana, 2012). As demonstrated by Li et al. (2009), the respective products of the sibP, sibK, sibC, sibQ, sibL, and sibG genes act sequentially in the formation of 3,5-hydroxy-4-methylanthranilic acid from L-tryptophan, in BRB081 the corresponding genes are ORFs 7, 12, 20, 6, 11 and 16 (Fig S1a); while the products of the sibU, sibV, sibS, sibZ, sibY, sibT, and sibW genes, sequentially convert L-tyrosine in 4-propenyl-2,3-dihydropyrrole-2-carboxylic acid, in BRB081 the genes respectively are ORFs 2, 1, 4, 25, 3, 26, 27 (Fig S1a). These two structures are substrates of NRPSs codified by ORFs 18 and 19 (sibD and sibE), that catalyze the formation of the diazepine ring of sibiromycin, all genes involved in their synthesis are represented in the putative BGC of Streptomyces sp. BRB081 (Table 5).

The NRPSs, encoded by the sibD and sibE genes, have a modular composition identical to the related NRPSs ICN76_028635 and ICN76_028630 in Streptomyces sp. BRB081. This conformity is also maintained in the sibirosamine synthesis pathway, involving the participation of the sibI, sibJ, sibM, sibN, sibO, and sibH genes, also represented in BRB081 by ORFs 14, 13, 10, 9, 8 and 15; however, in this last strain, there is a dTDP-glucose-4,6-dehydratase codified by ORF 23 (Fig S1b).

These observations indicate that both BGCs would synthesize the same diazepine ring and sibirosamine. Some features differentiate the two biosynthetic regions (Fig. 3), but probably do not affect the final product.

Fig. 3.

Comparison between A putative sibiromycin BGC from Streptomyces sp. BRB081 and B Streptosporangium sibiricum sibiromycin BGC (Li et al. 2009). The limits of A were defined based on homology. Correlated ORFs between A and B are represented by the same color. Dotted lines connect related and ordered ORFs in the same sequence

As mentioned above, one difference between two BGCs is the absence of rfbB gene in the S. sibiricum cluster, which codify dTDP-glucose-4,6-dehydratase necessary for the synthesis of 6-deoxyhexose of several glycosylated metabolites like this one. A copy of this gene located 9 kb upstream of sibiromycin BGC cluster would codify the enzyme responsible compensating for this absence inside this BGC (Li et al. 2009). On the other hand, a rfbB ortholog was observed in sibiromycin BGC of BRB081 (Fig. 3, ORF 23), and this agrees with the knowledge that all necessary genes for regulation, biosynthesis, and transport of a secondary metabolite must be contained in its BGC.

Another interesting observation made in the BRB081 cluster is the presence of an ORF for pyridoxamine 5′-phosphate oxidase (Fig. 3, ORF 21), that catalyzes the conversion of pyridoxamine phosphate to pyridoxal phosphate (Salvo et al. 2003). This active form of vitamin B6 acts as a cofactor for a variety of enzymatic reactions, such as the reaction catalyzed by kynurenine hydrolase (the SibQ protein). As SibQ activity is dependent on this cofactor, a copy of this gene inside sibiromycin BGC could ensure the proper supply of this cofactor during sibiromycin biosynthesis.

Finally, the ortholog of SibX is absent in BRB081 BGC cluster. This protein is a putative regulator member of PucR family; however, in this cluster, the ortholog of the regulator SibA is present (ORF 22), as well as ORF 5 (similar to sibR) that contains a helix–turn–helix motif and would be another regulator. Concerning to resistance are present in BRB 081 an UvrA excinuclease resistance gene codified by ORF 17 (sibF), and an additional mechanism of resistance (Fig. 3, ORF 24), present only in this strain that is a major facilitator superfamily transporter.

Search for sibiromycin BGC in other organisms

The approach to verify the presence of this BGC in other genomes available in public databases was to build a sequence similarity network (SSN) from NRPS protein MBL3808355.1, codified by ICN76_028635 (Table 5), and then to obtain a Genome Neighborhood Network (GNN) and Genome Neighborhood Diagrams (GND) from this protein and their relatives.

A first MBL3808355.1SSN was generated using parameters to group the retrieved 1000 proteins from Uniprot database in clusters of proteins sharing about 30% of identity (Score = 205) or more. In this case, only one cluster with all proteins was found (Figure S2, Supplementary Data), this result indicates all members could belong to the same family.

Then, a second SSN was obtained increasing the threshold from 205 to 380, the aims was to obtain some putative isofunctional clusters wherein MBL3808355.1 could be inserted. In fact, with this stringency, the proteins were spread into 35 clusters (Figure S3, Supplementary Data), MBL3808355.1 group with other 11 related proteins in cluster 10, including SibD from S. sibiricum. (Table S2, Supplementary Data). Reducing the initial database to a closely related group of proteins, we can use the Genome Neighborhood Tools (Zallot et al. 2019) to create a GNN and explore if any of these proteins are inserted in sibiromycin-like BGCs.

The GND generated with these data shows that the local genomic contexts of all cluster members (Figure S4B, Supplementary Data) are very similar to sibiromycin BGC from S. sibiricum and, by the way, Streptomyces sp. BRB081. In fact, BGCs from Streptomyces sp. AmelKG-E11A and Streptomyces sp. SID4919 (Figure S4B, Supplementary Data) have the same organization found in Streptomyces sp. BRB081 (Figure S4A, Supplementary Data).

Although the literature does not report the occurrence of the sibiromycin BGC in any other microorganism than S. sibiricum, we have located this BGC in 10 other genomes besides the BRB081 isolate, all of them belonging to the Actinomycetia class of Actinobacteria phylum (Table S2; Figure S4, Supplementary Data). The five genera represented were Streptomyces (4 spp.), Micromonospora (2 spp.), Couchioplanes (2 spp.), Actinomadura (1 spp.), and Thermomonospora (1 spp.).

Search of sibiromycin in cultivation extracts of Streptomyces sp. BRB081

The recovered yield of putative sibiromycin fraction was 0.1 mg/L. To target this molecule and its analogs in the LC–MS/MS analysis, ion-extraction chromatograms of m/z 476.3 [M + H]⁺ (sibiromycin), m/z 458.3 [M + H]⁺ (imine form) and m/z 476.3 [M + H]⁺ (methoxy-sibiromycin) were obtained (Fig. 4). Thus, the MS/MS spectrum of each extracted ion was compared to the literature to confirm their production by Streptomyces sp. BRB081.

Fig. 4.

Fragmentation spectra of sibiromycin and its analogs produced by Streptomyces sp. BRB081. In each spectrum, structures are represented showing the fragmentation of the detected compounds. Blue diamonds in each spectrum represent the precursor ion selected for fragmentation

The MS/MS fragmentation of m/z 476.3 [M + H]⁺ showed product ions at m/z 303.2 and m/z 174.2, consistent with the previously reported mass spectrum for sibiromycin (Sulc et al. 2011). Similarly, fragmentation of ions of m/z 458.3 [M + H]⁺ and m/z 490.3 [M + H]⁺ showed, respectively, product ions at m/z 285.2 and m/z 174.2, and at m/z 317.2 and m/z 174.2. These data also indicate the production of the imine form of sibiromycin and methoxy-sibiromycin (Sulc et al. 2011).

Discussion

The search for molecules with antitumor activity from marine microorganisms has added very interesting data to the literature. Some metabolites from marine microorganisms are responses to virus infections, displaying abilities to mediate specific inhibitory activities on a few key cellular processes, including apoptosis pathways, angiogenesis, and migration (He et al. 2009).

An initial chemical screening of Streptomyces sp. BRB081 extract revealed the production of the octapeptide anticancer surugamide and its analogs (Tangerina et al. 2020). In this work, the WGS of this strain and a further genome mining indicated the presence of genetic attributes responsible for the biosynthesis of this molecule and several others common to the genus, but also sibiromycin, which is unprecedented in this taxon.

The synteny-based assembly of Streptomyces sp. BRB081 was able to obtain a scaffold with 6,853,601 bp which represents 98.77% of the total genome. The completeness test done with both assemblies resulted in 99.18% of Actinomycetales core genes found and 0.32% of putative duplicated genes, these results indicate that almost all genetic information from this strain was recovered in this WGS project.

A comparison of the distribution of BGCs between BRB081 and phenetically related sequences reinforced the evidence that Streptomyces sp. BRB081 and S. rutgersensis NBH77 have a very close phylogenetic relationship (Figure S4, Supplementary Data). BRB081 and NBH77 have an almost identical composition of BGCs, which are arranged in the same order in their genomes. In fact, their secondary metabolic profiles only differ at the chromosomal ends, and this is a very common phenomenon in Streptomyces linear genomes, exposed to a very high degree of genetic instability and rearrangements at their ends (Volff and Altenbuchner 1998).

These data corroborate other studies that demonstrate a discrepancy between the repertory of secondary metabolites of phylogenetically identical Streptomyces but with distant geographic origins, suggesting that these chemical differences may result through niche specialization, as well as horizontal gene transfer (Vicente et al. 2018; Sottorff et al. 2019).

The sibiromycin BGC was not located in the genome of S. rutgersensis NBH77. However, through the construction of the sequence similarity network (SSN) and the genome neighborhood network (GNN), this BGC was identified in 10 other genomes of the Actinobacteria phylum. SSN allows the visualization of relationships between protein sequences, where the most closely related proteins are grouped into clusters. The GNN, on the other hand, provides a statistical analysis of the frequency of co-occurrence of queries and their genome neighbors for each SSN cluster (Zallot et al. 2019).

The microorganisms harboring a putative sibiromycin BGC were isolated mainly from soil or associated with insects; however, this group also includes isolated species from lichen (Streptomyces uncialis), hot springs (Streptomyces calidiresistens), and marine sponges (Actinomadura craniellae).

As there is no pattern between the niches of all these isolates, it is not yet clear against which environmental characteristics sibiromycin biosynthesis is started, nonetheless as mentioned above responses against bacteriophages would be a potential answer. However, these data are enough to demonstrate that much is unknown about the Actinobacteria phylum, and how genomics can be informative in screening studies.

The genomic evidence that BRB081 is capable of biosynthesizing sibiromycin was confirmed by the identification of this molecule in the culture broth of this marine isolate. In chemical analysis, low-yielding compounds with high biological activity can be suppressed by other main components of the extract. We believe that this may be the reason why sibiromycin and its analogs were not detected in the work by Tangerina et al. (2020). These results reinforce the importance of analyzing both genomic and metabolomic data for individual strains to maximize the finding of bioactive compounds, as recently discussed by Schorn et al. (2021).

Conclusions

The detection of sibiromycin BGC in Streptomyces sp. BRB081 and another ten Actinobacteria genomes, offers several chances to express genes in heterologous hosts to enhance the production of bioactive compounds, and also makes possible to introduce structural modifications of the compounds to develop unexpected biological activities. The incorporation of genome mining into natural products screening programs offers multiple possibilities, including the identification of BGCs responsible for low yield metabolites production, as demonstrated in the present work. But also, the prospects include the use of elicitors or heterologous expression systems to increase the production of the low yield metabolites, the identification of cryptic BGCs, that can be awake; as well as the possibility of introducing genetic modifications in the producing organisms, bringing a whole new chemical diversity to light.

Supplementary Information

Below is the link to the electronic supplementary material.

Funding

This research was supported by São Paulo Research Foundation (FAPESP) grants 2017/16606-6 and 2015/17177-6, CAPES (Finance Code 001, Brazil), and Pró-Reitoria de Pesquisa da USP (PIPAE 2021.1.10424.1.9). L.V.C.L. and M.J.P.F were funded by a fellowship from Conselho Nacional de Desenvolvimento Científico e Tecnológico—CNPq. V.M.B.L. were funded by a fellowship from Coordenação de Aperfeiçoamento de Pessoal de Nível Superior—CAPES.

Declarations

Conflict of interest

The authors declare that they have no conflict of interest in the publication. This research does not involve human/ animal participants.