Streptomyces extracellular vesicles are a broad and permissive antimicrobial packaging and delivery system

ABSTRACT

Streptomyces are the primary source of bioactive specialized metabolites used in research and medicine, including many antimicrobials. These are presumed to be secreted and function as freely soluble compounds. However, increasing evidence suggests that extracellular vesicles are an alternative secretion system. We assessed environmental and lab-adapted Streptomyces (sporulating filamentous actinomycetes) and found frequent production of antimicrobial vesicles. The molecular cargo included actinomycins, anthracyclines, candicidin, and actinorhodin, reflecting both diverse chemical properties and diverse antibacterial and antifungal activity. The levels of packaged antimicrobials correlated with the level of inhibitory activity of the vesicles, and a strain knocked out for the production of anthracyclines produced vesicles that lacked antimicrobial activity. We demonstrated that antimicrobial containing vesicles achieve direct delivery of the cargo to other microbes. Notably, this delivery via membrane fusion occurred to a broad range of microbes, including pathogenic bacteria and yeast. Vesicle encapsulation offers a broad and permissive packaging and delivery system for antimicrobial specialized metabolites, with important implications for ecology and translation.

IMPORTANCE

Extracellular vesicle encapsulation changes our picture of how antimicrobial metabolites function in the environment and provides an alternative translational approach for the delivery of antimicrobials. We find many Streptomyces strains are capable of releasing antimicrobial vesicles, and at least four distinct classes of compounds can be packaged, suggesting this is widespread in nature. This is a striking departure from the primary paradigm of the secretion and action of specialized metabolites as soluble compounds. Importantly, the vesicles deliver antimicrobial metabolites directly to other microbes via membrane fusion, including pathogenic bacteria and yeast. This suggests future applications in which lipid-encapsulated natural product antibiotics and antifungals could be used to solve some of the most pressing problems in drug resistance.

INTRODUCTION

Most modern medicine would be unimaginable without microbial specialized metabolites. These molecules have potent biological activities and, as a result, have been developed as antibiotics, chemotherapeutics, and other medicines (1). The crisis of antimicrobial resistance and the relatively poor performance of synthetic chemical libraries in antimicrobial discovery have spurred a renaissance in specialized metabolite research (2). Moreover, advances in genetic and metabolomic tools have demonstrated an abundance of novel chemistry yet to be explored in environmental bacteria (2–6).

Early reports of clinically valuable microbial specialized metabolites came with the discovery of penicillin from the Penicillium mold in 1928 and streptomycin from Streptomyces griseus in 1943 (7–9). Since then, the majority of specialized metabolites have been mined from the enormous actinomycete bacterial genus Streptomyces (6). Each strain of Streptomyces carries gene clusters for dozens of different specialized metabolites. Some of these compounds serve their producing organisms as siderophores and spore pigments (10), and many act as antimicrobials (11). Despite decades of research, mechanistic knowledge is limited to a relatively small fraction of these compounds and is even more scarce at the level of endogenous production and function (12).

Since their discovery nearly a century ago, it has been assumed that actinobacterial specialized metabolites act as free, soluble chemical agents and indeed, this is how they are employed clinically. The fact that many specialized metabolic gene clusters encode dedicated ABC transporters and efflux pumps has tacitly supported that view (13–15). However, this view was challenged in 2011 when Schrempf and colleagues (16) reported the ground-breaking discovery that actinorhodin, which had been studied for more than 60 years as a soluble compound (17, 18), was associated with lipid-encapsulated vesicles in secreted exudates from Streptomyces coelicolor. Since then, antibiotic-containing extracellular vesicles have been detected from Streptomyces lividans, Streptomyces sp. Mg1, and Streptomyces albus (19–22). These intriguing reports beg the question of how widespread this phenomenon might be among actinomyces.

It is increasingly apparent that extracellular vesicles are an integral part of the microbial world (23–25). Studies have focused primarily on Gram-negative genera, where they are predominantly derived from the outer membrane and referred to as outer membrane vesicles (24–28). Extracellular vesicles have now also been recognized in Gram-positive bacteria, forming from the cytoplasmic membrane by a poorly understood mechanism that is still under active investigation (29–34). Vesicles are known to carry a variety of molecular cargo, with most reports focused on their protein content. Importantly, the specific set of proteins contained in these vesicles differs from that of the periplasm or cytoplasm of the bacteria, suggesting they might be loaded according to specific targeting or enrichment mechanisms (32, 33, 35, 36).

Many roles have been assigned to extracellular vesicles, ranging from the removal of macromolecular waste to the delivery of specific cargo (30). Several pathogens deliver virulence factors such as α-hemolysin (Staphylococcus aureus) (37) and cholera toxin (Vibrio cholerae) (38) to host cells via extracellular vesicles. Pseudomonas aeruginosa releases vesicles that carry the quorum-sensing molecule PQS (39). Vesicles serve to facilitate horizontal gene transfer in Acinetobacter baumannii (40). They can also serve as antagonistic agents (41), carrying bacteriocins from Lactobacilli (42) or lytic enzymes and toxic metabolites from Myxococci (43), to kill other bacterial strains.

We asked what implications vesicle packaging might have for the transport and delivery of specialized metabolites, in particular antimicrobial metabolites. First, how widespread is this phenomenon in the most prolific specialized metabolite producers, the filamentous actinomycetes? Second, is this a specific or broad mechanism as it relates to spectrum of activity and metabolite properties? Third, and most importantly, how does vesicle encapsulation affect the delivery of specialized metabolites?

To address these questions, we examine 11 soil actinomycete isolates and 6 lab-adapted strains for antimicrobial extracellular vesicle production. We find that under laboratory conditions this is indeed common, 9 out of 17 strains, for actinomycetes. We identify the antimicrobial-specialized metabolites associated with extracellular vesicles from four strains and find that they have diverse chemical structures and biological activities. We demonstrate that vesicle activity is linked to the antimicrobials packaged. Importantly, we find that these extracellular vesicles fuse with the membranes of diverse bacteria and fungi, delivering antimicrobials directly to the cells. Vesicle-mediated delivery as a broad, permissive, and common mechanism represents a new insight into the enormous family of antimicrobial-specialized metabolites with far-reaching implications for their use in medicine and research.

RESULTS AND DISCUSSION

Filamentous actinomycetes frequently produce antimicrobial extracellular vesicles

If the packaging of specialized metabolites into extracellular vesicles is widespread, we reasoned it should be possible to detect metabolite-associated bioactivity in high molecular weight (MW) fractions of culture supernatants. Moreover, because vesicles are membrane-bound, these high molecular weight fractions should also contain lipids. Using centrifugal size filtration, we created a rigorously washed high MW fraction > 100 kDa and a matched fraction of low MW supernatant from 1-week liquid cultures of 17 antimicrobial-producing actinomycetes. This included common lab-adapted strains and environmental soil isolates (44). To test for lipids, we added the lipid-reactive dye FM 4-64 to each fraction. Of the 17 high MW fractions, 14 had increased fluorescence compared to the low MW fractions, consistent with the presence of extracellular vesicles (Table S1; Fig. S1A).

Strikingly nine of these high MW fractions also had antimicrobial activity (Table S1; Fig. S1A). As expected for activity arising from specialized metabolites, the spectrum of microbial inhibition of the high MW fractions differed across isolates. Moreover, matching antimicrobial activity was present in the low MW fractions presumably due to a portion of metabolites free in solution (Fig. 1A; Fig. S1A). Six high MW fractions, from the lab-adapted strain S. coelicolor M145 and the soil isolates WAC00187, 00218, 00237, 00240, and 00288, inhibited Gram-positive bacteria Staphylococcus aureus and Bacillus subtilis. Two high MW fractions, from WAC00265 and 00276, inhibited the fungi Candida albicans, and there was one, from WAC00303, with broad spectrum activity against both the bacteria and C. albicans. The distribution of antimicrobial activity between the high and low MW fractions differed between strains, with five of the isolates having equal or weaker activity in the high MW fraction (S. coelicolor, WAC00187, 00218, 00237, and 00240), whereas in four isolates (WAC00265, 00276, 00288, and 00303), the high MW fraction had the greatest antimicrobial activity (Fig. 1A; Fig. S1A). Finally, to confirm extracellular vesicle production, we examined the high MW fractions from isolates across the range of antimicrobial profiles, S. coelicolor (Fig. S1B) and three of the environmental isolates, WAC00276, 00288, and 00237 (Fig. 1B), using transmission electron microscopy. We observed abundant extracellular vesicles that ranged in size from tens to hundreds of nanometers in diameter. We further purified the extracellular vesicles using size exclusion chromatography (SEC) (45). Nanoparticle tracking analysis on these SEC-purified vesicles corroborated the electron microscopy results, revealing the majority of vesicles are between 50 and 150 nm, with mode sizes of 75, 65, and 85 nm for WAC00276, 00288, and 00237, respectively (Fig. 1C). Vesicle concentrations were calculated to be at 10¹⁰–10¹¹/mL of culture supernatant (Fig. 1C; Fig. S1C).

Fig 1.

Filamentous actinomycete isolates from soil can produce abundant antimicrobial extracellular vesicles. (A) Cell-free culture supernatant was divided into matched high (>100 kDa, washed 100×) and low MW (<100 kDa) fractions and tested for antimicrobial activity. WAC276 high MW fraction created a zone of inhibition against C. albicans, and WAC288 and WAC237 high and low MW fractions inhibited S. aureus. These high MW fractions were imaged by negative stain transmission electron microscopy revealing abundant extracellular vesicle structures (B). Vesicle purification via size exclusion chromatography on culture supernatant followed by nanoparticle tracking analysis was used to measure the concentration per milliliter (total vesicle concentration top right inset) and the size distribution of these vesicles (C). Results are representative of a minimum biological duplicate.

Extracellular vesicles package diverse antimicrobial-specialized metabolites

Taking our four strains with the range of bioactive profiles, we used liquid chromatography coupled to a tandem mass spectrometer (LC-MS/MS) to determine if the extracellular vesicles were associated with known antimicrobial-specialized metabolites. The vesicle fraction from Streptomyces coelicolor contained a clear peak of the molecular ion expected for the specialized metabolite actinorhodin (Fig. S1B), corroborating previous work (16, 46). For the environmental isolates WAC00276, 00288, and 00237, we used high performance liquid chromatography (HPLC) to fractionate butanol-extracted culture supernatant and obtained a semi-purified bioactive antimicrobial fraction. We also purified extracellular vesicles via size exclusion chromatography and extracted them with butanol or 1:1 isopropanol:acetonitrile. The vesicle extracts and the HPLC-purified bioactive fractions were then run on LC-MS/MS in comparison to blank solvent controls to identify antimicrobial metabolites present in the vesicles and in the bioactive fraction for each strain. The C. albicans inhibiting fraction and vesicles from WAC00276 contained a peak with molecular ions and an expected fragmentation pattern from the polyene candicidin (Fig. 2A through C) (47). The WAC00288 fraction and vesicles, with S. aureus inhibitory activity, contained peaks with molecular signatures for anthracyclines, namely hexaglycosylated β-rhodomycin analogs from the cosmomycin/ditrisarubicin family (Fig. 2D through F; Fig. S1D). The vesicles contained three dominant analogs (Fig. 2D; Fig. S1D), the major one with molecular ion and fragmentation patterns consistent with ditrisarubicin C (Fig. 2E and F) (48–50). The S. aureus inhibitory fraction and vesicles from the soil isolate WAC00237 had an abundant peak with the molecular ion and expected fragmentation pattern matching actinomycin X2 (Fig. 2G through I) (51). The metabolites eluted at very different retention times off the reverse-phase column, highlighting the wide spectrum of polarity and structural diversity in the compounds (Fig. 2). Candicidin was only present in WAC00276 and not in WAC00288 or WAC00237 vesicles, and similarly the anthracyclines and actinomycin X2 were only present in WAC00288 and WAC00237 vesicles, respectively (Fig. S1E).

Fig 2.

Vesicles package diverse antimicrobial metabolites. LC-MS/MS analysis identified clear peaks present in vesicle extracts (purified by size exclusion chromatography) not present in solvent blanks (black). (A–C) In WAC276 (green), a peak was present that matched the expected m/z (1,109.58) of the M + H ion of candicidin (A, extracted ion chromatogram of 1,109.580 ± 0.005 in SEC-purified WAC276 vesicles or blanks). This peak also had the expected mass fragments (B,MS^e spectrum of M + Na 1,131.56 peak at retention time (r.t.) 7.4 min in bioactive HPLC fractions of WAC276 supernatant) of candicidin (C). (D–F) WAC288 vesicles (orange) contained several peaks arising from related molecular structures, with m/z of M + H ions of 1,181.53, 1,165.53, and 1,149.54 (D, extracted ion chromatogram of 1,100–1,200 in SEC-purified vesicle extracts or blanks), and the expected mass fragments of hexaglycosylated β-rhodomycins (E, MS^e spectrum of vesicle peak at r.t. 6.5 min), including ditrisarubicin C (F). (G–I) WAC237 vesicles (yellow) contained a peak matching the m/z of the molecular ion of actinomycin X2 [M + H 1,269.62 (G) extracted ion chromatogram of 1,269.621 ± 0.005 of SEC-purified vesicle extracts or blanks), with fragments (H, MS^e spectrum of vesicle peak at r.t. 10 min) matching those expected from the molecular structure (I). Results are representative of at minimum biological triplicate batches of vesicles.

These vesicle-packaged antimicrobials differ in chemical structure, bioactivity, and biosynthetic pathway (Fig. 1 and 2; Fig. S1). In agreement with previous reports, but under different culture conditions, our results agree that Streptomyces coelicolor packages actinorhodin into vesicles (16, 46). Actinorhodin is a MW 634.5 Da, redox-active, planar molecule produced by polyketide synthases and has antimicrobial activity against some Gram-positive bacteria (52, 53). Our finding that soil isolate WAC00276 packages the antifungal compound candicidin into vesicles aligns with the report that the lab-adapted species S. albus releases vesicles containing candicidin (22). Candicidin is produced by many streptomycetes (54), a MW 1,109 Da amphiphilic polyketide that interferes with ergosterol in fungal membranes. We further identified that soil isolate WAC00237 packages actinomycin X2 MW 1,269 Da into vesicles. Actinomycins are large hydrophobic non-ribosomal peptides containing a core phenoxazine and two peptide lactone rings. This core can intercalate into DNA, leading to the inhibition of RNA polymerase. Finally, we find that soil isolate WAC00288 vesicles package several hexaglycosylated anthracyclines (cosmomycins/ditrisarubicins). Anthracyclines are also DNA intercalating but chemically distinct to actinomycins. These hydrophilic, highly glycosylated anthracyclines are assembled by polyketide synthases.

The diversity of these identified extracellular vesicle-packaged compounds, combined with the reports of lipophilic undecylprodigiosin and linearmycins in vesicles from streptomycetes (19, 20, 46), suggests that vesicle packaging is permissive in nature and can be used to package hydrophilic, amphiphilic, and hydrophobic compounds from diverse synthetic pathways, regardless of each compound’s ultimate molecular target or mode of biosynthesis. In answer to our first and second questions, our results suggest that extracellular vesicle packaging of antimicrobial-specialized metabolites is widespread in filamentous actinomycetes. This is remarkable because we used a single screening growth medium and time point and made no effort to optimize the expression of metabolites or vesicles. Specialized metabolite yields vary considerably with growth conditions in the laboratory, and many are not expressed at all (12, 55–57). Given the substantial association with vesicles we observed, we suspect that vesicle packaging of specialized metabolites is very common and possibly a ubiquitous capability in the actinomycetes. This is a dramatic departure from a decades-old paradigm of specialized metabolites acting primarily as secreted molecules in solution.

Specialized metabolite packaging is responsible for extracellular vesicle antimicrobial activity

To examine the relationship between specialized metabolite packaging and the antimicrobial activity of the extracellular vesicles, we utilized the UV-Vis absorbance properties of candicidin [peak at 390 nm (22)], the anthracyclines [495 nm (49)], and actinomycin X2 [445 nm (58)], enabling spectrophotometric detection. A crude vesicle preparation (washed >100 kDa fraction) was isolated from culture supernatants of environmental isolates WAC00276, 00288, and 00237 over time, tested for antimicrobial potency, and then metabolites were extracted by butanol and measured by absorbance. The level of antimicrobial activity in the vesicle fraction was notably correlated to the level of metabolites (Fig. 3A). As expected, metabolites and antimicrobial activity increased with days in culture, as did lipid fluorescence as an indicator of vesicle quantity (Fig. S2A). This correlation of vesicle antimicrobial activity with the levels of packaged metabolites continued through size exclusion chromatography purification (Fig. 3B). Extracellular vesicles are expected within the first three fractions of the column, and indeed these fractions contained spikes of lipids compared to the preceding void volume fractions and the later protein-rich fractions. The fractions also contained spikes of antimicrobial activity and spikes of the metabolites, and the vesicle fractions with the highest antimicrobial activity had the highest levels of metabolites (Fig. 3B). This correlation of packaged metabolites with the antimicrobial activity spectrum (Fig. 1 and 2) and potency (Fig. 3) suggests the vesicle antimicrobial activity is primarily derived from the specialized metabolite cargo.

Fig 3.

Vesicle antimicrobial potency matches the levels of packaged metabolites. (A) Correlation of antimicrobial activity and metabolite levels in crude vesicle fractions over days of culture. WAC276 (green squares), WAC288 (orange circles), and WAC237 (yellow triangles) culture supernatants were sampled over 8 days of culture (numbers on symbols indicate day), the high molecular weight (>100 kDa) fraction was purified and tested for antimicrobial activity (WAC276 fractions against C. albicans, WAC288 and WAC237 against S. aureus, serial dilutions of fractions tested for the inhibition of growth of liquid cultures) and then extracted with butanol and measured for absorbance at 390 nm for candicidin (in WAC276), 495 nm for anthracyclines (in WAC288), and 445 nm for actinomycin (in WAC237). Results are means ± SD of technical triplicate from an experiment representative of time courses done in biological triplicate. (B) Correlation of antimicrobial activity and metabolite levels in vesicle fractions purified by size exclusion chromatography. High molecular weight (>10 kDa) culture supernatant from WAC276 (squares), WAC288 (circles), and WAC237 (triangles) was loaded on qEV1 (Izon Sciences) SEC columns and 1 mL fractions were collected, assayed for the presence of lipids through fluorescence of FM 4-64 (purple), inhibition (black) of growth of liquid cultures of C. albicans (WAC276, tested at 1.7%, vol/vol) or S. aureus (WAC288 at 1.25%, vol/vol and WAC237 at 5%, vol/vol), and the presence of antimicrobial metabolites by absorbance (green for candicidin at 390 nm, orange for anthracyclines at 495 nm, and yellow for actinomycin X2 at 440 nm). Extracellular vesicles (EVs) are expected in the first 3 mL (fractions 1–3) after the void volume. Results are representative of at minimum biological duplicate. (C) Nanoparticle production over time by WAC288 wild type and genetically modified strains to abolish anthracycline production. High molecular weight fractions (>100 kDa) purified over time from wild-type (288 wt, orange circles) and three WAC288 strains knocked out for different essential genes in anthracycline (cosmomycin/ditrisarubicin) production, a padR regulator, a βKAS synthetic gene, and a SARP regulator, (black open symbols) were assayed for nanoparticle concentration per milliliter by nanoparticle tracking analysis. Results are means ± SD of technical triplicate from an experiment representative of biological duplicate. Wild-type results for each day were compared to each knockout by Welch’s unpaired t-test, adjusting for multiple comparisons using the Holm-Sidak method, and all P were <0.05 (*). (D) Vesicles purified from wt and the padR knockout by SEC were imaged by negative stain transmission electron microscopy. (E) WAC288 wt high molecular weight fractions create zones of inhibition against S. aureus, but fractions from the padR knockout do not, despite eight-fold concentration to match particle numbers of the wt.

To further answer the question of whether vesicle activity results from the presence of specialized metabolites, we focused on the anthracyclines from soil isolate WAC00288. These hexaglycosylated β-rhodomycins have been described as antibacterial, antifungal, antiphage, and insecticidal, covering a broad range of predators filamentous actinomycetes may encounter in the environment and are hence of considerable interest (48, 49, 59, 60). They are composed of a core anthraquinone decorated with six sugar chains (Fig. 2F), and their antimicrobial and cytotoxic activity depends on their ability to intercalate DNA (50, 59). We further purified WAC00288 vesicles by density through differential gradient ultracentrifugation and found that antimicrobial activity was found exclusively in fractions that also contained anthracyclines and vice versa (Fig. S2B). Moreover, when WAC00288 high molecular weight fractions were treated with detergent to lyse vesicles, anthracyclines and antimicrobial activity were then recovered from the low molecular weight fractions (Fig. S2C).

We next took advantage of modified strains of WAC00288 in which key regulatory and biosynthetic genes required for anthracycline (cosmomycins/ditrisarubicins) production are inactivated (49). These knockout strains have no growth defect and were cultured in parallel with the wild-type parent. We confirmed the absence of anthracyclines in all three mutant supernatants by mass spectrometry. Vesicles were still produced, determined by nanoparticle tracking analysis, transmission electron microscopy, and SEC purification (Fig. 3C and D; Fig. S3). Interestingly, there was a 10-fold reduction in final vesicle numbers in the mutant strains relative to the wild type (Fig. 3C; Fig. S3). There was also a complete loss of antimicrobial activity in vesicle fractions, including when adjusted to match wild-type vesicle levels (Fig. 3E). This loss of antimicrobial activity in the mutant vesicles, but clear connection between the amount of anthracyclines and antimicrobial activity in the wild-type vesicles (Fig. 3; Fig. S2), demonstrates that the anthracyclines are both required and responsible for vesicle antimicrobial activity.

In conclusion, the antimicrobial activity of the extracellular vesicles and the levels of cargo antimicrobial metabolites were highly associated in our strains (Fig. 3). Through rigorous purification and genetic inactivation, we demonstrated that the anthracycline cargo was essential for the antimicrobial activity of WAC00288 extracellular vesicles (Fig. 3); however, the metabolites were not required for extracellular vesicle production. Interestingly, the total yield of vesicles was higher when the cargo was present. This is similar to previous observations in Streptomyces sp. Mg1 and Pseudomonas aeruginosa where strains bearing deletions of the linearmycin or PQS cargo genes had reduced vesicle yields during purification (20, 39). In contrast, a violacein null mutant of Chromobacterium violaceum (61) produced greater vesicle yields relative to the wild type. It was argued that the hydrophobicity of these cargoes might alter vesicle production by modulating cellular membrane dynamics (20, 39). This seems less likely for the hydrophilic and highly glycosylated anthracyclines. An alternative view would be that in some cases, rather than influence the secretion of vesicles, the metabolite cargo enhances vesicle stability and recovery on purification. Given the wide variety of chemical cargo in extracellular vesicles that we (Fig. 2) and others have observed, and this repeated observation of an association between vesicle production and the production of specialized metabolite cargo (Fig. 3; Fig. S2), there may be regulatory mechanisms that link metabolite and vesicle biogenesis pathways.

Extracellular vesicles deliver antimicrobials to cells via membrane fusion

We next turned to examine the effect vesicle encapsulation has on the transport and delivery of specialized metabolites. We hypothesized that extracellular vesicles might deliver their antimicrobial metabolite content to target cells by direct fusion with cellular lipid membranes. This is particularly important for specialized metabolites with a cytoplasmic location of action, such as the anthracyclines, which intercalate into the DNA. The transfer of lipid-soluble fluorescent dyes is one method used to assess extracellular vesicle fusion with membranes (62). Labeled and unlabeled vesicles (SEC purified) were incubated with target cells. The cells were then pelleted and washed at low centrifugal forces that do not pellet vesicles. Two methods were used to compare the fluorescence of treated cells to negative controls of cells or vesicles alone: confocal microscopy and spectrometry (Fig. 4 and 5).

Fig 4.

WAC288 vesicles deliver multiple fluorescent metabolites to membranes or cytoplasm of B. subtilis. B. subtilis cells were incubated for 1 h with vesicles from WAC288 (W288), either labeled with external fluorescent dyes or unstained with native cargo of fluorescent anthracyclines. Cells were pelleted (5,000 × g, 5 min) and washed 100-fold with saline, then assessed for fluorescence. For microscopy, cells were placed on an agarose pad and imaged by confocal microscopy (red excitation 488 nm, emission 650–760 nm; green excitation 488 nm, emission 490–550 nm; or orange excitation 488 nm, emission 570–650 nm; and brightfield, T-PMT) (A and C). For spectrophotometer readings, cells were transferred to microtiter plate wells (red excitation 515 nm, emission 640 nm; green excitation 485 nm, emission 515 nm, left axis; orange excitation 500 nm, emission 590 nm, right axis) (B and D). Images grouped in panels are from parallel treatments in one experiment, placed on one agarose pad, imaged, and processed with the same settings. White bar, 5 µm. B. subtilis cells incubated for 1 h with unstained WAC288 vesicles or equal amounts of FM 4-64 (A and B), Vancomycin-Bodipy-FL (Vanc-B-FL) (C and D), or CFSE (D) labeled vesicles acquired fluorescence corresponding to the various molecular cargo. (B and D) Results shown are from technical triplicate of an experiment representative of experiments done in at least a biological duplicate. Comparisons between vesicle-treated cells and controls tested using Welch’s unpaired t-test, *P < 0.05.

Fig 5.

Streptomyces vesicles are a broad spectrum delivery mechanism for molecular cargo to pathogenic bacteria and yeast. Labeled (FM 4-64, red circles) or unstained (orange circles) vesicles from WAC288 (A–F), WAC237 (G and H), or WAC276 (I and J) were incubated with microbial cells, S. aureus (A, B, G, and H), Klebsiella pneumoniae (C and D), Cryptococcus neoformans (E and F), or C. albicans (I and J) and were able to transfer fluorescence from the molecular cargo to each cell type. Cells were pelleted (5,000 × g, 5 min) and washed 100-fold with saline, then imaged by confocal microscopy (red excitation 488 or 515 nm, emission 650–760 nm) and assayed for fluorescence with a spectrophotometer (red excitation 515 nm, emission 640 nm). Controls for expected background fluorescence were vesicles or cells alone. Results shown are technical replicates from an experiment representative of at least a biological duplicate. For results in technical triplicate, comparison between vesicle-treated cells and untreated cells tested using Welch’s unpaired t-test, *P < 0.05.

Vesicles from WAC00288 carrying the lipid-soluble dye FM 4-64 were able to transfer the red fluorescence to B. subtilis cells (Fig. 4A and B). This transfer of fluorescence was significantly greater than the negative controls of cells alone, stained vesicles alone, or an equal amount of unstained vesicles incubated with B. subtilis (Fig. 4A and B). Fluorescence was transferred to stationary or actively growing cells and could be detected even at the shortest incubation time tested of 15 min (Fig. S4A). Microscopy revealed the FM 4-64 fluorescence in a cell membrane-associated pattern on the B. subtilis (Fig. 4A). To validate this with a second molecule of different chemical composition (63), WAC00288 vesicles were labeled with vancomycin conjugated to bodipy-fluorescein; this binds tightly to lipid II in bacterial membranes through the vancomycin moiety. Again, WAC00288 vesicles were able to transfer the green vancomycin fluorescence to B. subtilis and in a membrane-associated pattern (Fig. 4C and D). This vesicle-mediated transfer and delivery of two distinct lipid-associated dyes to cells strongly support direct vesicle fusion to the cell membrane.

We then asked if the vesicle fusion could deliver the native cargo of specialized metabolites using the intrinsic fluorescent properties of anthracyclines. We reasoned that the hexaglycosylated β-rhodomycins are likely to be carried in the vesicle lumen due to the hydrophilicity of the six sugar moieties and the aqueous solubility of the purified compounds (Fig. 2D). Unstained, anthracycline-containing WAC00288 vesicles were incubated with B. subtilis. Notably, we found that the B. subtilis cells acquired anthracycline fluorescent properties (Fig. 4; Fig. S4). Confocal microscopy of the cells revealed anthracycline-associated fluorescence (red/orange), which was diffused throughout the cells, suggesting cytoplasmic localization (Fig. 4A and C; Fig. S4). This cytoplasmic location was especially clear in vancomycin-stained vesicle transfer experiments. In this case, the native anthracycline orange fluorescence could be imaged in parallel to the green vancomycin fluorescence (Fig. 4C). It was also clear when comparing fluorescence transfer from wild-type WAC00288 versus anthracycline-knockout vesicles (Fig. S4). As an informative negative control, the vesicles from WAC00288 knock-out strains did not contain or transfer anthracycline fluorescence but were able to transfer lipid-associated FM 4-64 fluorescence, suggesting membrane fusion still occurs (Fig. S4). Finally, in a further test of the ability to transfer aqueous lumenal content, WAC00288 vesicles were incubated with the membrane-permeable carboxyfluorescein diacetate succinimidyl ester, which is converted by cytoplasmically localized esterases to amine-bound fluorescent CFSE. Vesicles became fluorescent with CFSE and were able to transfer green CFSE fluorescence to B. subtilis cells (Fig. 4D).

Altogether this transfer of two distinct lipid-associated molecules and two distinct aqueous molecules confirms extracellular vesicle fusion to cell membranes and delivery of molecular cargo. Importantly, these results also demonstrate the direct delivery of vesicle cargo anthracyclines into target microbial cell cytoplasm, the cellular location necessary for antimicrobial activity.

Vesicles provide a mechanism to transfer antimicrobials to diverse microbial life forms

To determine if vesicle fusion with membranes could mediate antimicrobial delivery to a variety of organisms, including problematic pathogens, we incubated the streptomyces vesicles with further microbes, including the Gram-positive bacterium, S. aureus, the Gram-negative bacterium Klebsiella pneumoniae, and the yeast Cryptococcus neoformans. Strikingly, WAC00288 vesicles, either labeled with FM 4-64 or unstained carrying fluorescent anthracyclines, were able to transfer both the lipid-associated FM 4-64 fluorescence and anthracycline-associated fluorescence to all microbes (Fig. 5A through F; Fig. S5). Similarly, we purified WAC00237 or WAC00276 vesicles labeled with FM 4-64 and incubated them with S. aureus or C. albicans, respectively. Again, we found that the S. aureus and C. albicans acquired membrane-associated FM 4-64 (Fig. 5G through J; Fig. S5). In C. neoformans and C. albicans, the FM 4-64 is seen labeling the endocytic vesicles of the yeast due to the high endocytic turnover of yeast cellular membranes (64).

We conclude that Streptomyces extracellular vesicles are able to transfer fluorescence from multiple lipid- and aqueous-soluble molecules to diverse prokaryotic and eukaryotic microbes, most importantly from naturally packaged antimicrobials. This is strong support for the hypothesis that extracellular vesicles can serve as a delivery mechanism for specialized metabolite cargo via membrane fusion to other microbial cells. Notably, we find a versatile, broad, and permissive spectrum for fusion and content delivery. We demonstrated the transfer of multiple molecules (lipid carried vancomycin and FM 4-64, and lumenal anthracyclines and CFSE) from Streptomyces vesicles to other microbes. Metabolites can go to either the membrane (vancomycin) or the cytoplasm (anthracyclines) (Fig. 4), enabling interaction with molecular targets. This vesicle fusion and delivery were seen from three different environmental isolates, WAC00276, 00237, and 00288 (Fig. 5), suggesting it is a general property of Streptomyces extracellular vesicles. Remarkably, membrane fusion can occur in Gram-positive and Gram-negative bacteria as well as fungi (Fig. 5). Recent advances have highlighted the porous and heterogeneous nature of microbial cell envelopes, along with rapid shifts in membrane dynamics, presenting possibilities for how extracellular vesicles can both be released from and fuse with diverse cell membranes (65).

Conclusion

In summary, we examined extracellular vesicle packaging of antimicrobial-specialized metabolites in filamentous actinomycetes, and our results suggest that antimicrobial vesicle production is widespread, occurring in over 50% of strains we cultured (Fig. 1; Table S1; Fig. S1). Critically, we found that purified vesicles from four different strains are associated with highly chemically diverse specialized metabolites (Fig. 2; Fig. S1). Our results suggest that the antimicrobial activity of the vesicles is derived from the presence of the metabolite cargo (Fig. 3; Fig. S2 and S3), and we find diverse spectrums of antimicrobial activity against bacteria and fungi. We demonstrate that vesicles fuse with prokaryotic and eukaryotic microbial cells, delivering small-molecule cargo to either the cell membrane (FM 4-64 and vancomycin) or the cytoplasm (anthracyclines and CFSE) (Fig. 4 and 5; Fig. S4 and S5). Crucially, fusion was achieved in S. aureus, K. pneumoniae, C. neoformans, and C. albicans, each of which are priority pathogens that cause infections highly resistant to available antibiotics. These striking features have implications for how we understand the action of specialized metabolites in nature and for how to use them clinically.

This direct packaging and delivery of a broad variety of metabolites to a broad variety of microbes have important implications for the way we understand the action of Streptomyces-specialized metabolites in the environment. Extracellular vesicle packaging and release of specialized metabolites would protect the metabolites from extracellular degradation, and fusion to other cells would impact the effective concentration of metabolites delivered and the subsequent bioactivity. Vesicle packaging can also inform translational development. The vast majority of specialized metabolites that have been developed for clinical use have been developed as orally bioavailable drugs or injectable solutions, with the desired outcome being the systemic distribution of soluble compounds. The direct delivery via vesicle fusion of diverse antimicrobial-specialized metabolites that we observed, particularly to notoriously antimicrobial resistance species such as Klebsiella, Cryptococcus, and S. aureus (Fig. 5), suggests future therapeutic avenues. This lends credence to the development of either extracellular vesicles themselves (66–69) or lipid nanoparticle formulations (70, 71) as therapeutic vehicles for the delivery of small molecule antimicrobials to their targets. Indeed, the clinically used polyene amphotericin, closely related to candicidin (Fig. 2C), works better in a liposomal formulation; this is mostly attributed to improved pharmacokinetic and toxicity properties (72); however, it has also been suggested that the liposomal formulation improves delivery of the drug to the fungal membranes (73). With the benefits, safety, and feasibility of lipid particle formulation most conclusively demonstrated to date with the mRNA COVID-19 vaccines, the time is ripe for lipid formulations of antimicrobials to counter antimicrobial-resistant pathogens. We suggest that extracellular vesicle-packaged specialized metabolites from actinomycetes offer an attractive source of candidate molecules.

MATERIALS AND METHODS

All experiments were done at minimum in biological duplicate, with results shown from a representative experiment.

Actinomycete culture and supernatant size separation

Actinomycete strains were stored as glycerol (18%) spore stocks, and environmental isolates were from the Wright Actinomycete Collection (44). Strains were cultured with Maltose-Yeast Extract-Malt Extract (MYM) (74) medium, on agar or in liquid. For the antimicrobial vesicle screen, fresh spores were collected from 1-week agar cultures, and 10 mL of liquid MYM was inoculated with approximately 5 × 10⁶ CFU/mL and then cultured for 1 week. WAC00237, WAC00276, and WAC00288 wild-type and knock-out strains (49) were stored as aliquoted glycerol spore stocks, and liquid MYM cultures were inoculated directly from aliquots with ≈5 × 10⁶ CFU/mL. After incubation at 30°C with shaking at 220 rpm, cells were removed by centrifugation (3,000 × g, 20 min) and filtration (0.2 μm). The supernatant was either used fresh or stored at −20°C, then separated into high molecular weight and low molecular weight fractions with 100 kDa 0.5 mL centrifugal filters (Amicon Ultra) at 9,500 × g, concentrating the high molecular weight fraction 10-fold before washing 100-fold with saline. Low molecular weight filtrates were dried on a Genevac (EZ-2) for 1 h, before resolubilizing in water for 10-fold concentration.

Lipid assays with FM 4-64

Fluorescence was measured at excitation 515 nm and emission 640 nm (Synergy H1, Biotek). Matched high and low molecular weight fractions adjusted to supernatant concentration equivalents in 0.9% saline, or size exclusion chromatography fractions, were measured for fluorescence before and after the addition (15-min incubation) of 5 μg/mL of FM 4-64 (Invitrogen), and the background reading was subtracted to obtain the final measurement.

Test microbial culture and antimicrobial assays

Staphylococcus aureus ATCC 25912 (37°C), Klebsiella pneumoniae ATCC 13883 (37°C), and Bacillus subtilis 168 (30°C) were grown with Mueller Hinton II medium, and Candida albicans ATCC 90028 and Cryptococcus neoformans H99 were grown with Yeast Extract-Peptone-Dextrose medium at 30°C. The initial screen conditions were 10-fold concentrated size fractions from supernatant, 2 µL on lawns, and at 10%, vol/vol in liquid cultures in microtiter plates. Lawns were created from 1:100 dilutions of overnight cultures, liquid cultures were 1:1,000 dilutions of overnight cultures, and inhibition was measured after 18 h. To test purified vesicle fractions, samples were diluted in saline and then spotted against lawns or mixed 1:1 with liquid cultures. Liquid culture growth was measured by absorbance at 600 nm on a plate reader (Synergy H1, Biotek).

Transmission electron microscopy

Negatively charged carbon grids (CF400CU50, EMS) were incubated with samples for 60 s, washed three times with water, and then stained with 2% uranyl acetate for 30 s. Grids were imaged using a Talos L120C transmission electron microscope (software TEM Imaging & Analysis v5.0 SP4).

Vesicle purification by size exclusion chromatography (and density ultracentrifugation)

Culture supernatant was centrifuged to pellet cells (3,000 × g, 20 min), filtered through 0.2 μm membranes (PVDF or PES), and used directly or concentrated 5- to 30-fold using a 10 kDa centrifugal filter (Amicon). This concentrated fraction was then split into unstained controls and into portions incubated with fluorescent dyes. FM 4-64 was at 10 μg/mL for 15 min, Vancomycin-Bodipy-FL at 1 μg/mL for 30 min, and CFSE at 10 μM for 30 min. Samples were then applied in parallel to size exclusion columns (qEV1, 35 nm IZON Sciences) and separated with PBS elution, collecting 1 mL fractions. Fractions 1–3 after the void volume (4.5 mL) were combined as the vesicle fraction—these fractions had strong lipid signal from FM 4-64 fluorescence and high particle counts by nanoparticle tracking analysis. For WAC00288, this SEC vesicle fraction was further purified by differential gradient centrifugation. Step gradients were created using iodixanol (Optiprep), beginning with 2 mL of 15% iodixanol in PBS and layering 2 mL each of 20%, 25%, 30%, and 35% sequentially underneath. The bottom layer contained the SEC-purified vesicle fraction (unstained and FM 4-64 stained run in parallel) at 40% final iodixanol (3 mL). This was ultracentrifuged at 100,000 × g for 16 h [4°C, thin wall 13.2 mL tubes (331372), SW41Ti rotor, in an Optima XPN-80 centrifuge, Beckman Coulter], and 0.5 mL fractions were removed by pipette sequentially beginning from the top of the tubes.

Nanoparticle tracking analysis

Samples were iteratively diluted in saline and analyzed by nanoparticle tracking analysis (NanoSight NS300, Malvern) until particles were at quantifiable concentrations. For time-course comparisons, diluted samples were stored at −20°C then analyzed in one batch to keep acquisition and analysis settings consistent. Three 30-s runs were collected for each sample.

Metabolite assays, HPLC purification, and mass spectrometry

To determine the specialized metabolites responsible for antimicrobial activity, 20 mL of cell-free supernatant was extracted with 1:1 butanol (overnight, 4°C); the butanol layer was dried (EZ-2 Elite, Genevac), resolubilized in 3:1 acetonitrile to water, and injected on a C18 analytical column (250 × 4.6 mm, 5 µm, 100 Angstrom, Luna, Phenomenex) and separated on an Alliance HPLC (Waters) at 0.8 mL/min using a step gradient of increasing percentage B, from 5% to 97% over 6 min, then holding for a further 4 min: A—water + 0.1% formic acid and B—acetonitrile + 0.1% formic acid. Fractions were dried (Genevac) and resolubilized in DMSO or in water and tested for antimicrobial activity. The bioactive fractions were then examined by LC-MS/MS and assayed for absorbance and fluorescence spectral scans (Synergy H1, Biotek) to determine wavelength maxima. Absorbance maxima for the bioactive fraction from WAC00276 containing candicidin were at 390 nm, those from WAC00288 containing anthracyclines were at 495 nm, and those from WAC00237 containing actinomycin X2 were at 445 nm. Fluorescence maxima for anthracyclines were at excitation 500 nm and emission 590 nm. Following this, supernatant size fractions were extracted with butanol (1:1) overnight at 4°C, and the butanol fraction was transferred to microtiter plates for absorbance measurement (Synergy, Biotek). In SEC fractions, the absorbance was measured directly. For mass spectrometry of vesicles, either 200 µL of SEC-purified vesicles was extracted 1:1 with butanol, the butanol layer was dried and resolubilized in 50 μL 75% acetonitrile in water, or 20 μL of SEC-purified vesicles was incubated with 60 μL of 1:1 isopropanol and acetonitrile for 10 min. The precipitate was pelleted by centrifugation at 16,000 × g for 10 min, and 10 μL of the supernatant was injected into an Acquity UPLC BEH C18 column (1.7 µm, 2.1 × 50 mm) in parallel with blank controls of solvents. The separation gradient was at 0.2 mL/min, 5% B for 1 min, increasing linearly to 95% over 12 min, and held at 95% for 4 min: A—water + 0.1% formic acid and B—acetonitrile + 0.1% formic acid. Samples were analyzed by electrospray ionization with MS^e on a QTOF (Waters Xevo G2S-QTOF).

Fluorescence transfer assays

Matched unstained and stained vesicle fractions purified by SEC were incubated 1:1 with liquid cultures (overnight cultures were diluted 10- or 100-fold, incubated for 2 h, then mixed with samples) or blank medium, in technical duplicate or triplicate. Cells were pelleted (5,000 × g, 5 min), washed twice with saline (100× wash), resuspended, and fluorescence was read in microtiter plates (Synergy H1, Biotek). Excitation, emission (nm): 515, 540 FM 4-64; 500, 590 anthracyclines; and 485, 515 Vanc-Bodipy-FL and CFSE. Experiments were repeated in at least a biological duplicate with different batches of vesicles. The results were consistent with vesicles purified fresh or stored at −20°C or −80°C. After a freeze-thaw, vesicle samples were spun at 5,000 × g for 5 min, and the supernatant was transferred to a clean tube to ensure large aggregates of vesicles were removed.

Confocal microscopy

Washed cells from fluorescence transfer assays were spotted onto agarose pads (1.5% in saline) and imaged with a 100× oil objective on a Zeiss LSM 880 Elyra. Images were captured and adjusted for brightness and contrast using Zen (v2.1 SP3). For comparative analyses, samples were spotted onto the same agarose pad, and imaging and software settings were constant between samples. Biological duplicates were imaged for each condition, and all samples had multiple images captured across the spread of the cells on the agarose pad.

SUPPLEMENTAL MATERIAL

The following material is available online at https://doi.org/10.1128/jb.00325-23.

Supplemental material. jb.00325-23-s0001.pdf.

Table S1; Figures S1 to S5.

ASM does not own the copyrights to Supplemental Material that may be linked to, or accessed through, an article. The authors have granted ASM a non-exclusive, world-wide license to publish the Supplemental Material files. Please contact the corresponding author directly for reuse.