Insilico identification of silent primycin biosynthetic gene clusters within the family Pseudonocardiaceae

Abstract

In the global effort to discover or design new effective antibiotics to fight infectious diseases, the increasingly available multi-omics data with novel bioinformatics tools open up new horizons for the exploration of the genetic potential of bacteria to synthesize bioactive secondary metabolites. Rare actinomycetes are a prolific source of structurally diverse secondary metabolites that exhibit remarkable clinical and industrial importance. Recently several excellent genome mining tools have been available for identifying biosynthetic gene clusters, however in cases of poor-quality sequences and inappropriate genome assembly, these tools are not always able to identify the corresponding gene clusters. In this context, based on the previously characterized primycin biosynthetic gene cluster (PBGC) of the Saccharomonospora azurea SZMC 14600 strain an extended genome mining analysis was performed to advance the industrial application of unexplored taxa outside the Saccharomonospora genus. Further improvement of S. azurea PBGC revealed 28 clustered genes including core sequences for type I polyketide synthase. Application of PBGC core elements and targeted genome mining workflows revealed three species from the family Pseudonocardiaceae that can be considered potential candidates for primycin production, despite the fact that the discovered biosynthetic gene clusters were silent under the currently applied laboratory-culture condition. The findings presented here demonstrate the potential of an in silico toolkit and draw attention to the importance of awakening the dormant biosynthetic potential.

1. Introduction

Rare actinomycetes are known as a prolific source of structurally diverse bioactive secondary metabolites that exhibit remarkable clinical and industrial importance [1]. Representatives of the family Pseudonocardiaceae such as the vancomycin producer Amycolatopsis orientalis or the erythromycin producer Saccharopolyspora erythraea have been known for long as reliable sources of effective antimicrobial agents in combating threatening bacterial infections [2,3]. Although in the past decade, there has been a growing interest focused on the family Pseudonocardiaceae, the actual biosynthetic potential is still poorly characterized in the case of the related genera. Saccharomonospora azurea a Gram-positive filamentous bacterium of the Saccharomonospora genus is known as the only producer of primycin, a 36-membered marginolactone antibiotic that possesses high antimicrobial activity against frequent Gram-positive pathogens, including clinically prevalent multidrug-resistant strains [4]. Primycin is a mixture of homologous components characterised by a central lactone ring with a terminal guanidine moiety on a side chain. Up to date nine major and several minor components has been described differ in the arabinose moiety and the alkyl functional groups in R1 and R2 positions (Fig. 1) [5].

Fig. 1.

General chemical structures of primycins. The R1 substituent represents O-arabinose, -H, or -OH functional group in primycin A, B, and C component, while the R2 substituent represents n-butyl, n-pentyl, or n-hexyl functional group in component subgroups 1, 2, and 3.

It has been demonstrated that the guanidine-containing primycin molecule is synthesized by the bacterial type I polyketide synthase multienzyme complex, which is composed of multiple covalently linked catalytic domains grouped into functional modules [6]. In general, each PKS module harbours a minimal set of ketosynthase (KS), acyltransferase (AT), and acyl carrier protein (ACP) domains that extend the linear sequence of the growing polyketide intermediate by two carbon atoms. Additionally, PKSs may have optional ketoreductase (KR), dehydratase (DH), and enolyreductase (ER) domains to further modify the polyketide backbone and thioesterase (TE) domain to catalyze the release of the synthesized product [7].

Biosynthetic genes required for the production of a certain secondary metabolite are often arranged into contiguous clusters in the genome. These tightly linked sets of biosynthetic genes typically include genes encoding the whole biosynthetic process, such as precursor biosynthesis, product assembly, post-synthetic modifications, transport mechanisms, self-resistance, and regulation [8]. Our previously published structural genomic data revealed that several secondary metabolite biosynthetic genes, clustered together with the primycin PKS take an active part in the primycin biosynthesis. Among them amine oxidase, amidohydrolase, and acyl-CoA ligase gene products - members of the l-arginine precursor pathway - catalyze the formation of the 4-guanidinobutanoyl-CoA starter unit of the primycin biosynthesis, while S-malonyltransferase is responsible for transferring the activated 4-guanidinobutanoyl group to the loading ACP domain of the primycin PKS [9]. Acyl-CoA carboxylase is responsible for the assembly of unusual alkylmalonyl-CoA (butlymalonyl-CoA, pentlymalonyl-CoA or hexylmalonyl-CoA) PKS extender units incorporated into the primycin molecules. After release from the PKS enzyme complex, the synthesized primycin molecule may undergo several post-synthetic modifications. Conversion between guanidino and amino forms of primycin is catalyzed by agmatinase enzyme, by a nucleophilic attack on the amidino carbon similar to desertomycin [10], while in the case of primycin A1, A2 and A3 isoforms glycosyltransferase catalyze the transfer of the O-arabinose moiety onto the carbon atom in C-18 position analogue to ECO-0501 biosynthesis pathway [11].

Our in silico structural and functional comparison of primycin producer S. azurea strains revealed ABC superfamily multidrug transporter encoding genes positioned both in the upstream and downstream direction from the primycin PKS core genes. The performed transcriptomic analysis highlighted that these ABC transporters encoding genes are transcriptomically active during primycin biosynthesis, suggesting their contribution to self-resistance to the synthesized antibiotic. Similarly, a receptor histidine kinase and a cognate response regulator encoding genes, elements of the two components’ signal transduction system located within the primycin biosynthetic gene cluster showed overexpression in high-primycin producer strain confirming their importance in the regulation of primycin biosynthesis.

2. Materials and methods

2.1. Bacterial strains and culture conditions

Amycolatopsis orientalis DSM 40040, Kibdelosporangium aridum DSM 43828, and Saccharomonospora azurea DSM 44631, strains used in this study were purchased from Leibniz Institute, DSMZ - German Collection of Microorganisms and Cell Cultures, while S. azurea SZMC 14600 was originated from Szeged Microbiology Collection (SZMC). Bacterial cultures were stored as vegetative cell suspension supplemented with 20 % (v/v) glycerol at −80 °C. Culture conditions were carried out as follows: 1 mL bacterial cell suspension of S. azurea DSM 44631 and S. azurea SZMC 14600 was inoculated into 50 mL Luria-Bertani (LB) liquid medium contained 1 % (w/v) tryptone, 1 % (w/v) NaCl, 0.5 % (w/v) yeast extract (pH 8.0) and incubated for 48 h at 37 °C in an orbital shaker at 200 rpm. In the case of A. orientalis DSM 40040, K. aridum DSM 43828 strains the following growth medium (GM) [12] was used: 1.7 % (w/v) glucose, 1.1 % (w/v) bacto peptone, 0.3 % (w/v) yeast extract, 0.3 % (w/v) malt extract (pH 7.2) and incubated for 48 h at 37 °C in an orbital shaker at 200 rpm. Subsequently, 1 mL of accurately homogenized seed culture was transferred into 35 mL of primycin fermentation medium (FM) contained 4 % (w/v) soy flour, 4 % (w/v) water soluble starch, 0.3 % (w/v) NaCl, 0.5 % (w/v) CaCO₃, 0.3 % (w/v) stearic acid, 0.1 % (w/v) KH₂PO₄, 0.6 % (w/v) sunflower oil (pH 9.5) and cultivated for 7 days at 28 °C in an orbital shaker at 200 rpm. Bacillus subtilis ATCC 6633 used for agar well diffusion assay was purchased from the American Type Culture Collection (ATCC).

2.2. Antimicrobial assay

Antimicrobial activity of n-butanol-ethanol-distilled water 1:1:2 (v/v) (BEW) extracts of 5-day fermented cells was determined by agar well diffusion assay according to Kovács et al. [13]. Primycin concentration of cell extracts was compared to crystallized primycin-sulphate reference standard prepared in-house. Samples were obtained from three independent fermentations and triplicated as technical replicates.

2.3. Chromatographic analysis

High-performance liquid chromatography (HPLC) with diode array detection (DAD) and electrospray-mass spectrometry (ESI-MS) detection-based analysis of cell extracts were carried out on an Agilent 1200 Series HPLC System coupled to the Agilent 6120 Quadrupole LC/MS (Agilent Technologies) instrument according to Kovács et al. [13]. Primycin concentrations were determined from 0.5 mL fermentation media, collected on the fifth day of fermentation. All samples were prepared in three independent biological replicates and measured in technical triplicates. Statistical analysis was completed using one-way analyses of variance (ANOVA). Values are reported as mean ± SD (standard deviation) and results were expected statistically significant when p < 0.05.

2.4. Comparative genomics tools

Records of the annotated genomes are displayed by the following accession numbers in GeneBank: A. orientalis B 37 - CP016174.1; A. orientalis DSM 40040 - ASJB00000000.1; A. orientalis DSM 43388 - ASXG00000000.1; A. orientalis DSM 46075 - ASXH00000000.1; A. pittospori PIP 199 - JACBXD000000000.1; K. aridum DSM 43828 - FWXV00000000.1; K. aridum A82846 - QHKI00000000.1; K. aridum subsp. largum NRRL B-24462 - JNYM00000000.1; S. azurea SZMC 14600 - AHBX00000000.1; S. azurea DSM 44631 - AGIU00000000.2; S. azurea isolate bin47 - JAJUGW000000000.1. In silico DNA-DNA hybridisation (DDH) values between A. orientalis, A pittospori, K. aridum and S. azurea species were calculated by using the Genome-To-Genome Distance Calculator (GGDC) web server (https://ggdc.dsmz.de) [14]. Distance values were determined by the recommended Formula 2 for incomplete draft genomes.

2.5. Identification and in silico structural analysis of primycin PKS gene cluster

The primycin type I PKS gene cluster was identified and analysed by antiSMASH (Antibiotics & Secondary Metabolite Analysis Shell) [15]. Database searches for homologous genes and proteins were performed using the National Center for Biotechnology Information (NCBI) BLAST server [16]. Domain analysis and motif search were done by SMART (Simple Modular Architecture Research Tool) [17], SBSPKS (Structure Based Sequence Analysis of Polyketide Synthases) [18] and MEME (Multiple Em for Motif Elicitation) [19]. Multiple sequence alignments were performed by CLUSTAL W [20].

3. Results

3.1. Screening of primycin biosynthetic gene cluster

To test the hypothesis that a complete primycin biosynthetic gene cluster (PBGC) occurs outside the Saccharomonospora genus, the S. azurea SZMC 14600 draft genome was analysed by antiSMASH analysis tool. Accordingly, the S. azurea PBGC consist of 52 genes including type I PKS core biosynthetic genes along with the closely related additional biosynthetic genes, clustered together in a 126,698 nucleotide-long region. Due to the S. azurea SZMC 14600 incomplete genome assembly, genes representing the PBGC are located on two separate contigs (AHBX01000216.1 and AHBX01000133.1), however structural genomic analysis revealed that the genomic region of 169455–183316 and 30203–44064 in the case of contig 216 and contig 62 respectively overlap and share 100 % sequence identity. Therefore the assembly was revised and henceforth treated as one continuous contig in subsequent analyses (Fig. 2).

Fig. 2.

Elements of the primycin biosynthetic gene cluster, identified in S. azurea SZMC 14600. Two parallel lines indicate the antiSMASH identified primycin type I PKS (upper line) and detailed genomic position on contig 216 and contig 62 (lower line). Extension of the PBGC in 5′ and 3′ directions relative to the primycin type I PKS is indicated by colours as follows: claret – core biosynthetic genes, pink – additional biosynthetic genes, light blue – transport-related genes, green – regulatory genes and grey – other genes. Arrow orientation represents transcriptional direction.

Even though genes encoding the L-arginine precursor pathway of the primycin biosynthesis (acyl-CoA ligase - EHK88415.1, carbon-nitrogen hydrolase family protein - EHK88411.1, amine oxidase - EHK88410.1) are essential elements of the PBGC, the performed antiSMASH analysis did not identify these genes as part of the gene cluster. Nevertheless, an extended homology search carried out on AHBX01000133.1 contig, revealed the members of the L-arginine precursor pathway downstream in the immediate vicinity of the antiSMASH predicted PBGC. In total of 63 genes of the extended S. azurea SZMC 14600 PBGC and their closest homologues with the highest percent identity are listed in Table 1S. Although Table 1S shows only the top hits of homologues, representing twelve different taxa, further high scores were found in the case of several other species. Sequence homology search of the PBGC core elements, represented by 28 genes, K. aridum revealed the highest sequence identity in 27 cases, followed by A. orientalis, and A. pittospori species.

Table 1.

In silico genome-to-genome comparison of the high primycin producer S. azurea SZMC 14600 genome to the low primycin producer S. azurea DSM 44631 and the primycin biosynthetic gene cluster containing A. orientalis, A pittospori and K. aridum strains.

Organism	Strain	HSP length/total length (%)	identities/HSP length (%)	identities/total length (%)
S. azurea	DSM 44631	91.30	93.40	93.90
S. azurea	isolate bin47	84.60	92.20	88.70
A. orientalis	DSM 43388	14.80	20.60	14.90
A. orientalis	DSM 46075	14.70	20.50	14.90
A. orientalis	B-37	14.90	20.00	15.00
A. orientalis	DSM 40040	15.00	19.90	15.10
A. pittospori	PIP199	15.10	19.80	15.20
K. aridum	DSM 43828	13.90	19.90	14.20
K. aridum	A82846	13.90	19.90	14.10
K. aridum subsp. largum	NRRL B-24462	13.50	20.10	4.22

1 Length of all HSPs divided by total genome length in %.

2 Sum of all identities found in HSPs divided by overall HSP length in %.

3 Sum of all identities found in HSPs divided by total genome length in %, where HSP meaning high-scoring segment pairs according to https://ggdc.dsmz.de.

The PBGC-based sequence similarity of these representative species was further validated by the antiSMASH ClusterBlast analysis, shown in Fig. 3. Detailed sequence analysis of the corresponding species and their related strains that can be found in the public databases are presented in Table 2S. Surprisingly, in the case of S. azurea bin47, A. orientalis DSM 43388 and A. orientalis DSM 46075 strains, antiSMASH analysis could not identify PBGC-related genes.

Fig. 3.

Primycin biosynthetic gene cluster of S. azurea SZMC 14600 and homologous gene clusters found in S. azurea DSM 44631, K. aridum DSM 43828, A. orientalis B 37 and A. pittospori PIP199. Genes marked with the same colour are interrelated.

Table 2.

Sequence similarities between the primycin PKS catalytic domains in comparison of S. azurea SZMC 14600, S. azurea DSM 44631, K. aridum DSM 43828, K. aridum A 82846, A. orientalis DSM 40040 and A. orientalis B 37.

Domain		Average identity with S. azurea SZMC 14600 (%)
		S. azurea DSM 44631	K. ariudum DSM 43828	K. ariudum A82846	A. orientalis DSM 40040	A. orientalis B-37	A. pittosporPIP199
Type	Total number
KS	18	99.00	91.1	91.2	84.2	84.2	84.1
AT	19	97.7	82.2	81.7	71.9	71.8	71.0
DH	8	99.5	81.9	84.3	68.4	69.0	70.1
ER	5	99.2	88.6	88.2	80.4	80.8	79.6
KR	18	98.2	85.2	84.7	76.9	77.4	76.9
ACP	19	99.9	85.1	84.7	74.3	74.7	74.7
TE	1	99.0	86.0	84.0	75.0	75.0	–
	Average	98.9	85.7	85.5	75.9	75.9	76.1

3.2. In silico DNA–DNA hybridisation

Genome-wide in silico comparative analysis of the known primycin producer S. azurea strains revealed a high percentage identity (93.4 %) among the high-scoring segment pairs (HSPs). Recently identified S. azurea isolate bin47, which lacks the PBGC also shares a high degree of identity (92.2 %) with the high-primycin producer S. azurea SZMC 14600. In contrast, the comparison of the S. azurea SZMC 14600 genome to the newly identified potential primycin producer A. orientalis, A. pittospori and K. aridum strains resulted only a low-level identity (≤20 %) of HSPs (Table 1).

3.3. Primycin type I PKS

Analysis of the primycin PKS catalytic domains revealed a high percentage of average identity (98.9 %) between the two S. azurea strains. Identifying homologous sequences of the primycin PKS multienzyme in NCBI protein databases resulted in a higher level of sequence similarity in the case of K. aridum over A. orientalis and A. pittospori strains. Multiple sequence alignments of the catalytic domains of these newly identified primycin PKS megasynthases revealed an average 85.6 %, 76 %, and 76.1 % overall identity in the case of K. aridum, A. orientalis and A. pittospori to S. azurea SZMC 14600 respectively (Table 2). Even though several catalytic domains from the PBGC could be identified in the case of K. aridum subsp. largum NRRL B-24462 strain, the complete PKS domain organisation could not be assembled because of the low-quality sequencing data available in the NCBI database.

In good agreement with the formerly characterized primycin type I PKS of S. azurea SZMC 14600 the newly identified primycin type I PKSs are composed of one loading and 18 separate extender modules divided into 6 subunits in the case of each strain [6]. The antiSMASH analysis also confirmed malonyl-CoA substrate specificity at modules 4–10 and 12–16 and methylmalonyl-CoA substrate specificity at modules 1–3, 11, and 17 AT domains in each strain. Uniformly at module 18 each AT domain is bearing an unusual substrate-specific sequence motif (Q-GHSQG-R-GAGH) as was demonstrated in the case of S. azurea SZMC 14600 (Fig. 1S) [6]. Determination of the stereospecificity of ketoreduction generated by each KR domain predicts 7 A1-type (2R, 3S), 10 B1-type (2R, 3R), and 1 B2-type (2S, 3R) KR domain within the primycin type I PKS (Table 3S).

3.4. Primycin producing ability

Among the investigated species only the known primycin producer S. azurea strains were proved to possess antimicrobial activity against the Bacillus subtilis test organism. The rate of primycin production differed significantly among S. azurea SZMC 14600 (1683.14 ± 82.12 mg/L) and S. azurea DSM 44631 (238.70 ± 51.49 mg/L) strains in good agreement with our previously published data. In contrast, even though K. aridum DSM 43828 could grow well in the primycin fermentation media, no antibiotic production could be detected by the agar well diffusion bioassay. Although the growth of K. aridum cells was not associated with primycin production, dark-brown pigment product was present in the fermentation media as a sign of active secondary metabolism. In the case of A. orientalis DSM 40040, no cell growth was visible in the fermentation media, consequently no antimicrobial activity could be detected (Fig. 4). Primycin-producing ability was also confirmed by HPLC–DAD-ESI/MS analysis. S. azurea SZMC 14600 and DSM 44631 strains yielded 1890.40 ± 161.87 mg/L (Fig. 5) and 204.93 ± 1.92 mg/L primycin respectively.

Fig. 4.

Agar well diffusion assay of primycin extracts from S. azurea SZMC 14600, S. azurea DSM 44631, K. aridum DSM 43828 and A. orientalis B 37 strains. Numbers from 1 to 6 indicate serial dilutions of fermented cell extracts. Wells of crystallized primycin-sulphate standard correspond to the following concentrations: (1) 12.5 μg/mL; (2) 6.25 μg/mL; (3) 3.125 μg/mL; (4) 1.56 μg/mL; (5) 0.78 μg/mL; (6) 0.39 μg/mL.

Fig. 5.

Reversed-phase high-performance liquid chromatography (HPLC) and electrospray-mass spectrometry (ESI-MS) chromatograms of the high-primycin producer S. azurea SZMC 14600 cells extract. HPLC chromatogram (A) peaks with retention time of 5.608 min and 6.723 min correspond to primycin A1 (m/z 1078.7) and primycin C1 (m/z 946.5) respectively, based on molecular masses detected by ESI-MS (B). Data was acquired for 22 min over a 950–1150 m/z range in the positive ions mode.

4. Discussion

Industrial strain improvement practices of bioactive secondary metabolite producers microorganisms ranging from traditional mutagenesis and random screening to metabolic engineering are often time-consuming, cost- and labour-intensive processes [21]. Recent advances in whole genome sequencing and bioinformatic data processing have provided new opportunities to discover an immense reservoir of previously unrecognized biosynthetic gene clusters. A growing body of evidence confirms that diverse secondary metabolic gene clusters are present in genomes of rare actinomycetes, which are barely expressed or remain phenotypically “silent” under ordinary growth conditions [22]. Even though genome-wide comparative sequence analysis - i.e. in silico DNA-DNA hybridisation – revealed only low-level sequence similarities (∼20 %) across the analysed species of the family Pseudonocardiaceae, to our surprise, the detailed targeted sequence comparison revealed that PBGC is ubiquitously present and represents high-level sequence identities. The presence of PBGC including the PKS core biosynthetic genes was further supported by anstiSMASH prediction.

However the PKS core biosynthetic genes identification was hindered by the fact that the fragments of the primycin PKS were identified as separate entities on different contigs in the case of A. orientalis DSM 40040, A. pittospori PIP199 and K. aridum A82846 strains. Due to the manual reconstruction based on multiple sequences alignment of the PKS multienzyme catalytic domains we were able to identify complete primycin type I PKS in the case of all strains. Interestingly our previous finding related to S. azurea SZMC 14600 AT domain unusual substrate specificity (butylmalonyl-CoA, pentylmalonyl-CoA or hexylmalonyl-CoA) in module 18, was also characteristics in all the newly identified potential primycin producer species.

To implement the acquired knowledge obtained from the in silico identification of the PBGC genes into practice, A. orientalis DSM 40040 and K. aridum DSM 43828 strains were applied for primycin fermentation. Both strain was propagated well in the seed medium (GM), in which no primycin production was detectable. After GM-based propagation of the species, cell suspensions were inoculated into fermentation media (FM) optimized for primycin biosynthesis, nevertheless, dense mycelial growth could be observed only in the case of K. aridum DSM 43828. On the fifth day of fermentation, dark-brown soluble pigment was present in the fermentation media indicating obvious secondary metabolic activity, however cell growth was not associated with primycin production. In contrast, inoculation of FM with A. orientalis DSM 40040 cell suspension resulted in a complete lack of cell growth, consequently primycin synthesis could not be observed, suggesting incompatibility with the applied fermentation media.

Due to the growing number of quality genome sequencing data and advanced genome mining approaches, we can gain deeper insight into the secondary metabolic production potential of rare actinomycetes. Although the utilization of genome mining to discover the biosynthetic potential of microorganisms is a powerful strategy, it requires subsequent and extensive laboratory processing to integrate the acquired deep mechanistic understanding into advanced strain improvement practices [23]. Despite the fact that screening under primycin fermentation conditions resulted in a lack of antibiotic production, among the newly identified potential primycin-producing species, sheds light on the limitations and benefits of in silico identification of PBGC. In conclusion, the bioinformatics-based approach was able to identify promising antibiotic-producing new species and that fact could serve as a good starting point for future advanced strain improvement projects.

CRediT authorship contribution statement

Márk Kovács-Valasek: Writing – original draft, Visualization, Methodology, Investigation, Formal analysis. Csaba Fekete: Writing – review & editing, Writing – original draft, Supervision. Andrea Kovács-Valasek: Writing – review & editing, Visualization, Supervision, Methodology.

Ethics declarations

Not applicable.

Funding information

This research was partly supported by the GINOP-2.1.2-8.1.4–16 grant and by PannonPharma, Ltd., Pécsvárad, Hungary.

All data generated for this work are available as Supplementary Materials.

Declaration of competing interest

The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: Mark Kovacs-Valasek reports equipment, drugs, or supplies was provided by PannonPharma Pharmaceutical Ltd. Mark Kovacs-Valasek reports a relationship with PannonPharma Pharmaceutical Ltd. that includes: employment. If there are other authors, they declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary data

The following is the supplementary data to this article: