Cryptic specialized metabolites drive Streptomyces exploration and provide a competitive advantage during growth with other microbes

Significance

Iron is an essential nutrient. Bacteria utilize siderophores to obtain iron from their environment, often producing multiple siderophores. Whether these molecules have distinct biological functions is unclear. Streptomyces exploration promotes the induction of a unique repertoire of specialized metabolites, including siderophores. We compared the effect of the well-characterized desferrioxamine siderophore, with the alternative siderophore foroxymithine. These molecules shared functional redundancy during monoculture but had distinct spatial localization profiles and exhibited significant functional differentiation during co-culture with other microbes. Our results suggest that while desferrioxamine can serve as a common good, with other microbes having ferrioxamine uptake systems, the alternative foroxymithine siderophore effectively privatizes the iron surrounding a colony, enhancing the competitive fitness of its producer when growing in polymicrobial communities.

Abstract

Streptomyces bacteria have a complex life cycle that is intricately linked with their remarkable metabolic capabilities. Exploration is a recently discovered developmental innovation of these bacteria, that involves the rapid expansion of a structured colony on solid surfaces. Nutrient availability impacts exploration dynamics, and we have found that glycerol can dramatically increase exploration rates and alter the metabolic output of exploring colonies. We show here that glycerol-mediated growth acceleration is accompanied by distinct transcriptional signatures and by the activation of otherwise cryptic metabolites including the orange-pigmented coproporphyrin, the antibiotic chloramphenicol, and the uncommon, alternative siderophore foroxymithine. Exploring cultures are also known to produce the well-characterized desferrioxamine siderophore. Mutational studies of single and double siderophore mutants revealed functional redundancy when strains were cultured on their own; however, loss of the alternative foroxymithine siderophore imposed a more profound fitness penalty than loss of desferrioxamine during coculture with the yeast Saccharomyces cerevisiae. Notably, the two siderophores displayed distinct localization patterns, with desferrioxamine being confined within the colony area, and foroxymithine diffusing well beyond the colony boundary. The relative fitness advantage conferred by the alternative foroxymithine siderophore was abolished when the siderophore piracy capabilities of S. cerevisiae were eliminated (S. cerevisiae encodes a ferrioxamine-specific transporter). Our work suggests that exploring Streptomyces colonies can engage in nutrient-targeted metabolic arms races, deploying alternative siderophores that allow them to successfully outcompete other microbes for the limited bioavailable iron during coculture.

Among bacteria, the soil-dwelling Streptomyces demonstrate remarkable developmental and metabolic complexity. They have impressively diverse chemical synthetic capabilities and produce a staggering array of specialized metabolites. These natural products have been a rich source of antibiotics, alongside other molecules that have been repurposed for clinical and agricultural application (1–5). However, despite the demonstrated metabolic capabilities of the streptomycetes, most of their specialized metabolites are not produced under standard laboratory conditions (6). What conditions promote the synthesis of these compounds, and what ecological role these molecules facilitate, remain major questions in the field.

Specialized metabolite production is often coordinated with specific stages in the Streptomyces developmental cycle, suggesting there may be fitness benefits associated with their close coupling (7, 8). Classical development involves first establishing a branching vegetative mycelium. In response to nutrient limitation, these microbes transition into their reproductive growth phase (9). This involves raising aerial hyphae that go on to differentiate into chains of metabolically dormant spores. Initiation of this reproductive growth phase typically coincides with the onset of specialized metabolism; however, these two processes are often spatially segregated, with specialized metabolism being largely confined to the vegetative cells (10–12).

In 2017, it was discovered that Streptomyces growth and development is not limited to this sporulating life cycle (13). In response to specific environmental cues (e.g., low glucose, high amino acid concentrations, presence of fungi), Streptomyces species can exit their classical life cycle and enter an alternative exploratory growth mode. Exploration is characterized by a rapid outward expansion of vegetative-like mycelia on a solid surface and is accompanied by the emission of the small basic molecule trimethylamine (TMA). The release and diffusion of this volatile molecule results in a gradual rise in the pH of the environment surrounding the exploring colony (13). This pH increase impacts microbial growth and behavior in multiple ways: it enhances Streptomyces competition by reducing bioavailable iron and inhibiting the growth of other microbes in the vicinity, and it serves as a communication tool, being sensed by more distantly located Streptomyces colonies where it promotes the initiation of exploration by these species (13, 14).

Recent work aimed at better understanding the transcriptional changes occurring throughout the exploration cycle in Streptomyces venezuelae revealed differential expression of an operon whose products are responsible for glycerol catabolism (15). Glycerol supplementation of exploration-promoting growth medium led to a dramatically enhanced rate of exploration (Fig. 1A). In addition, glycerol supplementation led to the secretion of a vibrant orange pigment into the underlying medium, suggesting altered metabolic programs for these exploring cultures (Fig. 1A).

Fig. 1.

Gene expression while exploring on glycerol (YPG) is notably different from exploration on YP. (A) Phenotype of S. venezuelae cultures exploring on either YP or YPG medium, visualized either from the colony-side (left) or from the underside of the colony (right). (B) Differential gene expression between early and late timepoints in exploration on YPG. Genes are arranged by their position on the S. venezuelae chromosome, and the fold change in expression of each gene between the two conditions is plotted. Data points that are statistically significant are shown in red (P < 0.01). (C) Distribution of genes that showed a statistically significant change in expression (P < 0.01) between the early and late timepoints in the YP and YPG RNA-sequencing datasets. These genes are further divided into those that are up-regulated or down-regulated over time. (D) Differential gene expression between late timepoints in exploration on YP and YPG, plotted as described in (B). Biosynthetic gene clusters for chloramphenicol and an NRPS-containing cluster are boxed in blue and black, respectively.

Little is known about the specialized metabolic capabilities of exploring cultures. Given the apparent metabolic novelty associated with glycerol-supplemented exploration, we sought to understand the metabolite production associated with this growth mode. We determined that the orange pigment was a compound known as coproporphyrin, and that its accumulation was likely mediated by dysregulated heme synthesis imposed by iron limitation. We further discovered that glycerol-grown S. venezuelae exhibited potent antibiotic activity that was not seen for conventional exploring cultures. Using RNA sequencing, we determined that the transcriptional profiles associated with exploration on glycerol were distinct from those exhibited by exploring cultures grown without glycerol. We identified two specialized metabolic clusters that were specifically up-regulated during growth on glycerol-containing medium and showed that the products of both clusters (the antibiotic chloramphenicol and a cryptic siderophore) had antibacterial activity. The cryptic siderophore was produced at higher levels than the well-studied desferrioxamine siderophore and appeared to diffuse away from the colony in a way not seen for other siderophores. We found that this new compound, which we identified as the unusual siderophore foroxymithine, played a critical role during exploration, particularly when growing in association with yeast, and that loss of both siderophores completely abrogated exploration when yeast was present. Our work suggests that Streptomyces bacteria can deploy unique iron sequestration strategies involving specific roles for different siderophores to ensure survival when growing in the presence of competing microbes.

Results

Transcriptional Profiles for YPG-Grown Cultures are Markedly Different from YP-Grown Cultures.

Exploration of S. venezuelae on glycerol-containing exploration plates (conditions dubbed YPG, for yeast extract-peptone-glycerol) had pronounced phenotypic differences compared with growth on standard exploration-promoting conditions (termed YP, for yeast extract-peptone), including distinct metabolic outcomes (Fig. 1A). To gain insight into the transcriptional programs that underlie these differences, we isolated RNA from YPG-grown S. venezuelae at three timepoints, equivalent to those analyzed previously for YP-grown cultures (15). These included an early timepoint at 2 d postinoculation (prior to rapid colony expansion, analogous to 2 d on YP); a mid timepoint at 4 d (representing an intermediate stage when exploration was actively proceeding, corresponding to 5 d on YP); and a late timepoint at 7 d (where individual colonies had covered the majority of the available medium surface area and had developed robust wrinkling patterns, equivalent to 9 d on YP). Differential gene expression was observed genome-wide when comparing transcripts isolated from early- and late-timepoints in YPG-grown cultures (Fig. 1B), consistent with results obtained previously from YP-grown cultures (15). We observed substantial differences in transcription profiles between YP- and YPG-grown cultures when comparing equivalent late-exploration timepoints (Fig. 1 C and D). When we compared those genes that were significantly up- and down-regulated between the early and late timepoints for each growth condition, we found distinct gene sets (Fig. 1C). While there were some common trends observed during growth on the two media types (e.g., nitrogen metabolism genes were down-regulated, while sulfur metabolism and iron-sulfur cluster biogenesis genes were up-regulated over time for both YP- and YPG-grown cultures; SI Appendix, Fig. 1), the majority of differentially expressed genes were unique to a single condition. These results suggested that exploration on YP and YPG may have different genetic drivers.

The strong pigmentation that accompanied growth on YPG (Fig. 1A) prompted us to examine whether genes directing the production of specialized metabolites were up-regulated under these growth conditions. We found that two clusters were significantly up-regulated during growth on YPG: the chloramphenicol biosynthetic cluster and a nonribosomal peptide synthetase (NRPS) gene-containing cluster (Fig. 1D). To address whether the unknown pigment originated from the identified NRPS cluster (vnz_34740-34845), we created a deletion strain where the core synthetase-encoding gene was replaced with a hygromycin resistance cassette (ΔNRPS). Following the growth of this strain on YPG, we observed no reduction in pigment accumulation, suggesting that pigment production was independent of the NRPS cluster.

Exploration on Glycerol Is Associated with Porphyrin Accumulation.

As the genetic origin of the secreted pigment was not obvious from our RNA-sequencing data, we undertook a biochemical approach to identify and characterize this compound. We found it could be readily extracted from conditioned medium using methanol, which upon fractionation and concentration produced an intensely pink fraction. When analyzed by UV-visible spectroscopy, the pigment showed a single strong absorbance peak centered at 405 nm with minor peaks at 498, 535, and 574 nm (SI Appendix, Fig. 2). High-performance liquid chromatography-mass spectrometry (HPLC-MS) analysis of the purified pigment revealed two compounds in positive mode with m/z peaks at 655.28 and 716.18 (SI Appendix, Fig. 2). Taken together, these data were consistent with literature values reported for coproporphyrin in both the free and zinc-bound forms (16–18). Coproporphyrin is expected to be an intermediate in the heme biosynthetic pathway in Streptomyces (19–21). Surprisingly, the majority of the heme biosynthetic genes (with the exception of hemB, encoding porphobilinogen synthase) were not up-regulated on YPG relative to YP (SI Appendix, Fig. 2). Thus, the basis for porphyrin accumulation was not immediately clear.

Specialized Metabolism Is Altered during Exploration on YPG and Impacts the Growth of Other Microbes.

As the transcriptional dataset showed the up-regulation of two biosynthetic gene clusters over the course of exploration on YPG, we were curious about the antibiotic production capacity of these cells. Using a culture-based bioassay, we found that YP agar plugs conditioned with metabolites from S. venezuelae had negligible antibiotic activity against the gram-positive indicator organism Micrococcus luteus (Fig. 2A). In contrast, conditioned YPG agar plugs yielded robust inhibition zones. As one of the up-regulated gene clusters directs chloramphenicol synthesis (where chloramphenicol is a known antibiotic), we took advantage of a mutant that is deficient in chloramphenicol-production (ΔcmlR; 22) and tested the antibiotic production potential of this strain when grown on YPG. Surprisingly, growth inhibition of M. luteus was comparable between the wild type and mutant strains, suggesting that another antibiotic must be contributing to the observed inhibition (Fig. 2A). We returned our attention to the ΔNRPS mutant generated earlier. Conditioned medium from this strain had little antibiotic activity, with only a faint zone of inhibition observed, compared with the robust zone associated with conditioned medium from the wild type. We generated a double mutant strain (ΔcmlR ΔNRPS) and found that antibiotic activity was completely abolished in this background, suggesting that chloramphenicol was being produced, but that it made a relatively minor contribution to the observed growth inhibition.

Fig. 2.

Antibiotic activity of exploring S. venezuelae is due to cryptic siderophore activity. (A) Plugs of conditioned exploration medium from S. venezuelae strains of interest were tested for antibiotic activity against the gram-positive indicator organism M. luteus (Top); antibiotic activity against M. luteus supplemented with 100 µM FeCl₃ (Middle); and for siderophore activity in the Chrome Azurol S assay (Bottom). (B) Metabolite extracts from indicated strains were separated by flash chromatography. Resulting individual fractions were then tested for antibiotic activity against M. luteus. (C) (Top) Organization of the foroxymithine (fxm) NRPS biosynthetic cluster. Genes are color-coded according to functional category as predicted using antiSMASH (23). (Bottom, left, and center) Domain organization of the four modules within the core NRPS (fxmE); predicted substrate specificities for each A-domain are presented below each module [adapted from the graphical output generated by antiSMASH; (24) )] (D) Top: total ion chromatogram for the separation using an MS detector tuned to ions with an m/z of 576.26. Bottom: MS/MS fragmentation data for the associated peak. E) Proposed chemical structure of the siderophore (and known chemical structure for the siderophore foroxymithine). Dashed arrows indicate theoretical molecular ion fragments that produce m/z values observed in the experimental data shown in (D).

To verify that the YPG-grown exploring cultures produced two distinct bioactive compounds, we compared metabolite extract profiles from the wild type, ΔcmlR, and ΔcmlR ΔNRPS strains (Fig. 2B). Activity testing of individual fractions revealed that the wild type indeed produced two groups of bioactive molecules. The late-eluting active fractions were lost in the ΔcmlR mutant strain, and thus were expected to contain chloramphenicol. No activity was observed in any fraction from the double ΔcmlR ΔNRPS mutant, suggesting that the early-eluting fractions of the wild-type and ΔcmlR extracts corresponded to the product of the NRPS cluster.

The Product of the NRPS Cluster Is the Unusual Siderophore Foroxymithine.

Bioinformatic analysis of the NRPS-containing biosynthetic gene cluster (Fig. 2C and SI Appendix, Table S1) using antiSMASH (23) revealed a high degree of similarity to the salinichelin biosynthetic genes, where salinichelin is a hydroxamate siderophore produced by Salinispora spp. (25). This suggested that the final product of this biosynthetic cluster may be a siderophore. To test this, we compared the activity of our different mutants (ΔcmlR, ΔNRPS, and ΔcmlR ΔNRPS) for their ability to inhibit the growth of M. luteus when additional iron was added to the growth medium (Fig. 2A). We found that iron supplementation effectively rescued the growth of M. luteus in the presence of conditioned media agar plugs from wild-type and chloramphenicol mutants (strains producing the predicted siderophore).

We further probed the siderophore potential of this metabolite using a Chrome Azurol S (CAS)-based assay (Fig. 2A) (26–28). CAS is a metal-sensitive, colorimetric dye that changes from blue when iron is present, to pink when iron is absent. Conditioned media plugs from YPG explorers yielded large pink halos in strains where the NRPS cluster was intact, and these halos were abolished when the cluster was deleted. Collectively, these results suggested that exploratory growth on YPG stimulated the production of both chloramphenicol, and a normally silent siderophore.

Purified YPG metabolite fractions containing the active siderophore were analyzed by LC-MS, alongside equivalent fractions (eluted from the column under the same polarity conditions) isolated from the ΔNRPS strain. This revealed a parent ion unique to the wild-type sample with an m/z of 576.26 (positive mode) (Fig. 2D). The addition of excess FeCl₃ to the sample led to the appearance of a new peak with strong UV absorbance at 435 nm, and a parent ion with a corresponding m/z of 629.17 (SI Appendix, Fig. 3). Coupling the LC-MS data with phylogenetic analyses within antiSMASH (23), the presence of additional tailoring enzymes encoded within the NRPS cluster (Fig. 2C and SI Appendix, Table S1), and an understanding of the salinichelin biosynthetic pathway [in particular, the unusual final cyclization and product release mechanism that appears to be shared by the S. venezuelae NRPS (25)], we generated a proposed chemical structure for our NRPS siderophore (Fig. 2E). The cleavage events across the peptide bonds of our predicted structure were consistent with the fragment ion peaks observed in our collected MS/MS data, further supporting this proposed structure. Mining of the literature for characterized siderophores sharing these chemical properties led us to conclude that this molecule was foroxymithine, a siderophore known to be produced by a handful of Streptomyces species (29, 30).

Porphyrin Accumulation Is a Function of Iron Starvation.

The production of a dedicated siderophore during growth on YPG but not YP, suggested that the iron-limiting conditions normally imposed by exploration may be exacerbated on YPG. This prompted us to revisit our analysis of coproporphyrin (orange pigment) accumulation during exploration on YPG. In Streptomyces, heme is generated following the insertion of iron into coproporphyrin by HemH (ferrochelatase) and subsequent decarboxylation of this iron-containing complex by HemQ (SI Appendix, Fig. 2) (20). Given this, we would expect that under iron-limited conditions, flux through this pathway may lead to a bottleneck at the point of iron insertion, resulting in a build-up of coproporphyrin as the immediate precursor. To test this hypothesis, we assessed whether coproporphyrin accumulation was impacted when YPG plates were supplemented with additional iron.

Upon adding 100 µM FeCl₃ to YPG, the resulting colonies lost their pronounced orange pigmentation (Movie S1). To confirm that the lack of coloration did indeed reflect reduced coproporphyrin production, we extracted metabolites from cultures grown on YPG with and without iron supplementation. These extracts were fractionated and concentrated, and a UV-visible absorbance spectrum was acquired for each, over a range of 300 to 650 nm (SI Appendix, Fig. 4). While the most strongly colored fractions in the iron-supplemented samples registered small absorbance peaks at 405 nm (reflecting small amounts of coproporphyrin), equivalent fractions from the unsupplemented YPG extracts were much more concentrated and needed to be diluted threefold to fourfold for peak intensities to be accurately measured.

We further subjected these fractions to HPLC-MS, tuning for m/z values specific for unmetallated, zinc-bound (the form identified above; SI Appendix, Fig. 2), and iron-bound forms of coproporphyrin (m/z of 655, 716, and 708, respectively (17, 31)). Peaks corresponding to unmetallated coproporphyrin were observed in all samples; however, no peaks corresponding to iron-bound coproporphyrin were detected (SI Appendix, Fig. 4). Finally, to test whether reduced porphyrin secretion was specific to iron, we supplemented YPG with 100 µM of other biologically relevant metal cations (Mn²⁺, Ni²⁺, Zn²⁺, Mg²⁺) and found that only iron impacted coproporphyrin/orange pigment production (SI Appendix, Fig. 4). These data collectively suggested that coproporphyrin accumulation and secretion during growth on YPG occurred specifically in response to iron limitation.

Distinct Siderophore Distributions Are Seen during Exploration on Different Growth Substrates.

Iron limitation is intricately connected with the exploration process, as exploration itself generates iron-deficient conditions by raising the pH of the surrounding environment (13). Previous work has shown that iron limitation enhances exploration, and has suggested a role for the major siderophore desferrioxamine (product of the desABCD locus) in iron metabolism under classical exploring conditions (14). Our work here suggests that iron may be particularly limiting under YPG exploration conditions, and consequently, we wanted to probe the relative contributions of desferrioxamine and the cryptic foroxymithine siderophore to exploration under different conditions. We constructed a double siderophore mutant (ΔNRPS ΔdesABCD) and assessed the behavior of this strain during growth under distinct exploring conditions. On YP, both single siderophore mutants had a modest delay in the rate of colony expansion and development. Notably, the double mutant exhibited significant defects: colonies were much smaller and failed to establish the wrinkled core structures that are characteristic of wild-type exploration (Fig. 3A). Paradoxically, on YPG, all siderophore mutants displayed relatively minor exploration defects compared with the wild type (Fig. 3A). The ΔNRPS strain behaved similarly to the wild type, quickly forming established colonies, while the desferrioxamine mutant expanded at a slightly slower rate. The phenotype of the double mutant was similar to that of YP-grown cultures, having a wrinkled core with defined boundaries and flatter architecture at the colony edge; however, it expanded more rapidly than a typical YP-growing culture.

Fig. 3.

Spatial distribution of siderophore activity. (A) Wild-type S. venezuelae was spotted alongside the single and double siderophore mutants on YP and YPG. Representative colonies were imaged after 8 d of growth. (B) Wild-type and single/double siderophore mutant strains grown on YPG for 5 d (Top), and the colony area determined. The plates and colonies were overlaid with CAS agar discs for 3 h; the CAS discs were then removed (Bottom) and the resulting pink area (where iron had been chelated) for each was quantified. (C) Graph depicting the ratio of pink/iron-chelated area to colony area, for the strains shown in (A) and (B). The dotted line at 1.0 indicates equivalent areas for both. Bars depict the average of three replicates (each replicate value indicated with a dot).

While the phenotypic consequences associated with the NRPS deletion were not profound during growth on YP or YPG agar, we were intrigued by the iron sequestration patterns observed in our CAS assays (Fig. 2A); specifically, the large haloes associated with the NRPS-associated siderophore-producing strain. We questioned whether there may be spatial differences associated with desferrioxamine and foroxymithine during exploration. To examine this possibility, we tested our suite of siderophore-producing strains using a modified CAS assay, where this visual/colorimetric probe of iron availability in the media served as a proxy for the localization (and activity) of the siderophores produced by our different strains. We found that strains producing the cryptic foroxymithine siderophore (wild type and ΔdesABCD) consistently sequestered iron well beyond the colony boundary (Fig. 3 B and C). In contrast, strains producing only desferrioxamine (ΔNRPS) captured iron exclusively within the area occupied by the colony, and those lacking both siderophores had minimal iron uptake capabilities (Fig. 3 B and C). This suggested that foroxymithine was uniquely able to diffuse well beyond the colony borders, an observation that was consistent with the greater relative hydrophilicity predicted for foroxymithine relative to desferrioxamine (predicted LogP value of −4 for foroxymithine; experimentally determined LogP value of −2.1 for desferrioxamine; where LogP refers to the partitioning ratio between octanol and water as reported in PubChem).

Siderophore Competition Dictates Exploration Success.

We wondered whether the apparent diffusion/release of foroxymithine at the leading edge of the exploring colony could function to reserve iron ahead of the expanding colony and hypothesized that such capabilities might be particularly beneficial in a competitive environment, for example, when exploring S. venezuelae encounters other microbes. It had previously been shown that S. venezuelae could initiate exploration when grown in close proximity with yeast, so we tested the exploration of our siderophore mutant strains in association with Saccharomyces cerevisiae on YPD (yeast extract-peptone-dextrose) agar (13). We found the double siderophore mutant was entirely unable to grow when inoculated in association with S. cerevisiae, and that the NRPS mutant strain (lacking foroxymithine), exhibited significant growth delays compared with the wild type and desferrioxamine mutant strains (Fig. 4 A and B). This suggested that foroxymithine played a critical role in competitive interactions.

Fig. 4.

Cryptic siderophore activity is critical for competitive growth with yeast. (A) Wild-type S. venezuelae, as well as single and double siderophore mutants, were mixed together with the yeast S. cerevisiae, and the resulting cell mixtures were then spotted to YPD with and without iron supplementation. Representative colonies were imaged after 16 d of growth. Note that the only growth observed on the double mutant plate in the absence of iron supplementation is the yeast S. cerevisiae. (B) Solid media growth curves were prepared for the wild type and siderophore mutants spotted to YPD together with S. cerevisiae—with (solid lines) and without (dashed lines) iron supplementation—and tracked over the course of exploration. The average of six replicates for each condition were plotted. Error bars represent 1 SD. (C) YPG medium was conditioned by strains of exploring S. venezuelae with different siderophore production capabilities (as indicated by the labels along the top panel). After 7 d of growth, metabolites were extracted from the conditioned medium alongside an unconditioned medium control, which were subsequently used to supplement solid YPD. Normalized cocultures of wild-type S. cerevisiae with either wild type S. venezuelae (Top image panel) or the double siderophore mutant ΔdesABCD ΔNRPS (Bottom image panel), were spotted to the extract-supplemented plates alongside an unsupplemented YPD control and scored for exploration. Images were taken after 14 d of growth. *: + represents presence and − represents absence of siderophores predicted to be present in the YPG extract of interest. ¥: ++ represents moderate rescue of exploratory growth, + represents partial rescue of exploratory growth, − represents no rescue of exploratory growth.

To ensure that these effects were due to the loss of the respective siderophores and their associated iron acquisition capacities, we supplemented the YPD growth medium with either 100 µM iron (Fig. 4 A and B), or YPG-conditioned extracts from wild type or desABCD mutant strains (capable of producing foroxymithine) (Fig. 4C). We found that these could restore wild type-levels of exploration to the double mutant strain when grown in association with S. cerevisiae. In contrast, conditioned extracts from the NRPS mutant strain (producing desferrioxamine) only partially restored growth, and extracts from the double mutant and medium-alone control both failed to promote growth of the double mutant when cocultured with yeast (Fig. 4C). Collectively, these results suggested that the production of foroxymithine was of paramount importance for exploration when S. venezuelae is grown in association with other microbes, and further suggested that siderophore activities are not redundant, and the siderophore preferences and/or requirements of exploring colonies can differ depending on the exploration environment.

We were curious about the inability of the double siderophore mutant to grow adjacent to yeast, given that the double mutant strain could explore well under other conditions. An interesting aspect of S. cerevisiae biology is that while it does not produce siderophores, it encodes siderophore-specific transporters (32, 33). One of these transporters, Sit1, can take up ferrioxamines (33, 34). We hypothesized that when grown in competition with yeast, the desferrioxamines produced by S. venezuelae may be subject to piracy by S. cerevisiae, while foroxymithine may be resistant to yeast uptake. To test this, we compared the rate of colony expansion for our S. venezuelae siderophore mutants when grown with either wild type S. cerevisiae or an isogenic Δsit1 strain (Fig. 5). Importantly, the significant growth defect associated with the NRPS deletion (and loss of foroxymithine) was no longer observed (Fig. 5), suggesting that this compound is less critical when siderophore competition/piracy is removed. Furthermore, when the double siderophore mutant was cocultured with the Δsit1 mutant, it consistently exhibited modest exploratory growth, in contrast to its complete inability to grow when cocultured with wild type yeast (Fig. 5). These results suggest that the need for iron acquisition may promote microbial arms races, where survival in the face of competition requires alternative nutrient acquisition strategies.

Fig. 5.

Presence of a yeast ferrioxamine-specific transporter influences growth rates of exploring cultures. (A) Representative images of S. venezuelae siderophore mutants exploring in the presence of wild-type S. cerevisiae or the Δsit1 mutant strain. Photos were taken after 12 d of growth. (B) Growth curves measuring colony surface area for S. venezuelae wild-type and siderophore mutants spotted with wild-type S. cerevisiae. (C) Growth curves measuring colony surface area for S. venezuelae wild-type and siderophore mutants spotted with Δsit1 S. cerevisiae.

Discussion

Our work here has revealed that different modes of exploratory growth are associated with unique specialized metabolic signatures. For S. venezuelae growing on YPG, these metabolites conferred a significant competitive advantage relative to classical development or exploratory growth on YP, with increased antibiotic activity, and enhanced iron acquisition and nutrient modulatory capabilities.

In addition to expanding its iron capture strategies during growth on YPG through the production of an otherwise transcriptionally silent siderophore, S. venezuelae also appears to use differential distribution of its multiple siderophores as part of a multitiered approach to iron sequestration. It has previously been shown that exploring Streptomyces colonies release the basic volatile compound TMA. Given its volatile nature, TMA can diffuse broadly, and in raising the pH of the surrounding environment, reduces the levels of bioavailable iron and limits the growth of nearby microbes (13, 14). Foroxymithine also appears to readily diffuse away from its producing colony, effectively sequestering iron well beyond the edge of the growing colony (as well as within the colony area) and inhibiting the growth of adjacent microbes. Finally, desferrioxamine appears to function primarily to secure iron within the colony boundary (Fig. 6). This ability of foroxymithine to bind and reserve iron for its producer appeared to be most critical when S. venezuelae was growing in association with competing microbes, and particularly when growing with those having the ability to take up (but not produce) ferrioxamine siderophores. Ferrioxamine piracy has been observed in other microbial systems (35–37), suggesting that upon sensing nearby microbial competitors, S. venezuelae may activate the production of siderophores for which dedicated transporters are less broadly distributed (e.g., foroxymithine), to ensure a more unidirectional recovery of an otherwise common good, This may explain—at least in part—why many bacteria encode multiple siderophores. An analogous behavior has been observed for Bacillus subtilis, which releases the iron-binding pulcherrimin to create an iron-free zone that effectively protects the B. subtilis biofilm from invasion by other microbes (38). A key difference is that pulcherrimin release leads to the cessation of colony expansion in B. subtilis, while foroxymithine secretion promotes increased exploration of S. venezuelae.

Fig. 6.

Schematic overview of the nutritional and metabolic factors influencing Streptomyces exploration and competition with other microbes. Yeast can produce glycerol, which accelerates Streptomyces exploration. Exploration involves the release of the volatile compound TMA, which causes a rise in the pH of the surrounding environment, reducing the amount of bioavailable iron. Streptomyces respond to these low iron conditions with the release of siderophores (and reuptake when iron-bound), including desferrioxamine and foroxymithine. Desferrioxamine release is confined to the exploring colony area but can also be taken up by yeast through its Sit1 transporter. Foroxymithine release extends beyond the colony area and further contributes to reducing the amount of iron available to other organisms. This iron-depleted zone ultimately inhibits the growth of other organisms, together with other antibiotics released by the exploring colony like chloramphenicol. Double headed arrows indicate both secretion and uptake. Asterisks indicate molecules that reduce iron levels surrounding the exploring colony.

Iron competition has been reported to trigger antibiotic production in other bacterial systems (39), and siderophores themselves are receiving increasing attention for their antimicrobial and biocontrol properties (40). Indeed, the antibiotic activity observed for YPG-grown S. venezuelae was largely due to its foroxymithine production (Fig. 2A). Beyond the synthesis and secretion of the foroxymithine and desferrioxamine siderophores, we also found S. venezuelae produced chloramphenicol, as well as the orange pigmented coproporphyrin during exploration on YPG. Interestingly, chloramphenicol can also bind iron (41), and so it is conceivable that on top of its antibiotic properties, chloramphenicol can further limit the amount of free iron available to other organisms when produced by S. venezuelae. In the case of coproporphyrin, our results suggest that its secretion occurs as a consequence of iron limitation and a corresponding dysregulation of heme synthesis. Accumulation of coproporphyrin to visually detectable levels has been widely observed in different bacterial systems, often in response to iron (42–44) or oxygen (45, 46) limitation. A few functional roles for the molecule have been proposed, including it acting as a chelator for metal acquisition (16, 17), and mediating interbacterial interactions (47–49). Whether the release of coproporphyrin serves an equivalent function in S. venezuelae remains to be seen.

A key finding here is that glycerol supplementation dramatically enhances the exploration response and rewires the transcriptional program of S. venezuelae. Roles for glycerol in modulating microbial behavior have also been documented in other bacterial systems. For the probiotic bacteria Lactobacillus reuteri and Propionibacterium spp., glycerol supplementation enhances competitive fitness against Escherichia coli and fungi, respectively, by altering secreted metabolite profiles (50, 51). In Mycoplasma pneumoniae, glycerol activates virulence by feeding into hydrogen peroxide biogenesis (52), while in the fungal plant pathogen Magnaporthe grisea, glycerol uptake promotes cellular swelling, and the resulting turgor pressure allows the pathogen to break through the host cuticle (53). Curiously, glycerol has also been shown to induce or enhance the establishment of bacterial biofilms in multiple systems, often without a clear underlying molecular mechanism having been established (54–56). It will be interesting to see whether growing exploration-competent Streptomyces species on glycerol-supplemented media can unlock the production of novel siderophores and other metabolites with useful bioactivity.

From an ecological perspective, glycerol is the product of lipid degradation (e.g., catabolism of phospholipids from the membranes of dead cells) or is secreted from other microbes. In particular, yeasts like S. cerevisiae are well known for secreting glycerol as a heterofermentation product when glucose concentrations are high (57, 58). Therefore, it is tempting to speculate that in our laboratory context of YPG-exploration, S. venezuelae interprets the presence of glycerol in a glucose depleted environment, as an imminent encounter with yeast. In response to these stimuli, S. venezuelae initiates robust exploration and an altered specialized metabolic program, including the production of multiple siderophores to ensure adequate iron acquisition by the cells (Fig. 6). In the first report describing exploratory growth by S. venezuelae, it was observed that many wild Streptomyces isolates failed to explore when cocultured with S. cerevisiae (13). Our results here raise the intriguing possibility that the specific siderophore repertoire encoded within a given species may be a strong determinant of whether that strain will be able to effectively explore in yeast coculture.

Materials and Methods

Strains, Plasmids, Media, and Culture Conditions.

Strains, plasmids, and primers used in this study are listed in SI Appendix, Tables S2–S4, respectively. S. venezuelae NRRL B-65442 was grown in liquid MYM (1% malt extract, 0.4% yeast extract, 0.4% maltose) for overnight cultivation and solid MYM (2% agar) for spore stock generation and vegetative (nonexploratory) growth controls. For exploration experiments, 10 µL of an overnight culture of S. venezuelae (MYM, 10 mL) were spotted to solid YP medium (1% yeast extract, 2% peptone, 2% agar) additionally supplemented with 2% carbon source (e.g., dextrose, glycerol) and/or additional nutrients (iron (III) chloride, zinc (II) chloride, manganese (II) chloride, magnesium chloride, nickel (II) chloride; 10 µM, 100 µM, or 1 mM) where appropriate. In exploration experiments where the growth of multiple strains was being compared, overnight cultures were diluted with MYM to a normalized OD₆₀₀. For experiments assessing the growth of S. venezuelae in competition with S. cerevisiae, 10 µL of an overnight culture of S. venezuelae (MYM, 10 mL) normalized for OD₆₀₀ were premixed with 10 µL of S. cerevisiae and the resulting 20 µL mixture was directly spotted to relevant solid media. All Streptomyces and Saccharomyces cultures were incubated at 30 °C.

For culture-based antibiotic activity assays (detailed below), Micrococcus luteus was cultured overnight with shaking at 37 °C in liquid lysogeny broth (LB) medium (1% tryptone, 0.5% yeast extract, 1% NaCl). Overnight cultures were used to inoculate solid LB medium (1.5% agar; supplemented with 2% glycerol when testing activity of YPG-conditioned medium to eliminate background effects) at a concentration of 1% (vol/vol) and poured as 20 mL plates.

Construction of S. venezuelae Mutants.

Gene deletions were generated using ReDirect technology (59). The coding sequence of vnz_34785 (fxmE) on the cosmid vector Sv-6-D08 was replaced by an oriT-containing hygromycin resistance cassette. The resulting mutant cosmid was introduced into the nonmethylating E. coli strain ET12567/pUZ8002 followed by conjugation into S. venezuelae. Resulting exconjugants were screened for double-crossover events using antibiotic selection, and gene deletions were verified by PCR, using combinations of primers located upstream, downstream, and internal to the deleted regions (SI Appendix, Table S4).

Time-Lapse Videos of Exploring Colonies.

Cells (10 µL) were inoculated onto exploration medium, after which the plates were placed on an Epson Perfection V800 Photo scanner, programmed to acquire one image every hour, and incubated at 30 °C. The time-course images were then compiled to video format as sequential single frames.

Solid Media Growth Curves.

Photos of growing colonies were taken at set time intervals. All image analyses were performed using ImageJ. Scale was established for each image by setting the diameter of the Petri dish to a set distance of 10 cm. The perimeter of the colony was traced, and the area of the captured region was measured.

RNA Isolation, Library Preparation, and cDNA Sequencing.

RNA was isolated as described previously (13), from two independent replicates of S. venezuelae grown on solid exploration medium for specified incubation times (2, 4, and 7 d of growth for YPG). For all samples, ribosomal RNA (rRNA) was depleted using a Ribo-Zero rRNA depletion kit. cDNA and Illumina library preparation were performed using a NEBnext Ultra Directional Library Kit, followed by sequencing using unpaired-end 80 base pair reads using the MiSeq platform. All bioinformatic analyses were carried out using packages available through the free open-access platform, Galaxy (usegalaxy.org) using the default settings. Reads were aligned to the S. venezuelae genome using Bowtie2 (60), then sorted, indexed, and converted to BAM format using SAMtools (61). Transcript level normalization and analyses of differential transcript levels were conducted using DESeq2 (62). RNA-seq data has been submitted to the NCBI GEO repository and assigned the accession number GSE181041.

Culture-Based Assays for Antibiotic Activity.

LB agar inoculated with our general antibiotic-sensitive indicator organism, M. luteus, was prepared as described above. To assay conditioned solid medium, the opening of an inverted sterile 1 mL pipette tip was used to punch an agar plug from the center of an S. venezuelae colony. The agar plug was then transferred onto the surface of the indicator agar plate. To assay activity from liquid samples (i.e., total crude metabolite extracts/fractions generated by liquid chromatography), a 6 mm filter disk was placed on the surface of the indicator agar, to which 20 µL of the test sample was then applied. After sample application, indicator plates were incubated overnight at 37 °C. Following incubation, antibiotic activity could be observed as the zones of bacterial growth inhibition around the discs, relative to background growth over the rest of the plate.

Chrome Azurol S Assay for Siderophore Activity.

Samples containing siderophores of interest were assessed using a Chrome Azurol S-based assay (26–28). Briefly, sterile solid medium containing the colorimetric dye (2% agar, 100 µM Chrome Azurol S, 200 µM hexadecyltrimethylammonium bromide, 10 µM FeCl₃, 0.5 M 2-(N-morpholino)ethanesulfonic acid, pH 5.5) was poured as 15 mL plates. As described for the culture-based activity assay, plugs of conditioned media or filter discs loaded with liquid samples were placed on the surface of the colorimetric indicator plate. Prepared plates were incubated in the dark, at room temperature overnight. Following incubation, the development of a pink halo against the blue background of the plate around experimental samples served as a proxy for siderophore activity.

To measure siderophore activity for conditioned medium in situ, 15 mL of CAS agar was prepared and poured as described above into a 10 cm Petri dish. At relevant timepoints, cells from exploring colonies were scraped away from the surface of conditioned exploration medium and the CAS agar was transferred to lay flush with the surface of the medium, taking care to remove any trapped air bubbles. The assembled plates were then incubated in the dark at room temperature for 4 h, after which the overlaid CAS agar was removed and photographed. Surface area measurements for zones of siderophore activity were performed similarly to those prepared for solid medium growth curves (SI Appendix, Fig. 5). In ImageJ, the diameter of the CAS agar was set to represent a distance of 10 cm, the perimeter was traced for areas of the dyed agar that had undergone a color change, and the corresponding area within the boundaries was computed.

Metabolite Extraction from Solid Medium.

Conditioned agar from exploring cultures was diced into small pieces, cooled to −80 °C, and lyophilized for at least 2 d until dry. The solid material was then ground to a powder and transferred to a preweighed 50 mL conical tube. Reagent grade methanol was then added to the powdered sample at a ratio of 10 mL per g and the mixture was shaken overnight at room temperature in the dark. The following day, extracts were passed through a 0.4 µm filter to remove any debris and divided as 7 mL aliquots. Collected extracts were then concentrated to dryness under vacuum.

Metabolite Extraction for Exploration Rescue Experiments.

Strains of S. venezuelae with different siderophore production capabilities were spotted to YPG agar and incubated at 30 °C for 7 d (colonies were approximately equivalent sizes for all strains). Following removal of bacterial cells from the agar, a 5 cm by 5 cm square of the conditioned medium from directly underneath the original colony was harvested and subjected to the same lyophilization, methanol extraction, and concentration steps as described above. As a negative control, a 5 × 5 cm square of (unconditioned) YPG medium was subjected to the same treatment. Dried material recovered for each sample was reconstituted in 1.2 mL of 50:50 water/methanol and was used to supplement YPD agar (200 µL of extract per 40 mL). Supplemented YPD was then tested for its ability to promote exploration in a coculture of S. venezuelae and S. cerevisiae.

Metabolite Fractionation by Liquid Chromatography.

Dried extracts were resuspended in an equal volume solution of 50:50 HPLC grade water/methanol. When assaying for coproporphyrin production between different experimental conditions (i.e., growth on YPG supplemented with or without 100 µM FeCl₃), starting material was normalized by resuspending and combining three 7 mL aliquots in 1.4 mL of water/methanol; in all other applications, a minimal volume of water/methanol was added to dry material to maximize metabolite concentration. Following a 10 min centrifugation at >15,000 × g for 10 min at 4 °C to remove any suspended solids, 1 mL of concentrated extract was separated by flash chromatography with a 13 g RediSep Rf Gold C18 column at a flow rate of 30 mL/min. Separation was achieved using the following gradient program: 0 to 1 min, isocratic 10% A; 1 to 10, min a gradient of 10 to 100% A, 10 to 12 min, isocratic 100% A, where A is methanol and B is water. All fractions were then concentrated to dryness under vacuum and resuspended in 200 µL of HPLC grade solvent (early fractions 100% water, mid fractions 50:50 water/methanol, and late fractions 100% methanol). Fractions containing coproporphyrin were easily identified by eye, reproducibly eluting during the gradient over the range of 80 to 95% MeOH. Fractions containing chloramphenicol or foroxymithine were determined using the activity assays described above.

Chemical Analysis of Coproporphyrin by UV-Visible Spectrometry.

From resuspended fractions, 2 µL samples were analyzed using a NanoDrop ND-1000 to acquire a UV-visible absorbance spectrum over the range of 300 to 650 nm.

Chemical Analysis of Purified Metabolites by HPLC-MS/MS.

Samples were analyzed using an Agilent 1290 LC system coupled with a Thermo Scientific LTQ Orbitrap XL mass spectrometer (ESI-MS) under positive ionization mode. All samples (10 μL for each injection) were separated on a Zorbax Eclipse XDB C18 column (100 mm × 2.1 mm × 3.5 μm) at a flow rate of 0.4 mL/min. For analysis of coproporphyrin samples, the following gradient program was run: 0 to 14 min from 95 to 5% solvent A, 14 to 17 min isocratic 5% A, 17 to 20 min isocratic 95% A, where A is water with 0.1% formic acid and B is acetonitrile with 0.1% formic acid. For analysis of foroxymithine samples, the following gradient program was run: 0 to 14 min from 100 to 90% solvent A, 14 to 14.5 min from 90 to 5% A, and a gradient of 14.5 to 17 min isocratic 5% A, 17 to 17.5 min from 5 to 100% A, 17.5 to 20 min isocratic 100% A (solvents used are the same as for coproporphyrin analysis). Data were visualized using the Xcalibur 2.1 software.

Data, Materials, and Software Availability

All study data are included in the article and/or SI Appendix.