Streptomyces species produce SLPs
In previous studies, proteomic analyses have shown that phage-related proteins are expressed during the growth of S. coelicolor A3(2) [20, 21], although there is no information regarding its biological role and macromolecular structure. The gene cluster encoding these proteins is almost completely conserved in S. lividans, a close relative of S. coelicolor A3(2) (Fig. 1A). These gene clusters include the homologs of known eCIS genes encoding essential structural proteins (Fig. 1A), suggesting that phage-related genes in S. coelicolor and S. lividans encode eCIS-like nanostructures. Given that these eCIS-related genes are widely found in other Streptomyces species (Fig. 1A), we analyzed the phylogenetic relationships of the eCIS-related genes among actinomycetes and other bacterial classes. Based on a comprehensive annotation of the eCIS genes in the database of extracellular contractile injection systems [9], we constructed a phylogenetic tree of eCIS-related genes encoding tube proteins [15] conserved among proteobacteria, firmicutes, and actinobacteria (Fig. 1B). The eCIS-related genes in actinomycetes form a distinct clade that is phylogenetically distant from the functionally characterized eCISs (anti-feeding prophage, Photorhabdus virulence cassettes, and metamorphosis-associated contractile structures), which is consistent with that reported in previous studies [8, 9]. Notably, the eCIS-related gene of S. griseus falls into a subclade within the actinomycetes-specific clade, whereas those of S. lividans, S. coelicolor, and S. albus are a member of another subclade (Fig. 1B). In addition, an eCIS-related gene of S. avermitilis belongs to a clade that is more closely related to the eCIS genes of proteobacteria and firmicutes, implying that a horizontal gene transfer event might have occurred [8, 9].
Next, we investigated whether Streptomyces species produce eCIS-like nanostructures. Considering known Gram-negative bacterial eCISs are released into the extracellular milieu via cell lysis [15], we initially cultivated S. lividans and S. coelicolor in liquid culture and then ultracentrifuged the culture supernatant. However, we failed to find phage tail-like structures in the resuspended pellets by transmission electron microscopy (TEM), suggesting that these Streptomyces species do not release the putative nanostructures into the extracellular milieu under the tested conditions. Given this, we cultivated S. lividans and S. coelicolor on solid media and performed mild extraction from their mycelia using lysozyme and a detergent. As a result, we found a number of bullet-like nanostructures, which are very similar to known Gram-negative bacterial eCISs [11, 15] in the extracts (Fig. 2A and 2B). These structures were not observed in the deletion mutant of S. lividans for a gene encoding a protein (SLIV_17120, SlpS) that is homologous to Gram-negative bacterial eCIS sheath proteins (Fig. 1A and Supplementary Fig. S1). Because deletion of the sheath proteins was previously shown to abort the assembly processes of eCIS particles [11], our finding demonstrates that the phage tail-like structures of the tested Streptomyces species were synthesized from the conserved eCIS-related genes. Based on these observations, we named the nanostructures “Streptomyces phage tail-like particles (SLPs).”
Previous studies have reported that the expression of SLP genes in S. coelicolor is regulated by SCO4263 encoding an LuxR-type transcriptional regulator [18, 19]. SCO4263 contains the UUA codon, which is very rare in the Streptomyces genome that has a high G + C content, in its 5′ region, and its expression is completely dependent on bldA [17], a gene encoding Leu-tRNAUUA, which is capable of translating the UUA codon; therefore, SLP genes are regulated by bldA in S. coelicolor. The SCO4263 homolog in S. lividans, namely slpR (Fig. 1A), also contains a UUA codon in its 5′ region, strongly suggesting that bldA is a key regulator of SLP expression in S. lividans as well as S. coelicolor. In fact, SLP structures were not found in deletion mutants for slpR and bldA of S. lividans (Supplementary Fig. S1). This confirmed that bldA-dependent regulation of SLP genes is a common mechanism in S. lividans and S. coelicolor.
To examine whether other Streptomyces species produce SLPs, we cultivated three model Streptomyces species (S. albus, S. griseus, and S. avermitilis), all of which have eCIS-related genes in their genome (Fig. 1A). Their mycelia grown on solid media were subjected to mild extraction, ultracentrifugation, and TEM, as described above. As a result, we could observe a number of SLP-like nanostructures in the extract of S. albus, whereas such structures were not found in the other species (Fig. 2A). These SLP-like nanostructures were not found in the deletion mutant for XNR0535, a slpS homolog in the eCIS-related gene cluster of S. albus (Fig. 1A), suggesting that these nanostructures are the products of the SLP-like genes in S. albus. In addition, bioinformatic analysis using PHASTER [22] suggested the absence of intact or defective phage genes in the S. albus genome, further confirming the origin of the SLP-like nanostructures in S. albus. Importantly, a slpR homolog is present in eCIS-related gene clusters of S. albus but absent in those of S. griseus and S. avermitilis (Fig. 1A), indicating that regulation systems for the expression of latter gene clusters might be different from those of SLPs. This may, in part, explain why SLPs were not found in S. griseus and S. avermitilis under the tested conditions.
Loss of SLP genes affects the microbial competition between Streptomyces lividans and fungi
Next, we investigated the biological functions of SLPs using S. lividans as a model. It has long been known that bldA, a key regulator of SLP expression (Supplementary Fig. S1) [17], is an essential factor in the life cycle of Streptomyces species, which encompasses the following developmental stages: spore, vegetative mycelia, and aerial mycelia formation. Many genes essential for these morphological differentiations are regulated by bldA and, therefore, the deletion of bldA leads to a “naked” phenotype, wherein aerial mycelia and spores are lacking [17]. In addition, bldA also acts as a trigger for secondary metabolite production; many gene clusters responsible for secondary metabolite production contain UUA codons, suggesting a pleiotropic role of bldA in the Streptomyces life cycle [17, 23, 24]. Given these observations, we first examined the effects of the deletion of the SLP gene on morphological differentiation and secondary metabolite production in S. lividans. However, in our experiments, the spore formation rates and the liquid chromatography/mass spectrometry profiles of the extractable metabolites of ΔslpS mutant were comparable with those of the TK23 strain. These results support those of the previous study, wherein the absence of any detectable difference in morphological differentiation and secondary metabolite production between the TK23 strain and SLP-deficient mutant of S. coelicolor was reported [18, 19]. To further assess the phenotypic consequences of the deletion of SLP genes, we measured the growth of the TK23 S. lividans strain and two SLP-deficient strains (ΔslpS and ΔslpR). This revealed that slpR deletion leads to the fast-growing phenotype, whereas slpS deletion causes a slight increase in cell weight (Supplementary Fig. 2). In addition, to detect SLP expression and observe its distribution, we constructed an S. lividans strain expressing msfGFP-fused SlpS (SlpS-msfGFP). Microscopic analysis revealed that SlpS-msfGFP fluorescence is uniformly distributed in S. lividans colonies grown under standard conditions (Supplementary Fig. 3). These results are consistent with those of the previous study showing that SCO4253, a SlpS homolog in S. coelicolor, is expressed constantly during growth [18, 19]. The abundant and constant expression of structural proteins during the growth phase has also been reported in the T6SSs of Gram-negative bacteria [25], which may relate to their ecological importance. In addition, it has recently been proposed that some genes, including bldA-dependent ones, that are associated with the Streptomyces life cycle could mediate interactions between Streptomyces species and other organisms [26, 27]. These observations led us to hypothesize that producing SLPs may have a significance in the biological interactions that Streptomyces species would constantly face in their natural settings. Given that Streptomyces species are often found in complex microbial communities in natural environments [28], we speculated that the deletion of SLP genes lead to phenotypes that are disadvantageous to Streptomyces species in microbial competitions that could occur in such environments. For this reason, we co-cultured S. lividans and SLP-deficient mutants with various microorganisms and observed the colony morphologies. Among the tested microorganisms, S. cerevisiae and S. pombe were found to more severely invade the ΔslpS mutant colony than the TK23 strain of S. lividans (Fig. 3A and 3B). The slpS-complemented strain and ΔslpR mutant showed similar phenotypes to the TK23 strain and ΔslpS mutant, respectively, in the above competition assay, confirming the involvement of SLPs in the competitive interactions (Fig. 3A and 3B).
To gain more insights into the competition between S. lividans and fungi, we performed confocal laser scanning microscopy (CLSM) to visualize the colony boundaries. S. lividans and the fungi were visualized using Syto59 dye and GFP, respectively. The CLSM revealed that the colony boundary between the TK23 strain of S. lividans and S. pombe tends to be distorted, whereas distinct borders are formed between the ΔslpS mutant and the fungi (Fig. 4). In addition, the slpS-complemented strain showed similar phenotypes to the parental strain (Fig. 4). These results show that the outcomes of the microbial competitions between S. lividans and the fungi can be altered by the presence or absence of SLP genes.
Differential expression patterns of SLP genes and antibiotic biosynthesis genes under co-culture conditions
To further confirm the involvement of SLPs in the interkingdom competition, we analyzed SLP expression in S. lividans under co-culture conditions with fungal competitors. In the co-culture condition with S. pombe, SlpS-msfGFP fluorescence was clearly observed at the colony boundary between S. lividans and the fungi (Fig. 5A).
Furthermore, for simultaneous observation of the CLSM findings of the colony boundary and SlpS expression, we constructed an S. lividans strain expressing SlpS-mScarletI, the fluorescence of which can be detected separately from that of the Syto59 dye and GFP. When this S. lividans strain was cultured with fungal competitors, the fluorescence of SlpS-mScarletI was detected at the colony boundary (Supplementary Fig. 4), clearly indicating that S. lividans expresses SLPs in the contact region between these microorganisms.
During the microscopic observation of the expression of the fluorescent protein-labelled SlpS under the co-culture conditions, we noticed that SlpS-msfGFP fluorescence was absent at the colony boundary between S. lividans and B. subtilis 168, whereas S. lividans accumulated red pigments at the colony boundary, which strongly indicates the co-culture-dependent induction of secondary metabolite production, including that of the red antibiotic prodiginines (RED) [29–31] (Fig. 5B). In addition, in the competition assay with B. subtilis, the ΔslpS mutant of S. lividans showed a phenotype similar to that of the TK23 strain, suggesting that the absence of SLPs at the colony boundary does not affect this interbacterial competition (Fig. 5C). Given these results, we speculated that S. lividans increases antibiotic production in response to the interbacterial competition while decreasing SLP expression. We thus compared the transcription levels of the structural gene of SLPs and antibiotic biosynthetic genes under co-culture conditions with bacterial or fungal competitors. We co-cultured S. lividans with either S. pombe or B. subtilis and extracted total RNA to quantify the amount of the transcripts of slpS and redD, a gene encoding a pathway-specific activator protein for RED production in S. lividans [32]. Subsequent quantitative reverse transcription polymerase reaction (RT-qPCR) analysis revealed that the transcription level of slpS was approximately two times higher in the co-culture with S. pombe than in the single culture (Fig. 5D), indicating that SLP expression was induced in response to contact with the fungus, whereas the redD transcription levels were comparable between these culture conditions (Fig. 5E). In contrast to the co-culture conditions with the fungal competitor, the transcription level of slpS was rapidly decreased during co-culture with B. subtilis, which is consistent with the absence of SlpS-msfGFP fluorescence at the colony boundary between S. lividans and B. subtilis (Fig. 5B). In addition, significantly higher levels of redD transcription in the co-culture with B. subtilis (Fig. 5E) support the previous observation that the induction of RED production in S. lividans can be triggered by interbacterial competition (Fig. 5B) [29–31].
Besides RED, S. lividans produces several antibiotics such as calcium-dependent antibiotic (CDA), a channel-forming antibiotic that is effective against various Gram-positive bacteria, including B. subtilis [33, 34]. CDA production is regulated by cdaR encoding a pathway-specific activator protein CdaR [35]. The transcription levels of cdaR under the co-culture condition with B. subtilis showed similar tendency to that of redD (Fig. 5F), indicating that the co-culture with B. subtilis also triggered the upregulation of the biosynthetic genes responsible for CDA production. Moreover, although co-culture with B. subtilis showed these positive effects on the antibiotic production in S. lividans, co-culture with S. pombe did not alter the transcription level of cdaR (Fig. 5F). Together with the results of the microscopic analysis (Fig. 5A and 5B), these transcriptional responses demonstrate that the patterns of SLP expression and antibiotic production in S. lividans can differ depending on its competitor species.