2019 Volume 44 Issue 3 Pages 200-207
Microorganisms produce and secrete a variety of secondary metabolites including fatty acids, polyketides, terpenoids, alkaloids, and peptides. Among them, many molecules act as chemical signals that play important roles in inter-/intra-species microbial communication or the interaction with host organisms. In this review, I focus on our recent reports of the microbial signaling molecules involved in bacterium–fungus, bacterium–plant, and fungus–plant interactions. Their potential contribution to pest management is also discussed.
Secondary metabolites are low-molecular-mass organic compounds that, unlike primary metabolites, are not directly involved in the growth, development, or reproduction of the producing organisms. To date, numerous natural products have been identified from both terrestrial and marine organisms.1) In paticular, microorganisms are able to synthesize a large number of secondary metabolites.2) However, little is known about their biological roles for producer microorganisms. Thus, chemists and microbiologists have focused on elucidating the physiological and ecological roles of such metabolites. It is generally thought that these metabolites may contribute to protecting niches of producers in nature. That is, the “present” secondary metabolites seem to be continuously selected throughout the history of producer microorganisms and thus may potentially show strong and/or unique biological activities. We thus believe that the studies on microbial signaling molecules might provide insights for establishing a new methodology of pest management.
Many bacteria have developed close associations with eukaryotes including mammals, insects, and plants. The ability to enter into and inhabit eukaryotic cells has given the bacteria specialized niche environments that are stable, nutrient rich, and enemy free. In the past decade or so, it has been revealed that some bacteria inhabit fungal cells and form mutually beneficial relationships with their host fungi.3,4) These bacteria, called endobacteria, are now receiving much attention in the study of microbial ecology and the evolution of endocellular symbiosis. The associations of Burkholderia bacteria with Rhizopus/Mortierella fungi are well-studied examples of endobacterium–fungus interactions.5) For example, the phytotoxin rhizoxins of the plant pathogen R. microsporus are truly produced by its endobacterium, B. rhizoxinica.6) Recently, Keller et al. found that Ralstonia solanacearum, a plant pathogenic bacterium, enters fungal hyphae and induces chlamydospores in fungi.7) A mutant strain wherein rmyA, a ralstonin biosynthetic gene, is inactive resulted in the complete loss of chlamydospore development in fungi and a significant reduction in bacterial invasion into fungal hyphae. Thus, bacterial secondary metabolites may be key factors that control the endosymbiotic stage of such bacteria.
Because many bacteria and fungi compete with each other in nature, their antagonistic actions are frequently observed during coculturing. Mycophagous behavior, which arrests the growth of fungi, is a of the bacterial antagonism against the surrounding fungi.8) For example, various soil fungi, including the ectomycorrhizal fungus Laccaria bicolor and the arbuscular mycorrhizal fungus Glomus mosseae showed reduced fungal growth when co-inoculated with Collimonas fungivorans Ter331.9) Pseudomonas protegens strains are also known as plant-protective bacteria that reduce fungal and oomycete disease on several plants.10) These antagonistic activities may be due to the production of antibiotic compounds by the bacteria, but we have no idea which compounds truly function in nature. These examples imply that unknown secondary metabolites are involved in interspecies microbial interactions and may be a good source of agrochemical seeds.
In this review, I discuss the recent advances in bacterium–fungus, bacterium–plant, and fungus–plant interactions with a focus on important chemical signals. Specifically, I introduce our recent research that characterized ralfuranones/ralstonins produced by R. solanacearum OE1-1, collimonins produced by C. fungivorans Ter331, and trichorzins produced by the fungus Trichoderma harzianum HK-61.
R. solanacearum is a β-proteobacterium that causes a lethal disease, called “bacterial wilt,” in more than 250 plant species in tropical, subtropical, and warm temperature regions of the world.11) When this pathogen invades host vascular tissues, it grows vigorously and produces extracellular polysaccharide (EPS). The accumulation of EPS prevents water flow in vessels, eventually causing severe wilting symptoms in infected plants. The expression of EPS biosynthetic genes is controlled by PhcA, a LysR-type transcriptional regulator, the activity of which is regulated by quorum sensing (QS), a cell–cell communication mechanism mediated by diffusible signaling molecules (Fig. 1).11) We previously reported that the strains of R. solanacearum employ (R)-methyl 3-hydroxymyristate (3-OH MAME) or (R)-methyl 3-hydroxypalmitate (3-OH PAME) as their QS signals.12,13) In many plant pathogenic bacteria, the production of secondary metabolites is controlled by the QS system, and the metabolites contribute to bacterial virulence on plants. Therefore, the identification of new QS-controlled metabolites will provide insights into the mechanisms responsible for the virulence of R. solanacearum.
To identify phc QS-regulated secondary metabolites, we first compared the HPLC chromatograms of culture extracts of OE1-1 and ∆phcA. As a result, we detected five major peaks 1–5 in the EtOAc extract from a 4-day-old culture of OE1-1 but not in the chromatogram of ∆phcA. This confirmed that these metabolites were under the control of the phc QS. The UV spectra of 1 and 2 were similar to those of ralfuranones A and B (Fig. 2), which are secondary metabolites previously identified from strain GMI1000.14) This finding suggested that the compounds produced by strain OE1-1 were known and new ralfuranone derivatives. The ESI-MS of 3 showed an [M–H]− ion at m/z 281, indicating that its molecular weight is 16 Da greater than that of ralfuranone B. Thus, the compound was suggested to be an oxygenated derivative of ralfuranone B. 1D and 2D NMR data revealed that the compound was the 5-hydroxy derivative of ralfuranone B, and it was given the name ralfuranone J (Fig. 2).15,16) Analogously, compounds 4 and 5 were elucidated as ralfuranones K and L, respectively (Fig. 2).
Although ralfuranones were not toxic to host plants,15) this does not preclude ralfuranone involvement in the virulence of R. solanacearum. In order to establish whether ralfuranones were involved in the virulence of strain OE1-1, we created a ralA-deficient mutant of OE1-1 (ΔralA). The mutant strain did not produce any ralfuranones. Using wounded-petiole inoculation, we assessed the ability of ΔralA to cause wilt disease in tomato plants. The virulence of ΔralA on tomato plants was significantly weaker than that of strain OE1-1.15) The complementation of ralA gene restored the ralfuranone production and the virulence of ΔralA on tomato plants, suggesting the involvement of ralA in OE1-1 virulence. Therefore, the production of ralfuranones appears to be important for the full virulence of strain OE1-1 on host plants.
To investigate the functions of ralfuranones in more detail, we analyzed R. solanacearum transcriptome data generated by RNA-Seq.17,18) phcB, which is associated with 3-OH MAME production, and phcA genes were expressed in ΔralA at levels similar to those in strain OE1-1. In addition, ΔralA exhibited downregulated expression of more than 90% of the positively QS-regulated genes and upregulated expression of more than 75% of the negatively QS-regulated genes. These results suggest that ralfuranones affect the phc QS circuit. Ralfuranone supplementation restored the ability of ΔralA cells to aggregate. In addition, ralfuranones A (1) and B (2) restored the swimming motility of ΔralA to wild-type levels. However, the application of exogenous ralfuranones did not affect the production of the major exopolysaccharide, EPS I, in ΔralA. Quantitative real-time PCR assays revealed that the deletion of ralA results in the downregulated expression of vsrAD and vsrBC, which encode a sensor kinase and a response regulator, respectively, in the two-component regulatory systems that influence EPS I production.17) The application of ralfuranone B (2) restored the expression of these genes. Overall, our findings indicate that ralfuranones influence the phc QS circuit and virulence of R. solanacearum.
In order to discover unidentified secondary metabolites of R. solanacearum, we conducted a transcriptome analysis between strain OE1-1 and its QS mutants.17–19) We found that the expression of the hybrid polyketide synthase-nonribosomal peptide synthetase (PKS-NRPS) genes, rmyA and rmyB, is under the control of QS. These genes were initially detected in the genome of R. solanacearum strain GMI1000 and were expected to produce a derivative of syringomycins, phytotoxic lipopeptides of Pseudomonas syringae.20) Lipopeptides are a class of microbial metabolites composed of a fatty acid and a peptide and exhibit various biological activities.21) However, the details of this QS-controlled metabolite in R. solanacearum remained unclear for more than a decade. Keller et al. recently reported that those genes are responsible for the production of the lipopeptide ralstonins.7) Moreover, they found that R. solanacearum enters fungal hyphae and induces chlamydospores in fungi. A mutant strain wherein the rmyA gene is inactive resulted in the complete loss of chlamydospore development in fungi and a significant reduction in bacterial invasion into fungal hyphae. Thus, ralstonin may be a key factor in controlling the endofungal stage of R. solanacearum.
In order to isolate ralstonins, we prepared 1200 BG agar plates in which strain OE1-1 was grown for 3 days and extracted cells and media with acetone. After evaporation of the acetone, the aqueous concentrate was extracted with EtOAc. Ralstonins were only detected in the EtOAc extract by LC/MS. Thus, the EtOAc extract was separated on a silica gel column by eluting with n-hexane/EtOAc and EtOAc/MeOH. Ralstonins were eluted in 60% and 100% MeOH in EtOAc fractions. These fractions were combined and subjected to preparative reversed-phase HPLC, giving almost pure ralstonins A (6) and B (7) (Fig. 3).
The detail of the ralstonin structure elucidation was published in our previous paper.19) The ralstonin structures were unique in several aspects. They contained unusual amino acids such as β-hydroxytyrosine and dehydroalanine and the novel β-amino fatty acid 3-amino-2-hydroxyoctadecanoic acid (Ahod). Several lipopeptides with a β-amino fatty acid have been reported.22–24) In these cases, the amino groups of N-terminal β-amino fatty acids were connected to the carboxyl groups of the C-terminal residues, forming large cyclic peptides. However, the amino groups of Ahod in ralstonins were of the free form and the carboxyl groups in the C-terminal β-alanine were connected to the hydroxy group of homoserine, forming small cyclic depsipeptide moieties (Fig. 3). Thus, to the best of our knowledge, ralstonins are a previously undescribed type of lipodepsipeptide in nature.
We found that strain OE1-1 induced more chlamydospores of Fusarium oxysporum when they were cocultured. Thus, the chlamydospore-inducing activities of ralstonins on the fungus were evaluated using a disc diffusion assay. The formation of chlamydospores in F. oxysporum was induced by ralstonins A (6) and B (7).19) The phytotoxicity of ralstonins was evaluated. After the inoculation of ralstonins into tobacco leaves, necrosis was observed. A similar response was observed on the leaves of tobacco after the injection of a lipopeptide produced by the plant pathogen Pseudomonas corrugata.63) Thus, we confirmed that ralstonins exhibit strong chlamydospore-inducing activity and moderate phytotoxicity. We are now investigating the mode of action of ralstonins as chlamydospore inducers.
Understanding the molecular events involved in plant–microbe interactions may lead to new ways to save crops from the threat of pathogens. It is generally thought that bacterial pathogens sense certain molecules originating from host plants to detect the close presence of the hosts. When receiving such plant-derived compounds, bacterial pathogens evoke a variety of responses related to pathogenesis. For example, Agrobacterium tumefaciens, the causative agent of crown gall disease, senses a structurally diverse group of phenolic compounds from wounded plant tissues and then activates the transcription of virulence (vir) genes necessary for the transfer of T-DNA to host cells.25) Plant phenolic glucosides and sugars (such as D-fructose) activate the production of syringomycin, a lipopeptide phytotoxin produced by Pseudomonas syringae.26–28) Moreover, in rhizobia, plant flavonoids activate nodulation (nod) genes essential for the establishment of symbiosis with host plants.29) These examples demonstrate that proteobacteria, which have a close relationship with plants, have developed sophisticated mechanisms to detect the presence of their hosts by sensing specific plant-derived compounds.
We found that D-galactose and D-glucose, plant sugars, activate the production of ralfuranones/ralstonins in R. solanacearum. As a result, two new derivatives, ralfuranone M (8) and ralstonin C (9) (Fig. 4) were found in the culture extracts, and their structures were elucidated using spectroscopic and chemical methods.30) Ralstonin C (9) is a cyclic lipopeptide containing a unique fatty acid, (2S,3S,Z)-3-amino-2-hydroxyicos-13-enoic acid, whereas ralfuranone M (8) has an aryl-furanone structure in common with other ralfuranones. D-Galactose and D-glucose activated the expression of the biosynthetic ralfuranone/ralstonin genes and, in part, became the biosynthetic source of ralfuranones/ralstonins. Ralfuranones and ralstonins were detected in the xylem fluid of the infected tomato plants, and their production–deficient mutants exhibited reduced virulence on tomato and tobacco plants. Taken together, the activation of ralfuranone/ralstonin production by host sugars would contribute to R. solanacearum virulence.
Many natural polyynes, unique molecules with alternating triple and single carbon–carbon bonds, from plants, basidiomycetes, and insects have been reported.31,32) However, examples of polyynes of bacterial origin are very limited. The few partially characterized bacterial polyynes, such as cepacins from the human pathogen Burkholderia cepacia, caryoynencin from the carnation pathogen Burkholderia caryophylli, and Sch 31828 from Microbispora sp., possess a conjugated triple bond started with a terminal alkyne.33–35) Bacterial polyynes are very unstable and thus require caution when handled. In addition, these compounds possess one to several chiral centers, which are formed as a result of oxidation, although their absolute configurations have not been completely determined. Recently, the biosynthetic machinery of bacterial polyynes was identified, and the wider distribution of such metabolites in bacterial clades was suggested.36–38) However, although three decades have passed after the identification of Sch 31828, novel bacterial polyynes have not yet been isolated and characterized. Hence, the stable isolation and structure determination of bacterial polyynes are challenging and important topics in chemistry.
The genus Collimonas consists of mostly soil β-proteobacteria that are defined by their ability to grow at the expense of living fungal hyphae under nutrient-limited conditions and thus are known as “fungus-feeding bacteria.”39,40) So far, three species have been described, and all of them have a representative in which antifungal activity was demonstrated. The detailed analysis of the C. fungivorans Ter331 mutants defective in antifungal activity suggested that a gene cluster named cluster K is involved in the production of undescribed bacterial polyynes, named collimonins.41) These compounds may contribute to the unique interactions of C. fungivorans Ter331 with soil fungi.
The genes encoding the possible enzymes for collimonin biosynthesis within cluster K were named colABCDEFG.42) The col gene cluster starts with a long-chain fatty acyl-CoA ligase (colA) and possesses three desaturases (colB, colC, and colE), an acyl carrier protein (colD), and two hydrolases (colF and colG) (Fig. 5A). This gene combination was also observed in the cay gene cluster of B. caryophylli, which is capable of producing caryoynencin.36) Thus, the collimonin structures were also expected to be polyyne fatty acids with a terminal alkyne and hydroxy group(s). However, because of their high instability, collimonins had not been isolated, so their structures and biological activities remained elusive.
First, we determined the HPLC peaks of antifungal polyynes in the culture extract of Ter331. The EtOAc extract of 1200 WYA plates in which Ter331 was grown for 5 days was separated on an ODS column by eluting with H2O/MeOH. The fraction eluted with 60% MeOH in H2O showed antifungal activity against Aspergillus niger and gave four major HPLC peaks showing characteristic UV spectra for polyynes. Collimonins were unstable and easily polymerized to insoluble matter. By avoiding drying and exposure to oxygen, higher than ambient temperature, and light, we could isolate those compounds by two-step HPLC separation.
Collimonin A (10) generated complicated ESI-MS data, so its molecular weight was not determined from this data alone. The 1H NMR data accounted for 12 nonexchangeable protons, composed of one terminal alkyne proton, two (Z)-olefinic protons, five oxymethine protons, and two methylene groups. Analysis of 13C NMR and HMQC spectra showed 16 carbon signals and revealed the presence of a carboxyl group, (Z)-olefinic methines, a triyne, five oxymethines, and two methylenes. The characteristic UV spectrum suggested the presence of an ene-triyne moiety. The COSY and HMBC correlations were used to construct the planar structure of collimonin A (10), which possessed ene-triyne, trans-epoxide, and γ-lactone moieties (Fig. 5B).42) The planar structures of collimonins B (11), C (12), and D (13) were also determined (Fig. 5B).
To elucidate the absolute configurations of collimonins C (12) and D (13), we synthesized the isomers of methyl 6,7-dihydroxyhexadecanoate (14) (Fig. 5C) from the collimonins by catalytic hydrogenation and subsequent methylation and compared their HPLC retention times with those of synthetic 14. Compound 14 prepared from collimonins C (12) and D (13) showed retention times identical to those of synthetic 6S,7R- and 6R,7R-14, respectively.42) Thus, the absolute configurations of collimonins C (12) and D (13) were determined to be 6S,7R and 6R,7R, respectively.
To determine the absolute configuration of collimonin B (11), the crystalline sponge (CS) method was applied to the structure analysis.43) Click derivative of collimonin B was treated with a single crystal of the [(ZnCl2)3(tpt)2] complex [CS crystal; tpt=2,4,6-tris(4-pyridyl)triazine], and the guest-absorbed CS crystal was subjected to a diffraction study. The relative configuration of collimonin B (11) was revealed to be 4S*,5R*,6S*,7S*,8R*, although its absolute configuration could not be established (the Flack Parameter was 0.400(8)). Using J-coupling-based configuration analysis and (R)/(S)-α-methoxyphenylacetic acid derivatization,42,44) we definitely determined the absolute configuration of collimonin B (11) to be 4S,5R,6S,7S,8R.
The biological activities of collimonins against A. niger were examined. In a noncontact confrontation assay, Ter331 caused growth inhibition, branching, and pigmentation of A. niger hyphae. Collimonins A (10), C (12), and D (13) showed antifungal and hyphal branching activities.42) Collimonin B (11) induced the pigmentation of A. niger hyphae but did not show antifungal activity. The saturated analogue of collimonin C, compound 15 (Fig. 5C), did not show antifungal activity, confirming the importance of the ene-triyne moiety. Together, we concluded that collimonins are the causative agents of the antifungal and pigment-inducing activities of Ter331.
Our study is the first to reveal the absolute configurations of polyoxygenated bacterial polyynes. The isolation and analytical techniques used will provide a workflow for the identification of polyoxygenated polyynes from other bacteria. Collimonins A (10), C (12), and D (13) are the major antifungal agents of Ter331, and collimonin B (11) uniquely induced pigment production in fungal hyphae. Therefore, this study uncovered chemical and biological features of these enigmatic metabolites. However, it is still unclear how the bacterium organizes these compounds and other factors such as chitinase to feed/antagonize the surrounding fungi. We are now trying to answer this question by investigating the mode of action of collimonins.
Viruses are among the most detrimental pathogens of crops. Plant viral diseases cause serious economic losses in agriculture by reducing yield and quality. Because of the simple form of viruses, which consist of a segment of DNA or RNA encoding the genes required for their own multiplication in hosts and coat proteins, the chemical control of their diseases remains difficult or impossible.45,46) Therefore, the discovery of inhibitors of plant viral infections continues to be a priority.
We yielded several fungal strains from soil samples collected in Japan that produced unique natural products. For example, Penicillium simplicissimum ATC C 90288 produced insecticidal indole alkaloids, okaramines,47–52) and the unidentified ascomycete OK-128 produced paralytic cyclic peptides, PF1171s.53–55) We screened the fermentation extracts of fungal strains to find anti-plant viral compounds.
Anti-plant viral activity was examined by using a bioassay of Cucumber mosaic virus (CMV) and its local lesion host, cowpea (Vigna sesquipedalis cv. Kurodane-sanjaku), where its activity was evaluated by the inhibition of local lesions resulting from CMV infection.56) MeOH extracts of okara (the insoluble residue of the whole soybean) fermented with fungal strains were used for the bioassay. As a result of screening 200 strains, we found that T. harzianum HK-61 produces antiviral compounds against CMV. The MeOH extract of okara fermented with HK-61 was concentrated in vacuo, and the resulting aqueous concentrate was extracted with EtOAc. The EtOAc extract demonstrated inhibition of the CMV infection; however, the aqueous layer did not. Bioassay-guided purification of the extract by repeated column chromatography over silica gel and ODS resulted in the isolation of three peptaibols—trichorzins HA II (16), HA V (17), and HA VI (18) (Fig. 6). Their structures were confirmed by FAB-MS, NMR, and Marfey’s analysis.57)
Next, we examined the antiviral activity of the purified trichorzins. The compounds were added to hydroponic cultures of cowpea plants, and CMV was inoculated into the leaves. Trichorzin HA V (17) showed the strongest effects, 80.5% (5 µM) and 90.6% (10 µM) inhibition, against CMV.57) Trichorzins HA II (16) and HA VI (18) exhibited 42.6% and 68.5% inhibition, respectively, at a concentration of 10 µM. This is the first report of the anti-plant viral activity of trichorzins.
Peptaibols are characterized by an N-terminal acylated amino acid residue and a C-terminal amino alcohol on a lipophilic amino acid chain that includes many α,α-dialkylated amino acids, such as aminoisobutyric acid and isovaline.58) Numerous peptaibols have been isolated from Trichoderma spp. and several other fungi as antimicrobial substances. Previously, Yeo and co-workers isolated two peptaibols, peptaivirins A and B, from the unidentified fungus KGT142 as antiviral agents against infection by Tobacco mosaic virus (TMV) in the tobacco plant Nicotiana tabacum cv. Xanthi-nc.59) The 18mer peptaibols TvBI and TvBII from T. virens Gv29-8 elicited defense responses in the cucumber plant Cucumis sativus that resulted in resistance against several bacteria.60) Furthermore, trichokonins isolated from T. pseudokoningii SMF2 were revealed to induce defense responses and systemic resistance in N. tabacum var. Samsun NN against TMV infections.61) These results suggest that trichorzins may also induce defense responses in cowpea plants and cause resistance to CMV. Trichoderma is a ubiquitous fungal genus composed of some of the most versatile biocontrol agents against a wide array of plant diseases.62) Therefore, peptaibols may be involved in chemical communication between fungi and plants in nature. Trichorzins’ mechanism of action against CMV in cowpea plants would be an interesting future topic of study.
We identified and characterized the unique microbial communication signals involved in inter-/intra-species microbial communication or the interaction with host organisms. These results clearly indicate that microorganisms employ specific metabolites as signaling molecules. Of course, this knowledge will contribute to our understanding of the physiological and ecological roles of microbial metabolites. It is becoming difficult to develop new microbicides due to the limited base structures and food safety concerns. However, microbial secondary metabolites have been function in nature before we develop and use synthetic pesticides. I am not willing to deny the importance of synthetic pesticides, but we can still learn how to manage pests from natural ecosystems. I believe that the elucidation and chemical control of microbial communication would lead to new ways to combat important crop diseases.
I would like to thank the former and present students of my research group at Osaka Prefecture University and my collaborators for their considerable contributions to our work and stimulating discussions. I am grateful to Emeritus Professor Hideo Hayashi (Osaka Prefecture University), Professor Hisashi Miyagawa (Kyoto University), Professor Yasufumi Hikichi (Kochi University), and Professor Kohki Akiyama (Osaka Prefecture University) for their continuous support and motivation. Finally, I wish to acknowledge my family for their support and encouragement.