ABSTRACT
Among the plethora of unusual secondary metabolites isolated from Stachylidium bicolor are the tetrapeptidic endolides A and B. Both tetrapeptides contain 3-(3-furyl)-alanine residues, previously proposed to originate from bacterial metabolism. Inspired by this observation, we aimed to identify the presence of endosymbiotic bacteria in S. bicolor and to discover the true producer of the endolides. The endobacterium Burkholderia contaminans was initially detected by 16S rRNA gene amplicon sequencing from the fungal metagenome and was subsequently isolated. It was confirmed that the tetrapeptides were produced by the axenic B. contaminans only when in latency. Fungal colonies unable to produce conidia and the tetrapeptides were isolated and confirmed to be free of B. contaminans. A second endosymbiont identified as related to Sphingomonas leidyi was also isolated. In situ imaging of the mycelium supported an endosymbiotic relationship between S. bicolor and the two endobacteria. Besides the technical novelty, our in situ analyses revealed that the two endobacteria are compartmentalized in defined fungal cells, prevailing mostly in latency when in symbiosis. Within the emerging field of intracellular bacterial symbioses, fungi are the least studied eukaryotic hosts. Our study further supports the Fungi as a valuable model for understanding endobacterial symbioses in eukaryotes.
IMPORTANCE The discovery of two bacterial endosymbionts harbored in Stachylidium bicolor mycelium, Burkholderia contaminans and Sphingomonas leidyi, is described here. Production of tetrapeptides inside the mycelium is ensured by B. contaminans, and fungal sporulation is influenced by the endosymbionts. Here, we illustrate the bacterial endosymbiotic origin of secondary metabolites in an Ascomycota host.
INTRODUCTION
The term symbiosis, as defined by Anton de Bary in 1879, implies an intimate association between organisms of two or more species, broadly applied to a spectrum of beneficial (mutualism), neutral, or harmful relationships, including parasitism (1). Symbioses are ubiquitous in nature (2) and have the potential to influence the dynamics of natural populations by altering host ecology (3). From protists to humans, all animals and plants are inhabited by microbes, which comprise most of life's diversity (4). Symbiotic partners are usually regarded as separate individuals, hampering the assessment of interactive mechanisms within these holobiont systems (4–6).
Several studies have shown that prokaryotes live as symbionts in the intracellular space of hosts from Protozoa, Plantae, Animalia (7), and, less frequently documented, in Fungi (8). Endofungal bacteria are symbionts present inside mycelium, which can be horizontally transmitted or complete their life cycles within the host, being transmitted vertically from generation to generation (8, 9). They remain overlooked, mostly because they lack obvious phenotypes, and universal tractable experimental systems are yet to be established, including the means to cultivate the endobacteria in pure culture (10–12). Glomeromycotina, Mucoromycotina, Basidiomycota, and Ascomycota hosts harbor endobacterial symbionts (13). So far, the best-studied fungal endobacteria belong to the family Burkholderiaceae, which are associated with early-diverging lineages of terrestrial fungi within the Mucoromycotina (14). Widespread endobacterial existence in fungi has been suggested, either in temporary or long-term and intimate associations (15).
The sponge-derived fungus Stachylidium bicolor 293K04 (Ascomycota, Chaetosphaeriales) was shown to produce a wide variety of new secondary metabolites, comprising unusual polyketides (marilines A to C, marilones A to C, and derivatives), tyrosine derivatives (stachylines A to D), and the recently reported tetrapeptides endolides A to D (16–21). Endolides contain the residue 3-(3-furyl)-alanine, so far reported only twice in the literature. The first instance was in heptapeptides isolated from the Mucoromycotina fungus Rhizopus microsporus (22), which were later demonstrated to be produced by its endosymbiont Burkholderia endofungorum (23). The second instance was in bingchamides isolated from Streptomyces bingchenggensis (24). Both of these reports strongly support a bacterial biosynthetic background for the origin of this rare amino acid.
The challenge of the present study was to identify a potential endobacterium of Stachylidium bicolor 293K04 as the true producer of the endolides. We successfully isolated a bacterial strain from its mycelium. This strain was identified as Burkholderia contaminans, which can produce, under axenic conditions, the endolide A as well as onychocin D, the latter here described as a new natural product (isolation from fungal cultures). These results and the detection of a second endobacterium closely related to Sphingomonas leidyi suggest novel features in symbiotic biology, with endobacteria prevailing in a state of latency in defined fungal cells, some of which may participate in their vertical transmission.
RESULTS
Dissection of the putative holobiont Stachylidium sp. strain 293K04.To complete the taxonomic classification of the marine-derived fungal strain Stachylidium sp. 293K04 (20), we amplified the internal transcribed spacer (ITS)/28S rRNA region of all Stachylidium strains available in public culture collections, all of which are terrestrial (see Table S1.1 in the supplemental material). The obtained sequences were compared to that of Stachylidium sp. 293K04. Morphological features such as conidiophore and conidium size (data not shown) were also compared, resulting in its taxonomical identification as Stachylidium bicolor (Fig. 1).
Consensus tree from Bayesian-phylogeny inferences based on ITS/28S rRNA sequences of selected Stachylidium bicolor strains and related genera of Plectosphaerellaceae. Clade probability values are indicated at the branches. Ceratocystis fimbriata C1004 was used as the outgroup. Analysis of all Stachylidium bicolor strains available in public culture collections: black diamond, chemotaxonomic (tetrapeptide production); black square with white circle, “symbiotaxonomic” (detected presence of symbiotic Burkholderia sp.). The capacity of each strain to produce endolides A and B and onychocin D was verified by UVLC-MS analysis of their culture organic extracts (see Table S6). Symbiotic association with Burkholderia sp. was performed by 16S rRNA partial gene amplification from the respective conidial metagenomic DNA.
Stachylidium bicolor was previously shown to produce endolides A and B (Fig. 2) in solid and liquid media following cultivation periods of ca. 40 days (17, 25). Here, we isolated from the culture extracts of S. bicolor also the new natural product onychocin D (see details in Fig. S2.1 to S2.5 in the supplemental material). Onychocin D is a tetrapeptide lacking 3-(3-furyl)-alanine residues and corresponding to the major-yield tetrapeptide in the organic extracts from fungal liquid cultures (Fig. 2, compound 3).
Chemical structures of the tetrapeptides isolated from the marine-derived fungus S. bicolor 293K04: endolides A and B (compounds 1 and 2, respectively) (17) and onychocin D (compound 3): (−)-cyclo-[N-methyl-l-phenylalanyl, l-leucyl, N-methyl-l-phenylalanyl, l-valyl] (full spectroscopic data for compound 3 are available in Section S2 in the supplemental material).
Identification and isolation of endosymbiont bacteria.To test the hypothesis that a bacterial endosymbiont was present, we amplified 16S rRNA genes by PCR using the metagenomic DNA extracted from either conidia (fungal asexual spores) or mycelium of S. bicolor grown on solid BMS medium (see “Fungal culture conditions” in Materials and Methods). Both of the amplicon sequences clustered with bacteria from the Burkholderia cepacia complex, a group of species that cannot be distinguished by 16S rRNA gene comparison (see Fig. S1.1 in the supplemental material).
Standard bacterial isolation assays were extensively applied, which included different fungal cultivation conditions (e.g., pH, medium composition, and additives) and different isolation media (more than 400 independent trials), but no bacteria could be isolated. We successfully recovered a bacterial pellet from 15-day-old S. bicolor mycelium grown on oatmeal solid medium by exposure to tunicamycin (500 μg/ml) during the imposition of mechanical shear stress. Endobacteria were then recovered using 0.8-μm filters that retain the sheared fungal biomass but not bacteria. In these trials, isolation of endobacteria was successful when targeting immature conidiophores and hyphal tips, harvested from the margins of a growing colony. Eight colonies were isolated, and the taxonomic identity of all the isolates was determined by 16S rRNA gene sequencing, matching that of Burkholderia spp.
In addition, a second bacterial colony type was isolated from an independent 12-day-old culture of S. bicolor, grown on oatmeal solid medium. This colony type was identified as being closely related to Sphingomonas leidyi by 16S rRNA amplicon sequencing.
Taxonomic placement of Burkholderia sp. strain 293K04B and its environmental presence.The phylogenetic placement of the Burkholderia sp. strain 293K04B was estimated by multilocus sequence analysis (MLSA) based on the concatenation of 7 housekeeping genes (atpD, gltB, gyrB, recA, lepA, phaC, and trpB). Strain 293K04B formed a monophyletic group within the Burkholderia cepacia complex, and its closest relatives were B. contaminans LMG 23361T (100% sequence similarity) and B. lata ATCC 17760T (99.6%) (Fig. 3; see Fig. S1.2 in the supplemental material). In silico DNA-DNA hybridization experiments confirmed that strain 293K04B is a member of the species B. contaminans, with a reassociation of >96% with its closest relative, B. contaminans LMG 23361T (see Table S1.2 in the supplemental material). To understand the environmental distribution of bacteria phylogenetically related to B. contaminans 293K04B, we interrogated its 16S rRNA gene sequence using the data sets of all 16S rRNA gene amplicons available in the IMNGS (integrated microbial next generation sequencing) database (99% gene sequence similarity; ≥500 nucleotides [nt]) (26). The results show that they inhabit a wide range of ecological niches, including human tissues, soil, and aquatic environments, and as symbionts of other organisms (see Table S1.3 in the supplemental material).
Maximum likelihood (ML) tree illustrating the phylogenetic position of the Burkholderia sp. strain 293K04B and of related members of the Burkholderia cepacia complex, based on the concatenated sequences (2,773 bp) of seven housekeeping gene fragments: atpD (443 bp), gltB (400 bp), gyrB (454 bp), recA (393 bp), lepA (397 bp), phaC (385 bp), and trpB (301 bp). Gene sequences of the type strains were downloaded from the Burkholderia cepacia complex MLST databases (https://pubmlst.org/bcc/). Identities (IDs) of strains are provided in parentheses. The optimal evolutionary model of nucleotide substitution applied is T1M1 + G (gamma shape = 0.6490) + I (proportion of invariable sites = 0.5590). Bar, 0.05 fixed nucleotide substitutions per site. Bootstrap values above 50% (1,000 resamplings) are indicated at the branching points. Burkholderia terrestris LMG 22937T was used as the outgroup.
Tetrapeptide production detected in axenic B. contaminans cells in latency.Inspired by the bacterial origin of 3-(3-furyl)-alanine residues previously reported for peptides produced by an endofungal Burkholderia species, the capacity of B. contaminans to axenically biosynthesize the endolides A and B and onychocin D (Fig. 2) was investigated. We first screened organic extracts derived from B. contaminans cultures grown under very diverse conditions, varying the cultivation parameters, the growth medium formulations, and the presence of specific metabolic modulators (details are shown in Table S3.1 in the supplemental material). None of the resulting culture extracts yielded detectable amounts of the tetrapeptides by UV liquid chromatography-mass spectrometry (UVLC-MS) analysis, suggesting that their production might be cryptic in standard axenic liquid cultures. We then decided to explore the hypothesis that the endobacteria are mostly in a metabolically active but latent (nonculturable) state when in symbiosis. To test this hypothesis, we grew B. contaminans in liquid Trypticase soy broth (TSB) medium supplemented with 1% glycerol (vol/vol), which we had previously observed to increase the bacterial biomass up to 3-fold (our unpublished data). The presence of culturable cells in this seed culture was monitored daily by subculturing onto solid TSB medium. After 14 days of cultivation, we assumed that a putative latent state had been attained since culturable bacteria were no longer present. These cells were then placed at 4°C for 4 days. Their ability to produce the endolides was confirmed by incubating the cells under a variety of conditions for 2 days (see Table S3.2 in the supplemental material), followed by UVLC-MS analysis of the ensuing organic extracts. Under 6 of the 17 conditions tested, the tetrapeptides onychocin D and endolide A (Fig. 4) were detected. The successful conditions included incubation at three different pHs, TSB-diluted medium or with 40 μg/ml ciprofloxacin, and CaCO3 in water: the presence of the two tetrapeptides in one representative extract was validated using high-resolution (HR)/UVLC-MS; i.e., the derived chemical formulae and retention time matched those of the pure compounds (see Fig. S4.1 in the supplemental material). In addition, two other major compounds were consistently detected in B. contaminans cultures, from either latent or growing cells. These were predicted to be isomers of the siderophore pyochelin (27) with the chemical formula C14H16N2O3S2 (see Fig. S4.2 in the supplemental material).
UVLC-MS ion extraction chromatograms of the pure compounds onychocin D (a) and endolide A (b) and of an organic extract of Burkholderia contaminans cells in latency incubated in TSB diluted 1:10 for 2 days (c1 and c2; see LC-MS data in Fig. S4.1); UVLC-MS detection by pseudo-molecular-positive-ion extraction of onychocin D [ion (M+H)+ = 535+; retention time, 5.7 min] and endolide A [ion (M+H)+ = 515; retention time, 5.0 min].
In situ detection of endosymbiont bacteria.In conidia, we failed to obtain in situ hybridization signals, probably due to the poor permeability of their rigid cell walls. Therefore, to assess the potential vertical transmission of endobacteria, we performed in situ hybridization assays on a 3-day-old germinated conidium grown on solid BMS medium deposited onto a glass slide. Hybridization of the universal eubacterial probe EUB338 revealed that the germinated conidium harbored endobacteria (Fig. 5). In addition, we observed isolated hyphae potentially enriched with endobacteria (Fig. 5b and c, white arrows), the spacing of which suggests one bacterium cell per septate fungal cell. Costaining with the Non-Eub probe (i.e., negative control) confirmed the specificity of the EUB338 probe hybridization (see Fig. S5 in the supplemental material).
Microscopic organization of the endosymbiotic system of S. bicolor in germinated conidia. (a) Light microscopy of conidiophore structure and conidia on corn meal agar (bar, 10 μm); (b and c) in situ hybridization of the oligonucleotidic probe EUB338 in 3-day-old germinated fungal conidia demonstrating abundant signals from endobacteria as well as the presence of isolated hyphae harboring endobacteria, of which the spacing suggests one bacterium per septate fungal cell (white arrow); (b) differential interference contrast (DIC) channel overlaid with the 561-nm channel (EUB338); (c) 561-nm channel (EUB338). The sample was also costained with the Non-Eub probe (negative control), which revealed an absence of nonspecific binding (Fig. S5).
To observe and maintain the natural spatial arrangement of the margins of a mycelial network growing over a substrate, the mycelium was grown for 15 days over a porous layer of cellophane covering solid oatmeal medium. In addition to the EUB338 probe, we also used the genus-specific probes BUR (yellow) and SPH120 (blue) to covisualize endobacteria of the genera Burkholderia and Sphingomonas, respectively (Fig. 6). A small number of endobacterial “hot spots” were repeatedly observed in hyphal tips and in young mycelium showing both genus-specific (BUR and SPH120) hybridization signals and signals from the eubacterial probe EUB338 (Fig. 6a to d). In situ hybridization assays with SPH120 revealed a different type of organization in the inner part of the mycelium of a 12-day-old colony grown on solid oatmeal medium. Specifically, we observed the enriched and regular distribution of Sphingomonas in mature hyphae (Fig. 7), the spacing of which suggest one bacterium cell per septated fungal cell, in an arrangement similar to that observed in the imaging of young hyphae (Fig. 5c).
Microscopic organization of the endosymbiotic system of S. bicolor at the edges of a growing colony. In situ hybridization assays using the probes BUR, SPH120, and EUB338 on a 15-day-old mycelium grown over a layer of cellophane, showing endobacterial “hot spots” (EBH) in hyphal tips and in young mycelium, with colocalization of hybridization signals from Burkholderia sp. (yellow) and Sphingomonas sp. (blue), as well as from the general eubacterial probe EUB338 (red). (a) Differential interference contrast (DIC) channel overlaid with the 561-nm channel (EUB338); within dashed circles are the endobacterial hot spots; the dashed curved line indicates the hyphal front; the gray background corresponds to the cellophane layer. (b) EUB338, 561-nm channel; (c) BUR, 488-nm channel; (d) SPH120, 405-nm channel.
Microscopic organization of the endosymbiont Sphingomonas in mature mycelium of S. bicolor. In situ hybridization assays using the specific probe SPH120 on a 12-day-old S. bicolor mycelium showed the presence of enriched hyphae carrying endobacteria: the regular distance between the fluorescent signals suggest one fluorescent signal per septate fungal cell. (a) DIC channel overlaid with the 405-nm channel (SPH120); (b) 405-nm channel (SPH120).
Tetrapeptides and Burkholderia sp. in the terrestrial strains of S. bicolor.To determine the specificity of the symbiotic association between B. contaminans and the marine-derived S. bicolor, we undertook a chemotaxonomic analysis of all of the seven terrestrial strains of S. bicolor available in public collections. The presence of the tetrapeptides was detected consistently by UVLC-MS in the ensuing organic culture extracts (Fig. 1; see also Table S6 in the supplemental material). As negative controls, we also analyzed the culture organic extracts of the two strains of Plectosphaerelaceae included in the phylogenetic Bayesian tree of S. bicolor (Wallrothiella subicullosa JCM 23118 and Plectosphaerella cucumerina NBRC 9985), in which no tetrapeptides could be detected.
We successfully amplified 16S rRNA partial gene fragments from six of seven conidial S. bicolor metagenomes. These partial 16S rRNA gene sequences clustered within Burkholderia spp. and were closely related to B. contaminans, which was found as a best match in the EZTaxon database (Fig. 1).
Isolating S. bicolor 293K04 morphotypes unable to form conidia and potentially cured of B. contaminans.Attempts to cure S. bicolor of B. contaminans by treatment with 10 different antibiotics were unsuccessful (see Fig. S7 in the supplemental material). Ciprofloxacin was previously shown to cure R. microsporus strains from their Burkholderia endosymbionts (23, 28). Ciprofloxacin, at a concentration of 50 μg/ml, was sufficient to inhibit the growth of the axenic B. contaminans. However, prolonged fungal cultures in BMS medium supplemented with high doses of ciprofloxacin still produced tetrapeptides at concentrations similar to those found in nontreated cultures: (i) 200 μg/ml in solid medium for 6 months with 7 subcultures and (ii) 100 μg/ml in liquid medium for 1 month with 6 subcultures (Fig. S7). In addition, amplification of the 16S rRNA gene fragments from DNA extracted from the mycelium of the antibiotic-treated solid cultures revealed the prevalence of B. contaminans.
Rhizopus microsporus was reported to lose the ability to sporulate when cured from its Burkholderia species endosymbionts (23, 28). Furthermore, in situ imaging of the bacterium endosymbiont in mycelium of the basidiomycete Laccaria bicolor S238N showed that most hyphae did not harbor endobacteria (29). Inspired by these observations, we decided to attempt the isolation of S. bicolor morphotypes unable to produce conidia, as these could be potentially devoid of tetrapeptide production and potentially also free of its Burkholderia endobacterium. Therefore, a chemical-free mechanical microhypha fragmentation and isolation strategy was employed. Specifically, S. bicolor was grown under a range of different conditions and in different nutrient media under a mild shear stress (e.g., liquid media in Erlenmeyer flasks with baffles) followed by subculturing of hyphal microfragments onto solid medium to select colonies unable to produce conidia. Only seven colonies lacking conidium production (Fig. 8) and unable to produce the tetrapeptides were isolated, all of which originated from the same cultivation condition (see Fig. S8 in the supplemental material). We also attempted the amplification of 16S rRNA genes from DNA extracted from the mycelium of each of these colonies: no bacterial amplicons could be detected.
Stereomicroscopy imaging of colonies of S. bicolor wild-type strain (a, b) and a morphotype unable to produce conidia isolated by hyphal microfragmentation (c, d) grown for 23 days on solid media (potato dextrose medium and oatmeal solid medium, left and right panels, respectively). On both of the solid medium types, the typical coloration of the conidia is lacking in the isolated phenotype.
DISCUSSION
The results presented here support the proposal of a symbiosis of S. bicolor with its endobacterium B. contaminans based upon the consistent detection of endobacterium-derived tetrapeptides in the eight S. bicolor strains and the detection of Burkholderia species 16S rRNA gene partial sequences in their corresponding metagenomes (with the exception of one strain, which may be below the threshold for 16S rRNA amplification [Fig. 1]). Our results support that the tetrapeptides are a chemotaxonomic marker of the symbiosis between the host S. bicolor and its endobacterium B. contaminans. The hypothesis of a long-standing symbiotic relationship between S. bicolor and B. contaminans deserves further consideration.
The use of tunicamycin led to the isolation of endobacterial colonies from S. bicolor, subsequently identified as B. contaminans (Fig. 3) and related to S. leidyi. We also observed its efficiency for the isolation of other symbiotic bacteria from additional fungal hosts (our unpublished data). Among its known properties, tunicamycin functions as an antibiotic due to its capacity to inhibit the biosynthesis of teichoic acids in Gram-positive bacteria (30); these glucan-polymers are absent in Gram-negative bacterial cell walls such as Burkholderia spp. Tunicamycin is also a protein glycosylation inhibitor that acts as an endoplasmic reticulum (ER) stress inducer in eukaryotic cells (31), also known to trigger and mediate mitochondrial fission (32, 33). Imaging of the Mortierella elongata mycelium showed its endosymbiont Mycoavidus cysteinexigens surrounded by a tangled complex of membranes, possibly with the endobacterium undergoing cell division (34). This observation suggests a physical interaction of the endobacteria with the fungal ER, similar to that reported for the ER-transient interaction with mitochondria (MAM), chloroplasts, and eukaryote-endosymbiotic bacteria (for examples, see references 35, 36, 37, and 38). A similar hint can also be seen in the earliest transmission electron microscopy (TEM) images of endobacteria harbored in fungal cells, where an interaction of membrane structures from the host with the endobacteria cannot be discounted (23, 39–41). All these potential activities of tunicamycin may be involved in the release of the endobacterium from the mycelium of the host, yet the mode by which this compound assists the release of endobacteria remains unknown and requires further investigation.
The in situ imaging studies allowed for the identification of (i) endobacteria harbored in 3-day-old germinated conidia (Fig. 5), (ii) a small number of endobacterial hot spots, i.e., fungal cells containing both Sphingomonas and Burkholderia (Fig. 6), and (iii) hyphae enriched with a regular distribution of Sphingomonas (Fig. 7). The discovery of the endobacterial hot spots in the hyphal tips and young mycelium required the in situ hybridization of probes in mycelium while preserving its natural spatial structural network. The presence of only a small number of endobacterial hotspots (Fig. 5) shares similarities with the distribution of Paenibacillus sp. in the mycelium of Laccaria bicolor, which was observed only in the terminal cell of some hyphae, i.e., not spreading throughout the mycelium (29).
These organizational features, along with the presence of endobacteria in early-germinated conidia, are consistent with an ordered transport of endobacteria in the mycelium during the colonization of new substrates, possibly to ensure their vertical transmission to conidia and a rapid response to external cues in a changeable environment. The hypothesis of vertical transmission of endobacteria in fungal hosts has been suggested in studies focusing on the R. microsporus symbiotic models (42, 43). It remains to be addressed if the endobacterial hot spots found in the young mycelium of S. bicolor (Fig. 5) are also present in the dense mature mycelium, where the genesis of conidiophores mostly occurs.
An enriched and regular distribution of Sphingomonas was observed in hyphae of mature mycelium of S. bicolor (Fig. 7). The presence of one Sphingomonas cell per septated fungal cell is suggested based upon the regular separation of bacterial cells harbored in a hypha, despite the fact that the septa cannot be visualized in these images. A similar distribution was also observed when we used the general eubacterial probe EUB338 (Fig. 5, white arrow); this pattern was not yet observed when using the specific Burkholderia probe. The high yields of tetrapeptides that were isolated from mycelium grown on solid medium (17) would require a high density of B. contaminans cells to be harbored in the fungus. The hypothesis that certain hyphae carrying a high-cell density of one bacterium species is a conserved phenomenon in septate fungi deserves close attention in future.
In the present study, tetrapeptide production was never observed in standard axenic B. contaminans cultures; however, production was triggered following the induction of a putative latent state (Fig. 4). Production of onychocin D and endolide A was not observed in all cases, possibly due to low metabolite production or the requirement for a specific physiological state. So far, production of secondary metabolites in endofungal bacterial axenic cultures has been reported only twice: Burkholderia endofungorum axenically produced low yields of the heptapeptide rhizonin A (295 μg/liter), but rhizonin B could not be isolated (23); whereas B. rhizoxinica axenically produced the polyketide rhizoxin, losing this capacity soon after subculturing (28). Ciprofloxacin was inefficient to cure the fungus from B. contaminans, regardless of the fact that lower doses efficiently inhibited the growth of this bacterium when dividing in axenic cultures. Moreover, B. contaminans cells in a latent state were able to produce the tetrapeptides even when exposed to 40 μg/ml of ciprofloxacin. Together, these observations suggest that B. contaminans cells are mostly in latency when in symbiosis with the fungus, ensuring the production of the tetrapeptides. Bacterial cells in latency have been associated mostly with a survival state, e.g., the viable but nonculturable (VBNC) state (44, 45), and with an asymptomatic infection state, e.g., in Mycobacterium tuberculosis (46). Further studies are necessary to understand the metabolic traits of B. contaminans cells when in latency, especially when in an endosymbiotic association with the fungal host.
After extensive trials, we isolated seven mycelial colonies that were unable to produce both conidia and the tetrapeptides (Fig. 8). As previously suggested in other endosymbiotic associations, namely, for Burkholderia spp. harbored in R. microsporus hosts (23, 28), B. contaminans might control S. bicolor sporulation by an unknown mechanism. These nonsporulating fungal morphotypes constitute an exciting tool to further study the symbiosis of Burkholderia spp. in fungal hosts. These observations are consistent with the initial idea that the endobacterium localizes in some hyphae but does not spread throughout the mycelium, as suggested by the fluorescence in situ hybridization (FISH) imaging of mycelium from S. bicolor (Fig. 6 and 7) and from L. bicolor (29). Extensive studies are required to understand if our holobiont system can be reconstituted by colonizing the cured fungal phenotypes with axenic B. contaminans and/or S. leidyi, which is essential for the further unveiling of endosymbiotic features in superior fungi, in particular how endobacteria influence sporulation of the host.
Antibiotics, including ciprofloxacin at high doses and for up to 6 months, were ineffective at curing our fungal host from its endosymbiotic B. contaminans. Two aseptate R. microsporus strains were successfully cured of their endosymbiotic Burkholderia spp. using low concentrations of ciprofloxacin (23, 28). In addition, two basidiomycete hosts, namely, Piriforma indica (47, 48) and Ustilago maydis (49), could not be cured of their endosymbiotic bacteria when exposed to antibiotics. One feature in common with our ascomycete S. bicolor system is that both of these basidiomycete hosts have septate hyphae. Our in situ observations, specifically, the suggestion of hyphae enriched with S. leidyi (Fig. 7) combined with the isolation of morphotypes devoid of conidia that were free of tetrapeptide production and of Burkholderia (inferred by the 16S rRNA assays), reinforce the earlier suggestion of a conserved phenomenon in the organization of endosymbiont bacteria in septate fungal hosts. It will be critical to a better understanding of this mechanism to reveal if enriched hyphae originate from bacterial hot spots as a response to environmental cues that activate specific endobacterium growth and symbiotic function.
Burkholderia species are a remarkably diverse bacterial lineage and reside in a wide variety of niches, e.g., in soil, in the rhizosphere, as plant endosymbionts, or as human pathogens (50). Comparing the 16S rRNA sequencing data of our strain with publicly available metagenomes reinforces the existence of a wide distribution of species identical/similar to B. contaminans in several ecological niches (Table S1.3). Burkholderia contaminans was originally isolated from the Sargasso Sea after a metagenomic study yet was considered a contaminant because species of this genus were usually assumed to be terrestrial and grow poorly in seawater (discussed in reference 50).
Conclusion.The present study supports that B. contaminans is an intimate, possibly permanent, intracellular symbiont widespread in S. bicolor strains. Collectively, our data also reveal that the endosymbiont B. contaminans is responsible for the production of tetrapeptides isolated from the mycelium of S. bicolor. Furthermore, it suggests that B. contaminans exists mostly in a latent state when in symbiosis with the fungal host and that tetrapeptide biosynthesis likely occurs only in this state. The isolation and in situ visualization of a second endosymbiont bacterium, S. leidyi, provided invaluable information, supporting the existence of bacterial hot spots in hyphal tips and young mycelium, as well as the existence of an enriched mycelium. This second endobacterium does not appear to be linked to the production of the tetrapeptides. Several open questions should be addressed in the near future. What is the relationship between the two endobacteria within the fungal host? What is the symbiotic function of S. leidyi? Which bacterium is responsible for the loss of fungal sporulation? Finally, what is the ecological significance of the B. contaminans tetrapeptides, specifically for the fitness of the fungal host? We believe that the data presented here, together with the technical progress described, will help to promote the Fungi as a model kingdom for understanding the symbiotic biology of intracellular bacterial symbioses, especially how they impact evolutionary and ecological aspects of both the eukaryotic host and the intracellular prokaryotes.
MATERIALS AND METHODS
Fungal strain.The marine-derived fungal strain Stachylidium sp. strain 293K04 was isolated from the sponge Callyspongia flammea (collected at Bare Island, Sydney, Australia, in 1997). The fungal strain is deposited at the Institute for Pharmaceutical Biology, University of Bonn, Bonn, Germany. All Stachylidium strains available in public culture collections, namely, CBS 121802, CBS 449.88, CBS 292.72, IHEM 20007, NBRC 8948, NBRC 8949, and CF 023127 (Fig. 1; see also Table S1.1 in the supplemental material), were used to complete its taxonomic identification, along with morphological comparative studies.
Fungal culture conditions.Fungal cultures on solid media (20 g/liter agar; Fisher) were inoculated with 5-mm-diameter agar plugs of pregrown mycelia or by streaking conidia and incubated in Kuhner incubators (25°C, 70% humidity, without agitation) for 12 to 37 days. Culture media were BMS (biomalt with added artificial seawater: Villa Natura Kraftnahr biomalt, 20 g/liter; sea salts [Sigma], 40 g/liter), TSB (Trypticase soy broth; BD, Difco), or oatmeal (BD, Difco). Fungal cultures on liquid media (10 ml, EPA vials) were performed as previously described (25). Conidia from 25-day-old cultures grown on solid oatmeal were gently harvested with a loop using a saline (0.8% NaCl) solution containing 0.0025% Tween 80.
For chemotaxonomic comparative studies, S. bicolor strains (eight in total) and other Plectosphaerelaceae strains, viz. Wallrothiella subiculosa and Plectosphaerella cucumerina (same family as Stachylidium), were cultivated in BMS liquid (10 ml in EPA vials, one culture per time point at 12, 20, 28, and 40 days) or solid media (incubated as described above and harvested after 37 days of incubation).
Bacterial culture conditions.Colonies grown in R2A (Reasoner's 2A agar; Difco) or TSA (Trypticase soy agar; BD, Difco) solid medium (28°C, 70% humidity) were used to inoculate seed cultures, 50 ml in TSB medium, incubated during 2 to 3 days under controlled conditions (28°C, 70%, with agitation at 220 rpm), which were then used as inocula (3% [vol/vol] final culture volume) for different growth media (50-ml cultures in 250-ml Erlenmeyer flasks or 10-ml cultures in 40-ml EPA vials) and incubated at 28°C, without or with (220 rpm) agitation, for ≥2 days.
Organic extracts of microbial cultures.Fungal liquid cultures were extracted as previously described (25). Bacterial liquid cultures (10 ml) were extracted with methyl ethyl ketone (MEK) (2:1) under vigorous shaking; dimethyl sulfoxide (DMSO) was added to the organic fraction (500 μl), and MEK was evaporated under N2 flow; then, the cultures were concentrated 20× from the initial culture volume in DMSO; finally, the supernatants were cleared of cell debris by centrifugation (10 min, 1,467 × g). Solid bacterial/fungal cultures were first crushed, placed into EPA vials, and then extracted with acetone (1:1) and processed as described above for the fungal liquid cultures.
Isolation and identification of onychocin D.For isolation and identification of onychocin D, culture, extraction, chromatographic conditions, and structure elucidation were as described before (17). Briefly, to isolate compound 3 (Fig. 1), the vacuum liquid chromatography (VLC) fraction 3 was resolved first by normal-phase (NP)-HPLC (petroleum ether-acetone, 7.2:1; fraction 7 of 9) and then by RP-HPLC (75% methanol [MeOH], 6.6 mg; retention time, 13 min).
Isolation of endofungal symbiotic bacteria from the metafungus.Endofungal bacteria were isolated from 12- to 15-day-old mycelium of S. bicolor grown on oatmeal agar (ca. 200 mg) that was mechanically shear stressed in a volume of 500 μl containing tunicamycin (500 μg/ml; Sigma-Aldrich) for ca. 15 to 20 s and then filtered through 0.8-μm sterile filters (GVS). The endobacteria in the filtrate were recovered by centrifugation (1 min, 16,016 × g). The bacterial pellet was resuspended in 100 μl water and then plated on R2A medium plates, which were incubated (28°C, 70% humidity) and observed daily until colonies were visible.
Acquiring latent/nonculturable B. contaminans 293K04B cells.The cultivability of stationary-phase bacterial cells in TSB medium containing 1% glycerol (50-ml cultures) was monitored daily by streaking the liquid culture onto solid R2A, until bacterial growth was not detected (ca. 14 days), i.e., the culture was putatively in a metabolically viable but latent state. The pelleted cells (5-fold concentrated) were kept at 4°C and were then added (final suspension volume, 3%, vol/vol) to 10-ml liquid suspensions. These included a nutrient medium array comprising 17 distinct formulations (Table S3.2) to test the activation of secondary metabolism in the putative latent cells. These were then incubated for 2 to 3 days under stable conditions (28°C, 70% humidity, 220 rpm) before assessment of capacity to produce endolides (see below).
Identification of microbial secondary metabolites in the extracts.Extracts were analyzed by UVLC-MS as described previously (51) using the pure tetrapeptides as references. A likelihood match for compounds in the organic extracts used LC high-resolution mass spectrometry (HR-MS) as previously described (52).
Strategies to isolate S. bicolor mycelium unable to produce tetrapeptides.Preliminary assays to cure S. bicolor of its endobacteria used the following antibiotics (Sigma-Aldrich): ampicillin, penicillin G, rifampin, tetracycline, cephalothin, kanamycin, polymyxin B, erythromycin, novobiocin, and ciprofloxacin. Antibiotic culture assays were performed generally as described below for ciprofloxacin in solid cultures. The fungus was cultivated in BMS medium supplemented with ciprofloxacin at 200 and 100 μg/ml for solid and liquid media, respectively. Each of the seven subcultures in solid medium was done after 25 to 30 days, which included a parallel set without antibiotics (control). In the liquid cultures, mycelium was replicated weekly (six cycles). UVLC-MS extracts of the S. bicolor control and the 7th ciprofloxacin subculture were compared, and metagenomic DNA of the latter was extracted to determine the presence or absence of B. contaminans.
A chemical-free method of microhyphal isolation was implemented as follows. A range of cultivation parameters were tested: (i) different seed medium formulations; (ii) seed cultures attained between 1 and 10 days; (iii) induction of hyphal microfragmentation by incubating in Erlenmeyer flasks with baffles or use of borosilicate spheres (Sigma-Aldrich); and (iv) inoculation of the microhypha fragments in an array of solid media. Isolation of morphotypes unable to produce conidia was achieved only in a culture of liquid oatmeal medium grown in an Erlenmeyer flask with baffles for 4 days (early culture exponential phase); cultures were then used to inoculate oatmeal agar plates (Fig. S8). Nonsporulating mycelium morphotypes (i.e., white mycelium colonies) were then recultured on fresh solid media, namely, oatmeal and peptone-dextrose agar (PDA) (highly sporulating media for S. bicolor), to confirm the observed morphotype. Organic extracts from 20-day-old cultures were analyzed by UVLC-MS to confirm absence of tetrapeptide production.
Genomic/metagenomic DNA extraction.Axenic bacterial genomic DNA or fungal metagenomic DNA (from mycelium or conidia) was extracted using the cetyltrimethylammonium bromide detergent-polyvinylpyrrolidone (CTAB/PVP) extraction method (53). For fungal biomass, an early step of shear stress was introduced: vortexing (Fisher Scientific) with 1-mm-diameter borosilicate beads (ca. 1 volume biomass/1 volume beads/600 μl CTAB/PVP buffer [Sigma-Aldrich]; 2,500 rpm, 15 min).
Gene amplification by PCR.16S rRNA amplicon sequencing of DNA was accomplished using OneTaq polymerase (NEB) and the primers FD1 and RP2 (Sigma-Aldrich) as previously described (54). The PCR program consisted in 30 cycles with 1 min of denaturing, 1 min 30 s of annealing at 52°C, and 2 min of extension at 68°C, with a final extension of 10 min. Alternatively, we used the 515F-806R primers (55), using a PCR program consisting of 30 cycles with 30 s min of denaturing, 30 s of annealing at 52°C, and 30 s of extension at 68°C, with a final extension of 5 min.
For PCR amplification of the ITS1 to 5.8S-ITS2 sequences and of the 28S rRNA gene (initial 600 nucleotides) sequence, the 18S3 (56) and NL4 primers (57) were used as previously described (56).
Phylogenetic analysis.Species and genus affinities of fungi were inferred in a Bayesian analysis as previously described (56).
Bacterial phylogenetic analysis was based on the available 16S rRNA amplicon sequences of Burkholderia type (58) relying on multiple sequence alignments using the SINA alignment tool from the Arb-Silva website (59), correcting and omitting positions of uncertain alignments after visual inspection (60). Phylogenetic analysis based on maximum likelihood (ML) was performed using the program PAUP* version 4.0b10 (61). Since the lengths of the 16S rRNA gene sequences used here were uneven, analyses were performed by means of a pairwise deletion method for gaps and missing sites, using all available comparative data from each sequence pair (60). The optimal evolutionary model of nucleotide substitution was estimated through the program jmodeltest2 (62), observing the Akaike information criterion. Bootstrap analyses were performed using 1,000 replications.
The multilocus sequence analyses (MLSA) were based on the concatenated sequences (2,773 bp) of seven housekeeping gene fragments, atpD (443 bp), gltB (400 bp), gyrB (454 bp), recA (393 bp), lepA (397 bp), phaC (385 bp), and trpB (301 bp), retrieved from the sequenced genome of Burkholderia sp. 293K04B. Sequences for type and reference strains were retrieved from the Burkholderia cepacia complex MLST (multilocus sequence typing) databases (https://pubmlst.org/bcc/). Multiple sequence alignments were obtained using Muscle (63), considering the corresponding amino acid alignment for protein-coding genes and visually corrected as previously described (60). Phylogenetic analysis based on ML was done as described above.
In silico DNA-DNA hybridization was carried out with the online genome-to-genome calculator (GGDC 2.0) provided by the DSMZ (64). Reference genomes were retrieved from Integrated Microbial Genomes & Microbiomes (IMG/M) (65). The draft genome was annotated with Prokka (66) and RAST v.2.0 (67) using default parameters.
Microscopy-based studies of the metafungus.For light microscopy, stereomicroscopy, and confocal microscopy imaging, a Zeiss Axioskop coupled with an Olympus DP20 camera, a Leica MDG 28, and a Leica TCS ST5, respectively, were used. For in situ hybridization studies, different sources of fungal biomass grown in solid medium (BMS or oatmeal) were explored, e.g., colonies kept onto the cellophane ground; colonies grown from conidia in a thin layer of BMS-agar deposited onto a sterile glass slide; or mature mycelium carefully scraped with a scalpel from the center of a colony grown in solid oatmeal for 15 days. The fungal biomass was dehydrated and permeabilized with increasing concentrations of acetone (50%, 75%, and 100%; room temperature; 1 to 10 min for each cycle). The hybridization protocol was adapted from reference 68. Targeting the hybridization of bacterial 16S rRNA, oligonucleotide probes were selected from the Probebase database (code pB-xxx): the general eubacterial probe EUB338 (pB-159), the negative-control probe Non-Eub (pB-243) corresponding to the reverse complement of EUB338 (69), and the genus-specific probes targeting Burkholderia (Burkho, pB-347) and Sphingomonas (SPH120, pB-617). To allow multiple-probe hybridizations in the same biologic sample, we designed the 5′-end probes with fluorochromes displaying distinct excitation wavelengths maxima, similar to the four laser channels of the confocal microscope. The fluorochrome-probe sets (Biomers GmbH, Ulm, Germany) were SPH120-Pacific blue (excitation [exc.] with the 405-nm channel/emission [em.] range, 430- to 470-nm detection of Sphingomonas genus), BUR-Atto488 (488-nm exc. channel/500- to 540-nm em. detection of Burkholderia genus), EUB338-Atto565 (561-nm exc. channel/em. range, 575 to 620 nm), NonEUB-CY5 (635-nm exc. channel/em. range, 650 to 690 nm). Hybridization of the probes (20 ng/μl) took place with 30% formamide (Sigma-Aldrich) in the hybridization buffer (900 mM NaCl, 20 mM Tris-HCl, 1 mM EDTA, and 0.01% SDS) in the dark at 46°C for 2 h, followed by washing twice in wash buffer (20 mM Tris-HCl, 0.1 M NaCl), prewarmed to 48°C. The confocal microscope was equipped with a 10× dry objective and a 64× water immersion objective. All images were captured with 512 × 512 pixels.
ACKNOWLEDGMENTS
C. Almeida was the recipient of a fellowship from Fundação para a Ciência e Tecnologia, FCT, Portugal (fellowship SFRH/BPD/77720/2011). This research was financially supported by Fundación MEDINA, Spain, and by Project LISBOA-01-0145-FEDER-007660 (Microbiologia Molecular, Estrutural e Celular) funded by FEDER funds through COMPETE 2020—Programa Operacional Competitividade e Internacionalização (POCI) and by FCT grant UID/Multi/04551/2013 (Research Unit GREEN-it “Bioresources for Sustainability”).
We thank Gabriele M. Koenig (Institute for Pharmaceutical Biology, University of Bonn, Germany) for kindly providing S. bicolor strain 293K04 and acknowledge the institutional support of BIOISI (Faculty of Sciences, University of Lisbon, Portugal), as well as that of the IGC gene expression facility (Instituto Gulbenkian de Ciência, Portugal). We are thankful to James Yates (ITQB NOVA) for proofreading the manuscript.
We declare that we have no conflict of interests.
FOOTNOTES
- Received 23 March 2018.
- Accepted 23 May 2018.
- Accepted manuscript posted online 1 June 2018.
Supplemental material for this article may be found at https://doi.org/10.1128/AEM.00660-18.
- Copyright © 2018 American Society for Microbiology.
