Functional connexion of bacterioferritin in antibiotic production and morphological differentiation in Streptomyces coelicolor

Abstract

Background

Several two-component systems of Streptomyces coelicolor, a model organism used for studying antibiotic production in Streptomyces, affect the expression of the bfr (SCO2113) gene that encodes a bacterioferritin, a protein involved in iron storage. In this work, we have studied the effect of the deletion mutant ∆bfr in S. coelicolor.

Results

The ∆bfr mutant exhibits a delay in morphological differentiation and produces a lesser amount of the two pigmented antibiotics (actinorhodin and undecylprodigiosin) compared to the wild type on complex media. The effect of iron in minimal medium was tested in the wild type and ∆bfr mutant. Consequently, we also observed different levels of production of the two pigmented antibiotics between the two strains, depending on the iron concentration and the medium (solid or liquid) used. Contrary to expectations, no differences in intracellular iron concentration were detected between the wild type and ∆bfr mutant. However, a higher level of reactive oxygen species in the ∆bfr mutant and a higher tolerance to oxidative stress were observed. Proteomic analysis showed no variation in iron response proteins, but there was a lower abundance of proteins related to actinorhodin and ribosomal proteins, as well as others related to secondary metabolite production and differentiation. Additionally, a higher abundance of proteins related to various types of stress, such as respiration and hypoxia among others, was also revealed. Data are available via ProteomeXchange with identifier PXD050869.

Conclusion

This bacterioferritin in S. coelicolor (Bfr) is a new element in the complex regulation of secondary metabolism in S. coelicolor and, additionally, iron acts as a signal to modulate the biosynthesis of active molecules. Our model proposes an interaction between Bfr and iron-containing regulatory proteins. Thus, identifying these interactions would provide new information for improving antibiotic production in Streptomyces.

Background

Streptomyces are Gram-positive filamentous bacteria that thrives in various environments, such as soil, where they are highly abundant. These bacteria belong to the phylum Actinomycetota (previously known as Actinobacteria) [1] and have a complex life cycle: a vegetative branched mycelium emerges from a spore, and, under adverse conditions, differentiates into a spore-producing aerial mycelium [2]. They possess a remarkable ability to synthesise industrial compounds, such as extracellular enzymes, and natural products such as antibiotics (Streptomyces produces approximately 80% of the antibiotics used) [3]. Furthermore, the production of these active molecules is controlled by intricate regulatory networks that link their development, environmental signals, and metabolism [4].

Trace metals act as signals that influence the development of microorganisms, with iron being an essential component in bacteria [5]. The relationship between iron homeostasis, antibiotic production, and morphological differentiation has been previously reported for the genus Streptomyces [5, 6]. It is known that iron deprivation triggers actinorhodin (ACT) production in S. coelicolor [7] and a null mutant of the siderophore receptor desE in this species exhibits altered morphological differentiation and ACT production [8]. The impact of iron on antibiotic production has also been described for other Streptomyces species, such as S. hygroscopicus, where niphimycin production is enhanced by high iron concentrations [9]. However, low iron concentrations drive exploratory growth, a new mode of Streptomyces development [10], and co-cultures of Myxococcus with other Streptomyces species increase their secondary metabolite expression due to iron scarcity [7]. Thus, several metabolites and processes are dependent on iron [11].

Additionally, several elements are responsible for iron homeostasis with functions related to uptake, storage, and detoxification, among others. As iron levels are usually low in the ecological niches of many bacteria, adaptive responses have been developed for the uptake of this essential resource [12,13,14]. For example, three different trishydroxamate siderophores (coelichelin, desferrioxamine E, and desferrioxamine B) share the role of acquiring ferric iron during the vegetative growth of S. coelicolor and S. ambofaciens. In S. coelicolor, desferrioxamines, and coelichelin biosynthesis, reception, and assimilation are carried out by the desEFABCD and cchABCDEFEHIJ clusters, respectively. Another siderophore importer is codified by cdtABC [8, 15]. Once ferric iron is inside the cell, owing to the work of chelators, it is reduced to ferrous iron to diminish its affinity for the siderophore and is released for its use in cellular metabolism [12].

Alternatively, ferrous iron is potentially toxic due to the Fenton reaction of ferrous ions with H₂O₂ which produces hydroxyl radicals. Therefore, cells store iron mainly in ferritins and bacterioferritins (Bfr) to avoid toxicity. Iron storage proteins take up iron in the soluble ferrous form, which is deposited in their central cavity in the oxidised ferric form. This reservoir can be used when external iron supplies are restricted [12, 16]. Therefore, iron sequestration is one of the main protection systems to counteract the effect of the iron-based Fenton reaction [17, 18].

In S. coelicolor, the bacterioferritin Bfr and three Dps proteins (DNA-binding protein from starved cells) have a putative function related to iron storage. The main role assigned to bacterioferritins is to store excess iron in conditions of metal abundance, and to release this iron in response to iron starvation [16, 17]. However, the specific physiological role of the S. coelicolor bacterioferritin remains to be elucidated. To date, it is known that the S. coelicolor bfr gene (SCO2113) is induced under osmotic stress by SigB (SCO0600) [19] and that the protein Bfr has been detected in extracellular vesicles of S. coelicolor together with other stress-related proteins [20]. Moreover, the three Dps proteins of S. coelicolor (DpsA, DpsB, and DpsC) play a role in nucleoid condensation in spore formation. However, they are not induced under oxidative stress conditions [21] as described for Dps in other bacteria [22]. DpsA and DpsC can oxidise ferrous iron and store it as ferric iron inside their cavities [23]. Also, HbpS, the accessory and sensory component of the two-component system SenS/R, provides an additional defence mechanism in the presence of iron [18, 24].

All bacterioferritins are heme-containing 24-mer proteins with a spherical structure of about 500 kDa. Typically, there are 12 hemes groups per 24-mer located at each of the 12 two-fold interfaces between units, which can bind 2000–3000 iron atoms per 24-mer [17, 25]. The crystal structure of several Bfr isolated from different microorganisms has been resolved [26, 27] including the Bfr from S. coelicolor (PDB ID:5XX9). By employing cryo-electron microscopy and crystallisation, Jobinchen et al. reported the iron mineralization inside S. coelicolor Bfr and proposed the importance of specific residues and heme groups in the internalisation and efflux of iron ions [28].

Overall, iron homeostasis must be strictly regulated to co-ordinate the availability of iron in each stage of cell development. In S. coelicolor, this regulation is mediated by two DtxR proteins, DmdR1 and Dmdr2, and the antiparallel gene adm. It has been proposed that the protein Adm may act as a repressor of antibiotic biosynthesis and is under the control of RpoZ, a sigma factor that connects development and secondary metabolism [29]. IdeR is the homologue of DmdR1 and DmdR2 in S. avermitilis and has pleiotropic regulatory functions that allow S. avermitilis to adapt its growth in a complex environment in response to iron levels [30].

In previous work, we explored the regulatory network that triggers antibiotic production in S. coelicolor and the two-component systems (TCSs) responsible for signal transduction that act as global regulators of this production [31, 32]. Specifically, for bfr, we know that this gene is a direct target of AbrC3, a positive antibiotic response regulator characterized by our group [33]. Moreover, the expression of bfr is altered in null mutants of other studied TCSs, such as the orphan regulator Aor1 and the TCS AbrB1/B2 [34, 35].

In this work, a phenotypic study of a bfr deletion mutant in S. coelicolor is described, demonstrating the importance of the Bfr protein in antibiotic and morphological differentiation across different culture media. We also report a proteomic comparative study between this mutant and the wild-type strain, revealing significant proteomic changes observed in the bfr mutant and highlighting the role of Bfr under oxidative stress conditions. Moreover, we analysed the relationship between iron homeostasis and development in both the wild type and bfr mutant. Our results indicate that Bfr, and by extension iron homeostasis, is crucial for activating the regulatory network responsible for antibiotic production and morphological differentiation.

Methods

Strains, media, and growth conditions

Escherichia coli DH5α and ET12567 (a dam⁻ strain) were utilised to obtain DNA for transforming S. coelicolor. E. coli ET12567 (pUZ8002) [36] was used to transfer a pCRISPR plasmid derivative to S. coelicolor M145. For the antibacterial activity assay, E. coli DH5α, Staphylococcus epidermidis ATCC 14,990, and Micrococcus luteus CECT 245 were employed and grown on LB [37], BHI (Condalab, REF 1400), and YEPD [38], respectively. Bacillus subtilis (CECT 4522) was grown as an overlay on NA from Scharlau (Scharlab, S. L., REF 02-140) for the CDA bioassay [34].

For the phenotypic assays, S. coelicolor M145 (prototroph; SCP1⁻, SCP2⁻) and its Δbfr mutant strain were grown on modified NMMP [39], containing 4.5 mM NaH₂PO₄/K₂HPO₄ buffer pH 6.8 and three different iron concentrations (10 nM, 4 µM, and 2 mM), and the complex solid media YEPD, LB, R2YE, and MSA. R2YE was also used for transformation purposes, MSA for sporulation [40], and YEPD or LB for colony forming unit (CFU) quantification. NA was used for the CDA and coelymicin production assays. LB medium and NMMP, containing the three iron concentrations tested, were used for antibiotic quantification and for measuring iron and reactive oxygen species (ROS) levels. Proteomic experiments were performed on LB. TSB medium (Condalab, REF 1224) was used for pCRISPR-derived plasmid loss and DNA extraction.

Iron was added to sterilised NMMP medium using a freshly prepared and filtered 100 mM FeSO₄ stock in 10 mM HCl to prevent oxidation [41]. When necessary, the media were supplemented with antibiotics: E. coli media - ampicillin (100 µg mL^− 1), apramycin (50 µg mL^− 1), kanamycin (50 µg mL^− 1), chloramphenicol (25 µg mL^− 1), or nalidixic acid (25 µg mL^− 1) and S. coelicolor media - neomycin (20 µg mL^− 1) or apramycin (15 µg mL^− 1).

DNA manipulation and plasmid construction

Plasmid isolation, restriction enzyme digestion, ligation, and transformation of E. coli and S. coelicolor were conducted using the methods by Green et al. [42], and Kieser et al. [40], respectively. The plasmids used are listed in Additional File S1.

The bfr gene, along with its promoter, was amplified by PCR using oligonucleotides LGA001/LGA002 and genomic DNA from S. coelicolor M145 as the template. The 770-bp fragment was cloned into pKC796 [43] at a BamHI site yielding the integrative plasmid pKC796-Bfr. The same DNA amplicon (bfr and its promoter) was cloned into the multicopy plasmid pN702GEM3 [44], resulting in the pN-Bfr overexpression plasmid. Genomic DNA from S. coelicolor was extracted after grown in TSB cultures [45].

Mutant construction

To delete the bfr gene using the CRISPR/Cas9 system, a single guide RNA (sgRNA) comprising 20 bfr specific nucleotides, GCAGGACACGGATCGTCATA, was designed using the web resource CRISPy [46, 47]. Using the pCRISPR-Cas9 [48] as the DNA template and primers SAM-051 and MTS-014 (Additional File S2), the whole sgRNA was amplified by PCR. This fragment was cut with NcoI and SnaBI enzymes and inserted into pCRISPR-Cas9, digested with the same enzymes, to generate pCRISPR-Cas9-sgBfr. Additionally, 1 kb to the left and 1 kb to the right of bfr gene were amplified using genomic S. coelicolor M145 DNA and cloned into the plasmid containing sgRNA to be used as a template for repairing Cas9 cuts through homologous recombination. The 1-kb fragment upstream of the bfr gene (left, primers SAM-046/SAM-047) (Additional File S2), and 1 kb downstream (right, primers SAM-048/SAM-049) (Additional File S2) were amplified. These DNA fragments were used as the template in an overlapping PCR using SAM-046/SAM-049 primers. The resulting DNA fragment was digested with NheI, which produces sticky ends when cut with XbaI, and introduced into the XbaI site of the pCRISPR-Cas9-sgBfr to generate the final pCRISPR-Cas9-Bfr plasmid. All plasmids were sequenced and verified.

The three plasmids, the empty one (pCRISPR-Cas9), the one harbouring only the sgRNA (pCRISPR-Cas9-sgBfr), and the final plasmid with the guide and the template (pCRISPR-Cas9-Bfr) were introduced into E.coli ET12567 pUZ8002. Then, all three were transferred to S. coelicolor by interspecific conjugation [36] and apramycin-resistant colonies were selected. To induce the loss of the plasmids (which were thermosensitive), the colonies obtained were grown in liquid TSB under agitation (200 rpm) at 37 ºC for 2 days. Apramycin-sensitive colonies were again selected.

Genomic DNA was obtained from the S. coelicolor wild-type strain and putative M145 ∆bfr mutant colonies and the correct deletion of bfr gene was verified by PCR using outer primers (OP: LGA-012/LGA-013) and inner primers (IP: LGA-010/LGA-011) (Additional File S3).

Growth and antibiotic production analysis

Phenotypic studies were conducted on plates containing various solid media, inoculated with 5 × 10⁵ spores added in a 5 µL drop and incubated at 30 ºC for several days. The plates were monitored for changes in colour in the colonies formed and in the culture medium, as well as for morphology changes. Under standard conditions, RED antibiotic production can be detected after two days as red colonies, ACT production as a blue halo around the colonies, and aerial mycelium and spores appear white and grey, respectively. All assays were performed in triplicate.

Growth was assayed using liquid LB medium cultures (10 mL) inoculated with 4 × 10⁶ spores mL^− 1 grown at 28 ºC under agitation (200 rpm). The growth rate was determined by measuring the dry weight of the cultures in triplicate.

ACT and RED antibiotic production were quantified from liquid cultures (10 mL of medium with 4 × 10⁶ spores mL^− 1 incubated at 28 ºC) using the spectrophotometric method described in Yepes et al., 2011 [49]. All experiments were conducted in quadruplicate and measures were normalised by cellular dry weight.

CDA was detected as previously described [34]. Colonies of S. coelicolor M145 and ∆bfr strains, originating from a 5-µL drop containing 5 × 10⁵ spores, were grown on NA for 3 days at 30 ºC. A 5-mL overlay with Bacillus subtilis (0.2 mL, OD 0.25) in soft NA (0.5% agar) and 70 mM Ca(NO₃)₂ was added. The same overlay without Ca(NO₃)₂ was used as a negative control. The plates were incubated for an additional day at 30 ºC. The inhibition halo produced by adding Ca²⁺ was measured and compared between the two strains. This experiment was performed in triplicate.

Coelimycins were detected on solid NA supplemented with 300 mM of glutamic acid as previously described [50]. The pH was adjusted to 7.2 using NaOH. Colonies and patches of S. coelicolor M145 and ∆bfr strains with 5 × 10⁵ spores were cultivated on solid NA containing 300 mM Glu for 3 days at 30 ºC. Coelimycins were observed as a yellow halo around the colonies. This experiment was done in quadruplicate.

Iron quantification

S. coelicolor M145 and ∆bfr strains were grown in liquid LB for 5 days or NMMP containing different iron concentrations (10 nM, 4 µM, and 2 mM) for 8 days. Samples of 300 µL of mycelium were collected, washed twice with 10.3% sucrose, and frozen at – 20 ºC. Labile and total cellular iron were quantified using a colorimetric assay with ferene-S (Sigma, REF 82940) and modified using previously described protocols [51, 52].

Labile iron

Cellular pellets were lysed using a fresh lysis buffer (25 mM Tris pH 8.0, 10.3% sucrose, 3 mg mL^− 1 lysozyme) and incubated for 2 h at 37 ºC. A total of 500 µL of freshly prepared labile iron detection reagent (5 mM ferene-S and 10 mM ascorbic acid in 2.5 M ammonium acetate buffer pH 4.5) were added. After incubation for 30 min at 25 ºC, the samples were centrifuged for 15 min at 1500 g and the absorbance of ferene-Fe²⁺ complexes was measured at 564 nm.

Total iron

Cellular pellets were mixed with 500 µL of freshly prepared digestion reagent (2,25% w/v KMnO₄, 5% v/v HCl) by vortexing and incubated for 4 h at 65 ºC [53]. After 30 min at room temperature, cooled samples were treated with 500 µL of freshly prepared total iron detection reagent (5 mM ferene-S and 1 M ascorbic acid in 2.5 M ammonium acetate buffer pH 4.5) and incubated for 30 min at 25 ºC. Then, they were centrifuged for 15 min at 1500 g and the absorbance of the supernatant was measured at 564 nm.

Calibration curves were made using freshly prepared FeCl₃ standards (0.5–50 nmol) for both protocols. Measurements were made in quadruplicate and normalised by cellular dry weight.

Antibiogram assay

Three microorganisms were tested: E. coli on LB, S. epidermidis on BHI, and M. luteus on YEPD. Square Petri dishes (12 × 12 cm) containing 70 mL of medium were prepared with an overlay of 10 mL of soft agar (0.5%) and each microorganism (0.4 mL, OD 0.2). Plugs of NMMP (10 nM, 4 µM, and 2 mM iron) containing colonies (5 × 10⁵ spores) of S. coelicolor M145 and ∆bfr were obtained using a cork borer (0.7 cm in diameter) and assayed on the plates containing the lawn of microorganisms tested. Plates were incubated for 4 h at 4 ºC for antibiotics diffusion and then incubated for 2 days at 30 ºC. The resultant inhibition halos were measured. Bioassays were done in triplicate.

Quantitative proteomics

Samples of 10⁶ spores of S. coelicolor M145 and ∆bfr were grown over a cellophane disk (8 cm in diameter) deposited on the surface of solid LB medium for 3 and 4 days. Mycelium was collected using a sterilized spatula and washed twice with 10.3% sucrose. The entire process was performed with the samples on ice. The pellets were resuspended in 400 µL of protein extraction buffer (0.1 M Tris-HCl pH 7, 5 mM DTT, 6 M urea, 2 M thiourea, 1% CHAPS and protease inhibitor cocktail tablets cOmplete [Roche, REF 11 697 498 001] following manufacturer indications) and vortexed. The proteins were extracted by carrying out 4 cycles of sonication (10 min 30 s ON and 30 s OFF, high intensity) in a Bioruptor^® Standard (Diagenode) sonication system at 4 º C. Four biological replicates were extracted from each strain.

The quantity of proteins from the resultant extracts was observed on SDS-PAGE and measured using the Quick Start Bradford Dye Reagent (BIO-RAD, REF 5000205) assay. The Functional Proteomics facility at the Spanish National Centre for Biotechnology (CNB, CSIC, Madrid, Spain) [54] performed the quantitative proteomic experiment using peptide fractionation and TMT labelling. The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium [55] via the PRIDE [56, 57] partner repository with the dataset identifier PXD050869 and https://doi.org/10.6019/PXD050869.

The proteins identified and quantified with an adjusted p-value < 0.05 were analysed using the StrepDB [58], STRING database [59, 60], and Cytoscape software v 3.10.1 [61]. Protein function was studied using databases such as Google Scholar [62], UniProt [63, 64], NCBI Conserved Domain Database [65, 66], Kyoto Encyclopedia of Genes and Genomes (KEGG) [67, 68], and Database for Annotation, Visualization and Integrated Discovery (DAVID) [69, 70].

Oxidative stress tolerance

Previous assays were adapted to test colony survival after oxidative stress treatment [71, 72]. Samples containing 10⁶ spores of S. coelicolor M145 and ∆bfr were resuspended in germination medium (1.11 g/L CaCl₂, 10 g/L casamino acid, 10 g/L yeast extract) [39] and heat shocked for 10 min at 50 ºC. The spores were then treated with four H₂O₂ concentrations (0, 4, 8, and 20 mM) for 30 min at 30 ºC, washed twice with 10.3% sucrose, resuspended in new germination medium, and incubated for 3 h at 30 ºC. The cells were diluted 1:1000 and plated on LB solid medium. After three days, colony forming units (CFU) were counted and the survival ratio was calculated as the ratio between the number of colonies originating from the treated cells and the control cells. This experiment was carried out in quadruplicate.

Reactive oxygen species measurement

Reactive oxygen species (ROS) were measured from 10 mL-liquid cultures, containing 5 × 10⁵ spores mL^− 1 of both S. coelicolor M145 and ∆bfr strains incubated at 28 ºC under agitation (200 rpm), and on 10 mL of LB and NMMP supplemented with 10 nM, 4 µM and 2 mM of iron. Afterwards, 2’,7’-dichlorofluorescin diacetate (DCFH2-DA, Sigma, REF D6883) was added to each flask to a final concentration of 10 µM and the cultures were incubated in the dark for three hours. The rest of the process from this point was performed in the dark or minimum light conditions. The mycelium was washed twice with 10.3% sucrose, resuspended in 500 µL of 10.3% sucrose, and lysed using 1–1.15 mm glass beads Glasperlen (B. Braun Melsungen Apparatebau, REF 05861/711) in a FastPrep™ FP120 cell disruption system (Savant BIO 101) (4 cycles of 15 s at high intensity). Then, 100 µL of each supernatant were dispensed into a 96-well F-bottom microplate and the fluorescence (λ_excitation = 485 nm, λ_emission = 530 nm) was measured in a Varioskan^® Flash reader (Thermo Scientific). Four biological replicates were measured in triplicate. The fluorescence units were normalized by protein quantification by Bradford assay.

Iron chelators assay

LB medium supplemented with increasing concentrations (0, 50, 100, 200, and 300 µM) of 2,2’-bipyridyl (BIP) (Sigma, REF D216305) and bathophenanthrolinedisulfonic acid (BPS) (Sigma, REF 146617) was dispensed into Sterillin™ 25 well 100 mm plates (Bibby Sterilin Ltd, Stone, Staffs, UK). Colonies of S. coelicolor M145 and ∆bfr containing 5 × 10⁵ spores were cultivated at 30 ºC. A phenotypic analysis was carried out for 7 days and the colonies were cultivated in quadruplicated.

Statistical analysis

All statistical analyses were performed using the GraphPad Prism software v 8.0.1 (GraphPad Software, San Diego, CA, USA). Data are shown as mean ± SEM. The normality of data sets was checked using the Shapiro-Wilk test. Two-way ANOVA (for experiments with two factors such as iron concentration and strain) with Holm-Šídák test for multiple comparisons (following software recommendations), and t-tests (for experiments with only two sets of measurements) were performed. Statistical differences are represented using one (p < 0.05) or two (p < 0.01) asterisks.

The statistical analysis of the quantitative proteomic data was performed by the Functional Proteomics facility at the Spanish National Centre for Biotechnology using Proteome Discoverer software v2.5.

Results

Bfr plays a key role in antibiotic production and differentiation in S. coelicolor M145

The bfr gene (SCO2113) encoding a bacterioferritin (Bfr) was deleted from S. coelicolor M145 using CRISPR-Cas9 technology [48] and the plasmid pCRISPR-Cas9-Bfr. The correct deletion of the bfr gene was verified (see Methods) (Additional File S3).

The phenotypes of the mutant ∆bfr were compared with those of the wild-type (M145) strain on various rich solid media: MSA, LB, R2YE, and YEPD (Fig. 1). No clear phenotypic differences were observed on MSA for either strain. In contrast, the ∆bfr mutant exhibited a strong phenotype on the other three complex media: LB, R2YE, and YEPD. On these media, the following changes were observed: ACT production was almost completely inhibited, RED production was delayed, and cellular differentiation did not result in the formation of aerial mycelium in the mutant strain (Fig. 1A). Furthermore, all these ∆bfr strain phenotypes were complemented by the integrative plasmid pKC796-Bfr on LB and R2YE (Fig. 1B). Therefore, the absence of bacterioferritin was detrimental to antibiotic production and differentiation on these complex media.

Fig. 1

∆ bfr phenotype and complementation. (A) Comparison of antibiotic production and differentiation of S. coelicolor M145 wild-type strain with its ∆bfr mutant growing for 2, 6, and 10 days on several media (MSA, LB, R2YE, and YEPD). (B) Complementation of S. coelicolor ∆bfr mutant with the integrative plasmid pKC796-Bfr after 6 days on LB (left) and R2YE (right) media. The wild type (wt) and ∆bfr carrying the empty plasmid pKC796 were used as controls

Since LB medium presented the clearest and most reproducible phenotypes for the strains, it was also used for liquid growth assays and antibiotic quantification. For ∆bfr, RED (Fig. 2A), and ACT (Fig. 2B) production was severely diminished when grown in LB liquid medium and the differences detected between the wild-type and mutant strains were significant. Strikingly, the decreased antibiotic production observed for the strain lacking Bfr was accompanied by a significantly increased growth rate compared to the wild type (Fig. 2C). In fact, the spores of strain ∆bfr incubated at 28 ºC germinated at 8 h, while in the wild type, the germination tube was not clearly visible until 10 h. In addition, mycelia did not present any phenotypic alterations further along the growth curve (Fig. 2D).

Fig. 2

∆ bfr antibiotic production and growth rate in liquid medium. S. coelicolor M145 wild type (wt) is represented in black and the ∆bfr mutant in grey. Error bars show the standard error of the mean of the experiments carried out in quadruplicate. Asterisks show statistical differences between the two strains at each time point: *: p < 0.05; **: p < 0.01 (two-way ANOVA). (A) RED production in µM/mg at several time points in the growth curves in LB. (B) ACT production in µM/mg at several time points in the growth curves in LB. (C) Growth curves of S. coelicolor M145 and ∆bfr mutant in LB. (D) Germination of spores (red triangles) and mycelia of ∆bfr and wt at different culture times under an optical microscope. Bar: 300 μm

Overexpression of the bfr gene does not affect S. coelicolor M145 development

Since the deletion of bfr resulted in reduced antibiotic production compared to the wild type, the effect of overexpressing the bfr gene was also studied. The bfr overexpression plasmid, pN-Bfr, and the empty vector, pN702GEM3, were transformed in S. coelicolor M145. All transformants obtained for both strains were grown on LB, R2YE, and YEPD media supplemented with neomycin.

The overexpression of bfr on solid media did not significantly affect the final development of S. coelicolor (7 and 10 days) on any of the media assayed, although a slight delay in aerial mycelium formation was observed at 3 days (Additional File S4A). Additionally, less ACT was produced on YEPD at 10 days (Additional File S4 A). Liquid cultures in LB with neomycin were used to quantify ACT and RED production. Spores from S. coelicolor M145 harbouring pN702GEM3 and pN-Bfr plasmids were obtained and used as the inoculum. No statistical differences between the two strains were detected, except that the control strain with the empty plasmid (pN702GEM3) produced more RED than the strain overexpressing bfr (pN-Bfr) at day 3 (Additional File S4B).

Does iron concentration affect antibiotic production and differentiation in S. coelicolor M145 and its mutant strain ∆bfr?

Bfr proteins, including Bfr from S. coelicolor, are implicated in Fe metabolism [17, 28]. Thus, we set out to test whether different iron concentrations could influence antibiotic production and/or differentiation, and if any arising effect would occur differentially between the two strains, i.e., wild type and the mutant. For this purpose, minimal NMMP liquid and solid media with a standard concentration of Fe (4 µM), a low concentration (10 nM), and a high concentration (2 mM), were used for this experiment.

In liquid NMMP medium (Additional File S5), RED and ACT production was quantified every 2 days for 8 days. The results from these liquid cultures are shown in Fig. 3A and Additional File S6. In NMMP with 10 nM iron, RED production was higher in M145, with no statistically significant differences found (Fig. 3A left). Additionally, there was hardly any ACT produced in ∆bfr but a low amount in M145 (about 1.2 µM mg^− 1) (Fig. 3A right). In NMMP with 4 µM iron, RED production was similar in both strains (Fig. 3A left). However, in the mutant, ACT was produced earlier, at day 6, as compared to the wild type, and was even greater at day 8 (Fig. 3A right). Lastly, in NMMP with 2 mM iron, RED production in ∆bfr was approximately half than in M145 (which produces 0.27 µM mg^− 1 at day 4 and 0.21 µM mg^− 1 at day 8). ACT production in the mutant remained constant starting from day 4 (1.58 µM mg^− 1 at day 8); however, M145 presented the highest ACT (5.04 µM mg^− 1 at day 8) production in this iron concentration. As shown in Additional File S6, all differences in the quantifications of antibiotic production were statistically significant according to the ANOVA analysis.

Fig. 3

Effect of iron concentration in NMMP medium cultures of S. coelicolor M145 and ∆ bfr. (A) RED and ACT quantification of liquid cultures in NMMP with an iron concentration of 10 nM, 4 µM, and 2 mM in µM/mg of S. coelicolor M145 (wt) and ∆bfr. (B) The phenotype of colonies (5 × 10⁵ spores) of S. coelicolor M145 (wt) and ∆bfr on solid NMMP with 10 nM, 4 µM and 2 mM of iron. (C) Measurement of halo diameters of the antibiogram against S. epidermidis and M. luteus using plugs of S. coelicolor M145 (wt) and ∆bfr (colonies of 5 × 10⁵ spores) grown on NMMP with 10 nM and 4 µM of iron. Error bars show the standard error of the mean of the experiments carried out in triplicate. Asterisks show statistical differences between the two strains at each time point: *: p < 0.05; **: p < 0.01 (two-way ANOVA)

On solid media, antibiotic production and differentiation of the colonies were observed for 8 days (Fig. 3B). On NMMP with 10 nM iron, both strains produced almost no white aerial mycelium but could produce RED and ACT. On NMMP with 4 µM iron, there were no clear differences between strains, but ∆bfr produced more ACT at day 8 (seen as a bigger blue halo), similar to its increased production detected in liquid medium (in this case at day 6). Finally, on NMMP with 2 mM iron, neither M145 nor ∆bfr produced aerial mycelia and ACT production (blue halo) was not observed. Nonetheless, M145 produced yellow vegetative mycelia, whereas those of ∆bfr remained red.

These results show that iron affects secondary metabolism and development in S. coelicolor M145 and ∆bfr, and, under the tested conditions, this effect is different in solid and liquid media. Although it has been described that low iron concentrations enhance ACT production [7], under our experimental conditions, micromolar iron concentrations accelerated ACT production in ∆bfr, while millimolar concentrations of iron, which enhance antibiotic production in M145 in liquid medium, blocked antibiotic production and differentiation in both strains grown on solid medium.

In addition, the antibiotic activity against three microorganisms (S. epidermidis, M. luteus, and E. coli) of the wild type and the ∆bfr mutant strain grown on solid NMMP medium, with 10 nM, 4 µM and 2 mM of iron, was also tested using an antibiogram assay (Fig. 3C and Additional File S7). No antibiotic activity against E. coli was detected in any of the conditions tested. The activity against the two Gram-positive bacteria was stronger in cells grown on NMMP with 10 nM iron for the two strains. Moreover, ∆bfr grown on NMMP with 4 µM iron had stronger antibiotic activity against M. luteus than the wild type. Neither M145 nor ∆bfr showed antibiotic activity when they were grown on NMMP with 2 mM iron. This experiment showed that iron restriction enhanced the antibiotic activity of S. coelicolor and that this effect was independent of the presence of the bfr gene.

Differences between M145 and ∆bfr strains are not due to changes in intracellular iron concentrations

The phenotypic differences observed in antibiotic production and morphogenesis between the wild type and ∆bfr strain (as mentioned above) could be attributed to changes in the pool of labile iron (iron not incorporated into macromolecules) and/or in total iron concentration (all the iron inside the cell, free or associated with macromolecules) due to the absence of the Bfr storage protein. Therefore, both labile and total intracellular iron concentrations were measured in liquid cultures of both strains grown in 10 nM, 4 µM and 2 mM NMMP media.

Intracellular iron concentrations from cultures of both strains in NMMP with 10 nM iron were too low to obtain any reliable measurements. There were no statistically significant differences between the strains in the other two media (4 µM and 2 mM). The ∆bfr mutant had a higher labile iron content on days 4 and 6 in NMMP with 4 µM iron, and the total iron levels were similar to labile iron in this medium (values about 0.1–0.35 nmol mg^− 1) (Fig. 4A). However, iron levels were much higher in NMMP with 2 mM iron and total iron concentration was about twice as high as the labile iron concentration (Fig. 4B).

Fig. 4

Iron levels in liquid cultures of S. coelicolor M145 and ∆ bfr. Labile (left) and total (right) iron quantification in NMMP with 4 µM iron (A), NMMP with 2 mM iron (B), and LB (C) in nmol/mg. Error bars show the standard error of the mean of the experiments carried out in quadruplicate. Two-way ANOVA was performed for statistical analysis

This experiment was also conducted in LB medium (which contains about 17 µM iron [73]), where the ∆bfr phenotypes were more notable. As growth in rich media is faster for both strains, we measured the iron concentrations on days 2, 3, 4, and 5. The labile and total iron levels (about 0.25–0.5 nmol mg^− 1) were comparable to those obtained in NMMP with 4 µM iron and there were no statistically significant differences in iron levels between the two strains in LB (Fig. 4C).

In conclusion, despite the iron storage capacity of Bfr, the lack of this protein, in S. coelicolor, does not alter the levels of total or labile intracellular iron.

Different iron chelators alter development in S. coelicolor M145 and ∆bfr

Lee et al. reported a similar effect on ACT activation by reducing the concentration of iron in the culture medium and by using BIP as an iron chelator [74]. We aimed to investigate whether iron chelators could affect S. coelicolor M145 and ∆bfr when grown on LB, the complex medium used for antibiotic quantification. Two iron chelators were employed: BIP, which is permeable and sequesters both extracellular and intracellular iron, and BPS, which cannot enter the cell and only sequesters extracellular iron [75].

The results obtained for BIP are shown in Fig. 5A. In the wild type M145, ACT production decreased as the BIP concentration in the medium increased. Differentiation was slightly delayed with 100 µM BIP, and aerial mycelia formation was blocked with 200 µM. Differentiation and antibiotic production were abolished with 300 µM BIP. In the ∆bfr mutant, some aerial mycelia were induced with the BIP treatment (higher concentration equates to less aerial mycelium formation). Normal red colouration was delayed with 200 µM BIP and, at the highest concentration tested (300 µM), the mutant phenotype was similar to that of the wild type, with no differentiation or antibiotic production.

As shown in Fig. 5B, BPS had a stronger negative effect on differentiation for M145 at 3 and 4 days. BPS delayed aerial mycelium formation, and ACT production was slightly lower with 100 µM BPS, although this could be a consequence of the delay in differentiation. The two highest concentrations (200 and 300 µM) blocked white aerial hyphae formation, and M145 produced a vegetative mycelium with a dark border, with ACT production and a darker colour in 200 µM BPS. For the ∆bfr mutant, some aerial mycelium and ACT production were observed at days 6 and 7 with 50 µM BPS, but for the rest of the concentrations tested, the phenotype was quite similar: red colonies without aerial mycelium.

This experiment demonstrated that, on LB medium, the chelation of both intracellular and extracellular iron by BIP affects antibiotic production and differentiation in both the wild type and ∆bfr mutant, while the chelation of extracellular iron by BPS primarily affects aerial mycelium formation with no major changes in the mutant.

Fig. 5

Effect of iron chelators on differentiation and antibiotic production in S. coelicolor M145 and ∆ bfr. The phenotype of colonies (5 × 10⁵ spores) of M145 and ∆bfr on LB supplemented with increasing concentrations (0, 50, 100, 200, and 300 µM) of BIP (A) and BPS (B) on different days

Role of bacterioferritin in oxidative stress

The absence of bacterioferritin in the cell could also lead to a compromise in oxidative stress tolerance, as has been observed previously in other microorganisms [71, 72]. Therefore, a survival assay was conducted by treating the spores of both strains with various concentrations of H₂O₂ (0, 4, 8, and 20 mM) and counting the colony forming units (CFUs) formed on YEPD medium. Surprisingly, the ∆bfr strain had a better survival rate than the wild type after treatment with 4and 8 mM H₂O₂ (Fig. 6A). Indeed, 2–3 CFUs could be observed on plates inoculated with ∆bfr spores treated with 20 mM H₂O₂, whereas at this concentration, M145 could not survive (Fig. 6A).

Fig. 6

Role of bfr gene in oxidative stress. (A) The survival rate of S. coelicolor M145 (wt) and ∆bfr after treatment with H₂O₂, calculated as the ratio of colony forming units (CFUs) per the number of untreated cells. (B) ROS measurement in NMMP with 4 µM and 2 mM iron at day 8. (C) ROS measurement in NMMP with 4 µM iron at days 6 and 8. (D) ROS measurement in LB at different days. ROS measurements are in AU/mg protein and the phenotypes of the cultures are shown in the upper part of the graphs. Error bars show the standard error of the mean of the experiments carried out in quadruplicate. Asterisks show statistical differences: *: p < 0.05; **: p < 0.01 (two-way ANOVA)

Bfr influence reactive oxygen species concentration in S. coelicolor M145

As previously mentioned and contrary to expectations, S. coelicolor ∆bfr demonstrated greater tolerance to oxidative stress, with its intracellular iron levels being akin to those of the wild type. The increased tolerance of ∆bfr to H₂O₂ indicated this protein may be associated with oxidative stress. Furthermore, the link between antibiotic production and oxidative stress had been studied before in S. coelicolor [76]. Our hypothesis was that the phenotypic differences observed between the M145 and ∆bfr strains might be related to reactive oxygen species (ROS) production.

To test this hypothesis, ROS production was measured by using 10 µM DCFH₂-DA in both strains grown in liquid NMMP with 10 nM, 4 µM, and 2 mM iron. ROS were quantified at the final time point (8 d) for all three media, although ROS concentration could not be measured in NMMP with 10 nM iron (Fig. 6B). Quantification was also performed on day 6 in NMMP with 4 µM iron (Fig. 6C) when ∆bfr accelerated ACT production (Fig. 3A).

The results showed that in NMMP with 4 µM iron, ROS concentration was similar in both strains on day 6 (when ∆bfr accelerates ACT production), but at 8 days, when both strains exhibit similar ACT production, it decreased in M145 and significantly increased in ∆bfr (Fig. 6C). ROS concentration in ∆bfr was also higher than in the wild type in NMMP with 2 mM iron on day 8 when both strains produce ACT, but M145 produces more antibiotic (Fig. 6B).

ROS production was also measured in LB medium, on day 3, when both strains have similar ACT production, and on day 4, when ACT synthetized by M145 is greater than in ∆bfr. As shown in Fig. 6D, ROS concentration was similar on day 3, but higher in ∆bfr on day 4.

These experiments indicated that the absence of Bfr led to an increase in ROS concentration in S. coelicolor which could be related to reduced ACT production.

The deletion of the bacterioferritin gene changes the proteome in S. coelicolor M145

As Bfr is not, in principle, a gene expression regulator, the phenotypes observed with the bfr gene deletion might be a consequence of changes in S. coelicolor metabolism. Therefore, to explore how the absence of Bfr triggers the observed cellular behaviour, a comparative proteomic analysis of the wild type and the ∆brf mutant was performed. The assay was conducted on cells obtained from cultures at 3 and 4 days on solid LB medium, where differences between M145 and ∆bfr were more remarkable (Additional File S8 A). Protein extraction was done as described in the Methods section.

The analysis performed by the Functional Proteomics facility of the Spanish National Centre for Biotechnology (CNB) identified 3521 proteins (approximately 43% of the proteome). The samples showed clear differential clustering for each strain at each culture time using principal component analysis (PCA) (Additional File S9). By setting the threshold to an adjusted p-value < 0.05, 204 proteins were differentially represented in ∆bfr compared to M145 at 3 days (90 proteins were more abundant in ∆bfr and 114 proteins were less abundant) and 159 proteins at 4 days (51 proteins were more abundant in ∆bfr and 108 proteins were less abundant). Additionally, 125 of the differentially represented proteins were detected on both days (Fig. 7A and Additional File S8B).

All differentially represented proteins detected in this study are listed in Additional File S13 (day 3) and Additional File S14 (day 4).

To study how the lack of Bfr could change metabolism, all differentially represented proteins were grouped according to their functions (Fig. 7B). The main results obtained for each functional category are detailed below.

Fig. 7

Proteomic study of S. coelicolor ∆ bfr versus M145 on LB medium. (A) Venn diagram showing the total number of differentially represented identified proteins between the wild-type and mutant strains at days 3 and 4 of growth culture (adjusted p-value < 0.05). (B) Graphical summary of the number of proteins identified (with adjusted p-value < 0.05) grouped according to their assigned or predicted function

Secondary metabolism

One of the most evident characteristics of the ∆bfr mutant is a delay in ACT and RED production on LB medium. We detected 15 of the 22 proteins from act cluster (SCO5071-5092) that were less abundant in ∆bfr compared to M145. These included proteins responsible for ACT biosynthesis and the regulator ActVI-ORFA (SCO5071) [77]. From the red cluster, we only detected four proteins that, contrary to expectations, were more abundant in ∆bfr. RedS (SCO5585), whose function is unknown, was detected on both days, while three proteins involved in biosynthesis, RedP (SCO5888), RedQ (SCO5887) [78], and RedF (SCO5898) [79] only passed the threshold (p < 0.05) on day three (Fig. 8A).

Fourteen proteins from the calcium-dependent antibiotic cluster (CDA) (SCO3210-SCO3249), involved in its antibiotic biosynthesis, were more abundant in ∆bfr. A CDA bioassay confirmed that ∆bfr produced more CDA than the wild-type strain (Fig. 8B).

Other molecules produced by S. coelicolor include yellow coelimycins (yCPKs) P1 and P2 and the antibiotic coelimycin A (abCPK) [80]. Although the ScF (SCO6272) protein, which converts yCPKs to abCPK [50], was more abundant in ∆bfr than in the wild type, four biosynthetic proteins were less abundant in ∆bfr. Lower levels of yellow coelimycins were detected in ∆bfr when both strains were grown on NA medium supplemented with glutamate (Fig. 8C).

Proteins involved in the synthesis of other secondary metabolites also showed changes in abundance. Nine proteins from the cryptic deoxysugar cluster (SCO0381-SCO0401), homologous to the ste cluster found in Streptomyces sp. 139 and thought to be involved in the biosynthesis of ebosin (an exopolysaccharide) [81, 82] were less abundant in ∆bfr. Other less abundant proteins in the mutant included EctC (SCO1866), involved in the biosynthesis of ectoine, which is implicated in osmotic stress [83], and geosmin synthase (SCO6073) [84]. However, ScbB (SCO6264), involved in γ-butirolactones synthesis [85], was more abundant in the mutant (Additional Files S13 and S14).

Fig. 8

Proteins linked to secondary metabolism detected in the proteomic study. Proteins marked with * were only detected at day 3, and those marked with # were only detected at day 4. Represented ratios are the mean for proteins detected at both days. (A) Abundance ratios of proteins from act and red cluster. (B) Abundance ratios of proteins from cda cluster and CDA production assay (measurement of inhibition halos against B. subtilis). Error bars show the standard error of the mean of the experiments carried out in triplicate. Asterisks in the graph show statistical differences between the two strains: *: p < 0.05; **: p < 0.01 (t-test). (C) Abundance ratios of proteins from cpk cluster and yCPKs production assay

Proteins involved in antibiotic regulation

bfr is a direct target of the response regulator AbrC3, which activates antibiotic production [33] and alters its expression in mutants lacking other components of regulatory TCSs related to antibiotic production [32, 34]. Therefore, we also examined which proteins involved in antibiotic regulation altered their abundance in the ∆bfr mutant.

The ecr (expression coordinated with red) cluster (SCO2517-2519) includes a TCS (EcrA2/SCO2518 and EcrA1/SCO2517) that activates RED [86], and EcrB (SCO2519), a paralogue of the ACT efflux pump ActB/ActII-ORF3 (SCO5084) [87]. Two of these three proteins were detected in higher abundance in the mutant: EcrA2, the HK of the system, and EcrB (Fig. 9A), which is consistent with the higher abundance of some of the proteins from the red cluster in the mutant.

Another HK more abundant in the mutant is MtrB (SCO3012) (Fig. 9A). The MtrA-MtrB-LpqB system [88] activates the RR MtrA, which in turn activates aerial mycelium formation [89] and represses antibiotic production regulators [90].

The abe (antibiotic enhancement upon overexpression) cluster (SCO3287-3290) comprises four genes that may function as a signalling transduction system and enhancers of ACT production [91]. Three proteins from this cluster had lower abundance in the ∆bfr mutant: AbeB (SCO3288) and AbeC (SCO3289) are membrane proteins, and AbeD (SCO3290) has a cyclic nucleotide-binding domain (Fig. 9A).

The role of XRE/DUF397 proteins in antibiotic production in S. coelicolor has recently been studied [92]. In our analysis, three XRE proteins, SCO4678, SCO5125, and SCO6629 were less abundant in the ∆bfr mutant (Fig. 9A).

Additionally, two regulatory proteins were less abundant in the ∆bfr mutant: NsdB (SCO7252) and EshA (SCO7699). NsdB (negatively affecting Streptomyces physiological differentiation) represses antibiotic production in some culture media [93], and EshA is required to keep ppGpp levels high enough to enhance ACT production and aerial mycelium in certain culture media [94] (Fig. 9A).

The ∆bfr mutant also exhibited changes in proteins from uncharacterised TCSs that could regulate antibiotic production or other processes (see Additional Files S13 and S14). Moreover, several proteins from the so-called “conservons” have been assigned regulatory roles, as explained below.

Conservons

cvn operons are gene clusters with a conserved structure. These genes encode proteins that could function as eukaryotic G signalling proteins: a histidine kinase-like protein (CvnA), two hypothetical proteins (CvnB and CvnC), and an ATP or GTP-binding protein (CvnD). S. coelicolor possesses 13 conservons, four of which include an additional gene encoding a cytochrome P450 protein (CvnE) [95].

From the first conservon, cvn1 (SCO5541-SCO5544), CvnA1 (SCO5544), and CvnB1 (SCO5543) were more abundant in the ∆bfr mutant. This conservon is involved in the coordination between glucose concentrations and development by regulating the sigma factor SigU, which represses differentiation and antibiotic production [96] (Fig. 9B).

The cvn7 operon is involved in the interaction of S. coelicolor with other microorganisms. This cluster includes an additional gene, cvnF7 (SCO6798), which encodes a cyclic nucleotide-binding protein, and it is homologous to cvnF8 (sco6939), which is involved in secondary metabolism and cryptic cluster regulation [97]. CvnF7 was less abundant in the ∆bfr mutant (Fig. 9B).

The cvn9 operon (SCO1626-1630), also known as rar (restoration ofaerial mycelium in amfRmutant), encodes proteins characterized as repressors of differentiation and secondary metabolism under normal conditions [95]. RarA/CvnA9 (SCO1630), RarB/CvnB9 (SCO1629), RarC/CvnC9 (SCO1628), and RarE/CvnE9 (SCO1626) were less abundant in the ∆bfr mutant compared to the wild type (Fig. 9B).

Finally, three proteins from an uncharacterised conservon, cvn11 (SCO0584-0588), were also less abundant in the ∆bfr mutant: CvnA11 (SCO0588), CvnD11 (SCO0585, only detected on day 4), and CvnE11 (SCO0584) (Fig. 9B). CvnA11 has been described as a modulatory protein for several genes [98] and a regulator of nitrogen metabolism [99].

Fig. 9

Proteins linked to secondary metabolism regulation detected in the proteomic study. Proteins marked with * were only detected at day 3, and those marked with # were only detected at day 4. Represented ratios are the mean for proteins detected at both days. (A) Abundance ratios of other proteins which regulate secondary metabolism. (B) Abundance ratios of proteins from conservons

Oxidative stress and respiration

As mentioned earlier, the ∆bfr mutant demonstrated better tolerance to oxidative stress (Fig. 6A) and had a higher intracellular ROS concentration (Fig. 6B, C and D). Consequently, proteins involved in oxidative stress were analysed to determine the presence of a stress response or a related pathway highly represented in the mutant.

Since the levels of ACT biosynthetic proteins were lower in the mutant on LB, resulting in reduced ACT production, the response to redox metabolites, such as ACT and its precursors, was also diminished in the mutant. SoxR (SCO1697) activates the expression of five genes in response to ACT or its derivatives to protect against side effects [100, 101]. Four of the five proteins encoded by these genes were less abundant in the mutant on day 3 (Additional File S10A): EcaA (SCO7008), an ABC transporter; EcaB (SCO1909), a monooxygenase which could modify ACT; EcaD (SCO4266), an NADPH-dependent quinone reductase positively regulated by ActVI-ORFA (also less abundant in ∆bfr, see “Secondary metabolism” section) [77]; and SCO2478, an NADPH-dependent flavin reductase.

Furthermore, all detected proteins related to oxidative respiration were more abundant in ∆bfr (Additional File S10B): bcc-aa₃ supercomplex subunits QcrB (SCO2148), CtaE (SCO2151) and CtaC (SCO2156) [102]; the cytochrome maturation protein ResA (SCO4472); respiratory chain complex I subunits NuoH (SCO4569), NuoL (SCO4573) and NuoM (SCO4574); and ATPase subunits AtpI (SCO5366), AtpB (SCO5367) and AtpE (SCO5368) [76, 103]. Moreover, GabD (SCO7035), a succinate-aldehyde dehydrogenase linked to NADH production in S. coelicolor [76], was also more abundant in the mutant. This could result in higher O₂ consumption, promoting the faster growth observed in the ∆bfr mutant and early hypoxia.

In this context, it is known that the two-component system OsdK/DevS-OsdR/DevR (SCO0203-0204) activates a dormancy response under hypoxic conditions [104]. Two proteins activated by OsdR were more abundant in the ∆bfr mutant (Additional File S10 A): SCO2367, a secreted serine peptidase [104], and SCO6164, a paralogous of DksA from E. coli [105], a transcription factor which modulates RNA polymerase by ppGpp levels [106, 107] (Additional File S10 A).

Tdd8 (SCO2368) is a protein involved in calcium homeostasis that represses genes involved in oxidative stress response and dormancy [108]. Although Tdd8 was not detected in our analysis, some proteins repressed (directly or indirectly) by Tdd8 were more abundant in the mutant compared to the wild type (Additional File S10 A). Two of these proteins repressed by Tdd8 are involved in nitrogen metabolism and are positively regulated by OsdR. These proteins are NarG2 (SCO0216) and NarI2 (SCO0219), the catalytic and the γ subunits, respectively, of the nitrate reductase Nar2, the main nitrate reductase of S. coelicolor, [109]. By contrast, NarH1 (SCO6534), the catalytic subunit of Nar1, was detected on both days but was less abundant in the mutant. Nar1 is the nitrate reductase which reduces nitrate in spores [110]. Endogenous nitric oxide in S. coelicolor has been linked to RED and ACT production, under certain levels, and a delay in differentiation [111, 112].

Another two proteins repressed by Tdd8 and overrepresented in the proteomic assay were SCO168 (a cyclic nucleotide-binding protein from family CRP/FNR) and SCO0169 (a putative cystathionine β-synthase) [108]. These proteins are related to oxidative and nitrosative stress protection because cystathionine is an intermediate in cysteine biosynthesis, which is a precursor of mycothiol, involved in these types of stress [113, 114]. Moreover, another protein, SCO2763, linked to mycothiol, was detected on day 3 and was more abundant in the mutant. SCO2763 is an ABC transporter activated by SigR in oxidative stress from thiol groups, transporting mycothiol conjugates [115].

Two more proteins directly related to oxidative stress were detected on both days (Additional File S10A). MetQ (SCO1557), a lipoprotein more abundant in the ∆bfr mutant than in the wild type, is involved in methionine import together with MetI and MetN. The three genes encoding these proteins are overexpressed in oxidative stress [24]. Conversely, AhpC (SCO5032) was less abundant in the mutant. This protein is an alkyl hydroperoxide reductase induced by OxyR in oxidative stress, acting in defence against endogenous hydrogen peroxide [116].

Regarding other proteins less abundant in ∆bfr compared to the wild type, we found two membrane proteins with thioredoxin domains (Additional File S10A); thioredoxins are involved in various oxidative stress responses. These proteins are SCO2035, which has been described as more abundant in multinucleated mycelia [117], and SCO2989 which has not been studied and only surpasses the p < 0.05 threshold on day 3.

Cellular envelope stress and transporters

The cellular envelope is another significant source of stress. The ECF sigma factor SigE controls the expression of genes that encode membrane proteins in response to this stress [118]. Three proteins whose genes are regulated by SigE were more abundant in the mutant: SCO4134, a putative oxidoreductase similar to Pseudomonas putida 4-cresol dehydrogenase, SCO4471, a lipoprotein involved in antibiotic and lysozyme resistance, and SCO5213, a metallopeptidase [118] (Additional File S11A).

Two proteins from an ABC transporter associated with stress from antibiotics affecting the cell wall were less abundant in the ∆bfr mutant: SCO3110 (only detected on day 4) and SCO3111 [119] (Additional File S11A).

Many membrane proteins with transport functions (predicted or described) had altered abundance levels in ∆bfr (Additional File S11A). Two transporters were more abundant in the mutant and were detected on both days: SCO1655, an oligopeptide transporter [120], and KdpB (SCO3717), a subunit of a potassium importer [121]. On day 3, a putative metal and/or siderophore transporter, SCO0996 [122], was less abundant in the mutant.

Osmotic stress

As mentioned in the “Secondary metabolism” section, one of the proteins required for ectoine production in osmotic stress (EctC/SCO1866) [83] was less abundant in ∆bfr, similar to DpsA, which is known to play a role in osmotic stress [21, 123] (Additional File S11B). Osmotic stress in S. coelicolor is mainly regulated by SigB [124]. SigB was not detected in this proteomics analysis, but its anti-sigma factor, RsbA (SCO0599), which releases SigB in osmotic stress [125], was less abundant in the mutant. This could indicate that SigB is more active in ∆bfr than in the wild type, but we did not observe an active osmotic stress response. The abundance of proteins regulated by SigB was variable: SCO3343, a protein with a GsmA (sporangiosporematuration cell wall hydrolase) domain [126], and SCO6014, a putative amino acids transporter [127], were more abundant in the mutant, while SCO5207, a putative regulatory protein from cystathionine beta-synthase family [126], was less abundant.

Another protein with a role in osmotic stress, SmpA (SCO2529), was less abundant in the ∆bfr mutant (Additional File S11B). SmpA is a metalloprotease that activates the catalase CatB, required for osmotic protection and differentiation [128].

Differentiation and development

Another important phenotype shown by the ∆bfr mutant was delayed differentiation on LB (Fig. 1). In this way and as expected, some proteins involved in Streptomyces development were less abundant in ∆bfr (Fig. 10). As Table 1 shows, these proteins were: DpsA (SCO0596), SCO1756, RpfC (SCO3098), BldN (SCO3323), RsbN (SCO3324), BdtA (SCO3328), RsfA (SCO4677), BldM/WhiK (SCO4768), SCO4920, SCO5029, EshB (SCO5249), FilP (SCO5396), BldB (SCO5723), WhiH (SCO5819), and CabC (SCO7647).

Table 1 Proteins linked to development detected in the proteomics study of S. Coelicolor ∆bfr versus M145

Regarding proteins more abundant in the ∆bfr mutant compared to the wild type, we only detected Cis1 (SCO4242) on day 4, which is involved in programmed cell death [134] (Fig. 10).

Fig. 10

Proteins linked to development detected in the proteomic study of S. coelicolor ∆ bfr versus M145. Abundance ratios of proteins. Proteins marked with * were only detected at day 3, and those marked with # were only detected at day 4. Represented ratios are the mean for proteins detected at both days

Primary metabolism

Proteins from glycolysis and the Krebs cycle were not detected in the proteomics study. Four proteins related to carbohydrate transport and degradation were more abundant in ∆bfr (Additional File S12A): XylF (SCO6009) and BxlE1 (SCO7028), which are involved in the binding and import of xylose [142] and xylobiose [143], respectively, and CsnB (SCO2024) and DagB (SCO3487), a chitosanase [144] and an agarase [145], respectively.

AccB (SCO5535), an essential component from acetyl-CoA carboxylase that produces malonyl-CoA for fatty acids biosynthesis [146], was less abundant in ∆bfr (Additional File S12A). FabH (SCO2388), a β-ketoacyl ACP synthase involved in fatty acids biosynthesis [147], was also less abundant in the mutant but was only detected above the threshold (p < 0.05) on day 4. Two proteins related to cholesterol metabolism were more abundant in ∆bfr: SCO4781 (only detected on day 4), homologous to Mycobacterium tuberculosis ChoD (cholesterol oxidase) [148], and SCO5420, a putative cholesterol esterase that can degrade detergents to use them as carbon source [149].

Two proteins involved in amino acids biosynthesis were less abundant in the mutant (Additional File S12A): SerA (SCO5515, only detected on day 3), for serine biosynthesis [150], and ArgG (SCO7036), for arginine biosynthesis [151].

Ribosome and DNA

Eight ribosomal proteins, seven from 30 S subunit (RpsD/SCO1505, RpsT/SCO2563, RpsR1/SCO3908, RpsL/SCO4659, RpsN1/SCO4715, RpsI/SCO4735, and RpsP/SCO5591) and one from 50 S subunit (RpmB1/SCO5564), were less abundant in the mutant (Additional File S12B). This suggests that ∆bfr could be less efficient during translation.

A few proteins related to DNA processing were also detected (Additional File S12B). AdnA (SCO5183), involved in DNA recombination [152], was one of the most abundant proteins in the ∆bfr mutant. Another protein more abundant in ∆bfr was SCO3619, which belongs to YbaB/EbfC family of nucleoid-associated proteins and could have a role in DNA repair [153].

Finally, LexA (SCO5803), which represses the SOS response under normal conditions [154], was less abundant in the mutant on day 3 (Additional File S12B).

Discussion

Iron as a signal in antibiotic biosynthesis and development in S. coelicolor M145

Antibiotic regulation in Streptomyces is a complex network of regulatory proteins linked to environmental signals [4]. Iron is one of these key signals, and the relationship of iron, proteins related to iron metabolism, and secondary metabolism in Streptomyces has been well-documented [5, 6, 29, 74, 155]. Previous studies on S. coelicolor’s TCSs, which play a major role in cellular signalling, have validated the relationship between these regulatory systems and the bfr gene, which encodes for a bacterioferritin [34, 35, 155]. In other actinomycetes, the link between iron homeostasis and secondary metabolism has also been described, as in S. avermitillis, where the DtxR-family regulator IdeR is required for the proper production of avermectins and activates, among other genes, the expression of bfrA to store excess iron [30]. On the other hand, Bfr from Saccharopolyspora pogona plays a role in iron storage and oxidative stress tolerance, and enhances butenyl-spinosyn production [71].

In our study, the absence of the bacterioferritin in S. coelicolor caused a delay in differentiation and antibiotic production on complex media (Figs. 1 and 2). In the minimal medium NMMP, where iron concentrations can be controlled, ACT and RED production differed between liquid and solid media. Lee et al. reported that low iron conditions increased ACT production on CYE medium [74]. In our study, low iron conditions in solid NMMP enhanced antibiotic production (Fig. 3B), while high iron concentrations inhibited development on solid medium and the production of both antibiotics. However, in liquid NMMP with high iron concentrations, cultures of the wild-type strain M145 enhanced ACT and RED production but not the mutant (Fig. 3A). Liquid non-sporulating cultures exhibit greater expression of genes linked to oxidative stress, among others [156], so antibiotic production in liquid cultures could be linked to increased stress and, possibly, to the differential expression of some regulatory genes.

Growth assays in the presence of the iron chelators BIP and BPS demonstrated that both extracellular and intracellular iron play a role in S. coelicolor wild-type development and antibiotic production (Fig. 5). The lack of significant changes in the ∆bfr mutant, regardless of the presence of the chelator, could be attributed to a possible role of Bfr as an iron donor for regulatory proteins containing Fe-S clusters (Fig. 11). Since the absence of Bfr negatively affected the iron levels required for regulatory proteins, they would remain inactive whether iron was chelated or not (see below).

Fig. 11

Proposed model of the role of Bfr in S. coelicolor M145. (A) In the presence of Bfr, proteins with Fe-S clusters can receive iron and participate in the regulation of other proteins. Secondary metabolites (ACT, CPKs, etc.) are produced under normal conditions. ROS are produced at standard levels. (B) In the absence of Bfr, regulatory proteins with Fe-S clusters do not receive enough iron, so differentiation and secondary metabolism are impaired. More ROS are produced and S. coelicolor activates oxidative respiration, growth, programmed cell death (with CDA biosynthesis), hypoxia, and stress responses

The presence of only one bacterioferritin in S. coelicolor M145 and its role in iron storage

While in S. coelicolor the bacterioferritin Bfr, the subject of this study, is the only putative canonical iron storage protein, other bacteria, such as E. coli, Campylobacter jejuni, Pseudomonas aeruginosa, and M. tuberculosis, possess two types of these proteins, bacterioferritin and ferritin, each with a specific role in each organism. In M. tuberculosis, which belongs to the phylum Actinomycetota, like Streptomyces, there is one bacterioferritin (BfrA), RV1876, which is 62% identical to the one studied in this work, and one ferritin (Ftn, previously named “BfrB”), RV3841, which does not have a homologue in S. coelicolor. Both proteins protect M. tuberculosis from oxidative stress. However, Ftn has a stronger iron storage capacity and plays a role in higher iron concentrations. By contrast, BfrA acts under low iron conditions to maintain basal iron levels. Only the expression of Ftn is induced with iron by IdeR, the homologue of DmdR1 in Mycobacterium [16]. Interestingly, in Mycobacterium abscessus subsp. massiliense, an emerging pathogen that causes non-tuberculosis diseases, two genes for ferritins have been described, although there are no genes present for bacterioferritin [157]. In P. aeruginosa, the bacterioferritin (BfrB) that coexists with a ferritin (FtnA, previously named “BfrA”) is responsible for iron accumulation and requires specific interaction with a bacterioferritin-associated ferredoxin (Bfd) to mobilise iron. In P. aeruginosa, FtnA seems to provide iron to the catalase KatA [27]. The interaction with Bfd also occurs in Erwinia chrysanthemi [158]. Therefore, there are specific roles for each type in each microorganism; for example, in E. coli the most important ferritin is FtnA [27].

The deletion of iron storage proteins in other microorganisms or the disruption of iron release from these proteins changes intracellular iron concentrations [27]. In P. aeruginosa, the lack of BfrB (the main iron storage protein) reduces total iron levels, prompting the cell to decrease iron uptake. Conversely, inhibiting the BfrB-Bfd interaction results in an irreversible flux of iron towards BfrB, thereby decreasing the labile iron concentration [27]. However, in our study, which tested NMMP with three different iron concentrations and LB, there were no differences in total and labile iron concentrations between the ∆bfr mutant of S. coelicolor and the M145 strain (Fig. 4).

Regarding the putative transcriptional regulation of bacterioferritins, the promoter region of bfrA from S. avermitilis contains iron boxes (GCAGGACACGGATCGTCATA) and is activated by IdeR, the homologue of DmdR1 and DmdR2 from S. coelicolor [30]. However, no iron boxes have been detected in the promoter of the bfr gene of S. coelicolor [159]. The absence of these boxes in the S. coelicolor bfr promoter and the lack of differences in iron concentrations between M145 and ∆bfr mutant could indicate that, in S. coelicolor, Bfr might have other functions beyond iron storage. Dps proteins (DpsA, DpsB, and DpsC in S. coelicolor), which also possess iron storage capacity [23], could be compensating (among others) for the lack of Bfr, as described in other organisms [158] (Fig. 11B).

The role of bacterioferritin in S. coelicolor M145 in oxidative stress and iron loading to regulatory proteins

The absence of iron storage proteins in other organisms causes susceptibility to oxidative stress since ROS derived from the Fenton reaction cause damage to cellular structures (71, 72). Here, we report the contrary response from S. coelicolor M145: the ∆bfr mutant shows increased tolerance to oxidative stress (Fig. 6A), even when intracellular ROS concentration is higher in the mutant in both complex (LB medium, Fig. 6D) and minimal (NMMP, Fig. 6B and C) media.

These results lead us to consider that Bfr from S. coelicolor could act as a modulator of oxidative stress, storing iron inside its cavity to prevent Fenton reaction and maintaining normal oxidative stress levels. In this way, Bfr from Desulfovibrio vulgaris acts as a protective system against ROS and is activated by the oxidative stress regulator PerR, while Ftn does not play a role in ROS protection [160]. Moreover, like Ftn from P. aeruginosa, which provides iron to another protein [27, 161], Bfr from S. coelicolor could play a similar role. Parallel functions have been linked to other ferritins: in E. coli, FtnA can store iron from damaged Fe-S clusters and this iron is transferred to the protein IscA (with the thioredoxin reductase system), which transfers it to the scaffold protein IscU from the Isc system, for Fe-S biosynthesis [162]. In E. chrysanthemi, iron from siderophores is captured by the Suf system, which synthesises Fe-S clusters and incorporates this iron into Bfr [158]. One of the ferritins of Salmonella enterica sv. Typhimurium, FtnB, can use stored iron to repair damaged Fe-S clusters [163]. Thus, in S. coelicolor, Bfr could be transferring iron to regulatory proteins involved in differentiation, antibiotic production, and other processes (Fig. 11A). Several regulatory proteins have Fe-S clusters or are iron coupled, like the Wbl family, involved in differentiation [164], and proteins involved in redox sensing and resistance, such as CatR [165], which is also involved in iron homeostasis [166]. The source of iron for Fe-S clusters assembly is still unknown [167, 168]. In S. coelicolor, the contiguous gene, SCO2114, codes for a putative ferredoxin Bfd, which could help Bfr to reduce stored iron and transfer it to regulatory proteins. According to our proposed role of Bfr, the Fe-S or iron-containing proteins would be inactive in the ∆bfr mutant, and the proteins regulated by these iron-containing upstream regulators (proteins from conservons, some TCSs, metabolism regulators, secreted proteins, etc.) would change their abundance and produce the phenotype seen in the mutant (Fig. 11B).

On the other hand, oxidative stress related to respiration regulates antibiotic production in S. coelicolor, and these antibiotics act as modulators of programmed cell death (RED, CDA) and antioxidants (ACT) [76, 169]. In this context, the accelerated production of ACT in NMMP with 4 µM iron in the ∆bfr mutant could be produced by higher ROS production, which triggers ACT to neutralise them, leading to concentrations similar to those found in the wild type on day 6. However, as Bfr is absent, the concentrations of ROS continue to increase in the mutant on day 8 (Fig. 6B). In solid NMMP with 2 mM iron, iron could produce an excess of ROS, above a certain threshold, which blocks differentiation and secondary metabolism. By contrast, in liquid NMMP, this iron could be less available, and/or cells are more tolerant to stress due to the greater expression of genes linked to oxidative stress [156], and iron enhances ACT production in M145, but produces too much oxidative stress in ∆bfr (Fig. 6C).

The proteomics study performed on LB medium, where ∆bfr mutant also showed higher ROS concentration, did not show a general oxidative stress response enhanced due to the lack of Bfr. Instead, proteins related to hypoxia [104] and activated in oxidative stress [108] (Additional File S10A), oxidative respiration (Additional File S10B), and secondary metabolites (CDA) related to programmed cell death (Fig. 8B) [169] were more abundant (Fig. 11). These mechanisms could be a response to a higher ROS concentration. Thus, the ∆bfr mutant would try to consume more oxygen, causing a higher growth rate, but also activating dormancy and stress responses. This response could not be enough to permanently decrease ROS concentration, so high ROS levels, and/or the lack of functional iron clusters inside regulatory proteins (as explained before), would be blocking downstream regulatory proteins, differentiation, and some metabolic processes, such as fatty acid biosynthesis and some secondary metabolites production (deoxysugar, coelymicins, etc.). Despite an altered development and the hypothetical absence of these functional regulatory proteins, the cellular response of the mutant would allow it to survive in higher oxidative stress conditions without significant changes in iron concentrations.

ACT production could also be reduced by the absence of SoxR in its active form: SoxR is a [2Fe-2 S] protein [170], which activates proteins for ACT detoxification [100, 101], and these proteins were less abundant in the ∆bfr mutant on day 3 (Additional File S10 A), which could indicate a decrease in SoxR activity.

These assumptions regarding the role of Bfr in S. coelicolor should be further studied by checking whether Bfr can interact with Fe-S clusters containing regulatory proteins or other iron-associated proteins. Additionally, it would be interesting to study the proteomes of S. coelicolor M145 and ∆bfr in minimal media with different iron concentrations, with a special focus on the accelerated production of ACT with 4 µM iron and the differences seen between liquid and solid NMMP with 2 mM iron. The molecular mechanism of the role of Bfr in S. coelicolor needs to be further explored to understand the role of iron in S. coelicolor metabolism and enhance the production of bioactive metabolites.

Conclusions

The bacterioferritin from S. coelicolor plays an important role in antibiotic production and development. Although its function in iron storage appears to be replaceable, Bfr could also have a role modulating reactive oxygen species (ROS) levels and facilitating iron transfer to regulatory proteins. The absence of Bfr significantly impacts the regulation of proteins involved in secondary metabolism and differentiation, among others. Additionally, the influence of iron on antibiotic production varies between solid and liquid minimal media, and Bfr plays a role in this effect.

Therefore, to optimize the production of bioactive molecules and discover new ones, it is essential to consider iron, its storage proteins, and oxidative stress as components of the complex regulatory network of antibiotic production in actinomycetes.