Scott A.
Jarmusch
*ab,
Diego
Lagos-Susaeta‡
c,
Emtinan
Diab‡
a,
Oriana
Salazar
c,
Juan A.
Asenjo
c,
Rainer
Ebel
a and
Marcel
Jaspars
*a
aMarine Biodiscovery Centre, Department of Chemistry, University of Aberdeen, Old Aberdeen AB24 3UE, Scotland, UK. E-mail: m.jaspars@abdn.ac.uk
bDepartment of Pharmaceutical Biosciences – Pharmacognosy, Uppsala University, Biomedicinskt Centrum, Uppsala, Sweden 751 23. E-mail: scott.jarmusch@farmbio.uu.se
cCentre for Biotechnology and Bioengineering, University of Chile, Beauchef 850, Santiago, Chile
First published on 9th November 2020
Siderophores are iron-chelating compounds that aid iron uptake, one of the key strategies for microorganisms to carve out ecological niches in microbially diverse environments. Desferrioxamines are the principal siderophores produced by Streptomyces spp. Their biosynthesis has been well studied and as a consequence, the chemical potential of the pathway continues to expand. With all of this in mind, our study aimed to explore extremotolerant and lupine rhizosphere-derived Streptomyces sp. S29 for its potential antifungal capabilities. Cocultivation of isolate S29 was carried out with Aspergillus niger and Botrytis cinerea, both costly fungal phytopathogens in the wine industry, to simulate their interaction within the rhizosphere. The results indicate that not only is Streptomyces sp. S29 extraordinary at producing hydroxamate siderophores but uses siderophore production as a means to ‘starve’ the fungi of iron. High resolution LC-MS/MS followed by GNPS molecular networking was used to observe the datasets for desferrioxamines and guided structure elucidation of new desferrioxamine analogues. Comparing the new chemistry, using tools like molecular networking and MS2LDA, with the known biosynthesis, we show that the chemical potential of the desferrioxamine pathway has further room for exploration.
Rhizosphere-associated microbes are one of the many strategies currently being explored as biocontrol agents. Bacterial communities in the rhizosphere contribute to the extremely complex flow of nutrients as well as the export of antibiotics and other secondary metabolites into the soil.6,7 Model organisms from genera like Bacillus and Pseudomonas have traditionally been promising sources due to their association to the soil and their production of secondary metabolites like lipopeptides but also plant growth promotors.8,9 Actinobactera, which are also predominantly soil-associated microbes, have also been explored as biocontrol agents due to their prolific antimicrobial potential.10 This combined with the fact that extremotolerant microbes have adapted to the harsh conditions and the potential of these adaptations to be translatable to biotechnologies, extremotolerant microbes look to be promising sources as biocontrol agents.11 In Chile, the wine industry is the main agricultural export, estimated at ∼$2 billion USD in 2017,12 thus driving a large research interest around protection of wine-producing grapes against various phytopathogens.
Siderophores are iron-chelating compounds that aid metal uptake in iron deficient environments.13 These secondary metabolites are produced by all microbes and four types of siderophore-mediated social interactions have been observed: uptake among clonal cells (sharing), cheating or piracy (which can only happen amongst microbes with the same uptake receptor), competition via locking away and metabolite competition (whichever microbe makes the most cost efficient and effective chelator).14 Desferrioxamines (DFO) are the most widely observed group of siderophores in Streptomyces spp., used principally for Fe(III) scavenging,13,15 as well as some other heavy metals.16–18 Since the vast majority of organisms require Fe(III) in order to maintain proper cell wall function,13 as one scavenges for itself, it simultaneously starves other competing organisms, rendering them potentially unviable.
The biosynthetic machinery of desferrioxamines has been well studied19–21 and as a consequence, the chemical potential continues to expand. The substrate specificity of DesC-like proteins, similar to acyl CoA-dependent transferases, seems to be the key factor in producing diverse desferrioxamine scaffolds.19,21 It has been shown that DesC can accommodate for acetyl, succinyl, and myristol-CoA, with the latter providing insight into the production of acyl desferrioxamines.19 Additionally, aryl desferrioxamines, first isolated from Micrococcus luteus KLE101122 and later from Streptomyces sp. MA37,23 showed further substrate flexibility, where the latter communication described a new biosynthetic gene cluster (BGC) which contained a DesC homolog (LgoC). This enzyme promiscuity fits ideally with the Screening Hypothesis first proposed by Firn and Jones; secondary metabolism stems from metabolic pathways that ‘maximize diversity and minimize cost’.24
Due to the modular nature of desferrioxamines and the presence of peptide bonds throughout their structure, mass spectrometry-based studies have been utilized extensively to not only study their biological impact but also for elucidation of new structures. Previous studies have established standard fragmentation patterns of these siderophores, even as far as elucidating new structures solely based on mass spectrometry.25–31 This standardized fragmentation makes the discovery of siderophores much easier thanks to metabolomics-based tools like Global Natural Products Social Molecular Networking (GNPS). It allows for correlations to be drawn within your dataset and against the GNPS library that relates similar MS/MS spectra to one another, forming clusters or families of similar features.32
With all of this in mind, our study originally aimed to explore extremotolerant, rhizosphere-derived Streptomyces sp. S29, a novel strain, for antifungal secondary metabolites. Cocultivation of S29 with Aspergillus niger and Botrytis cinerea, both costly fungal phytopathogens in the wine industry, were carried out in a fashion to allow S29 to dominate the culture, thus eliciting a chemical response from the addition of each fungus. Bioassay-guided fractionation lead to no lead compounds as potential antifungal agents yet post-fermentation streaks of inoculum showed no fungal growth. High resolution LC-MS/MS followed by GNPS molecular networking was used to analyse the data sets. The results indicated that not only is Streptomyces sp. S29 extraordinary at producing hydroxamate siderophores, totalling 24 new analogues and 8 previously identified ones confirmed through MS/MS structure elucidation, but uses siderophore production as a means to ‘starve’ the fungi of iron. When grown in coculture with the fungi, 22 additional new analogues and 17 additional previously described were identified, clearly showing a chemical response to the fungi in culture environment. Using MS2LDA to further explore desferrioxamine chemical space, we observed that it is potentially 2.5 times larger than previously anticipated. These results point to the potential of Streptomyces sp. S29 as a biocontrol agent to prevent fungal infection and the potential for new desferrioxamine chemistry.
![]() | ||
Fig. 1 Preliminary antifungal screening on Streptomyces isolate S29 (right) grown with Botrytis cinerea (left). Image courtesy of Diego Lagos-Susaeta. |
Extracts and fractions were evaluated for the presence of two DFOs (B and D1) during the fractionation steps, in order to trace the fractions that contained the highest abundance of siderophores. Presence of DFOs was further evaluated at the final stages of separation based on the molecular families generated from molecular networking, utilizing the known DFOs in the GNPS library as anchor points, which further aided structure elucidation. Based on the LC-MS results, the butanol partitions contained the largest abundance of DFOs, and therefore this data was used to generate a molecular network using GNPS.
Aryl desferrioxamines related to the previously described acyl ferrioxamine 1 (also known as legonoxamine A) and 2 were also found distributed amongst varying molecular families across the molecular network and do not seem to link into their own cluster such as that seen for the DFO-Ds and DFO-Bs. For the remaining small molecular families that contain DFOs, we observed a group of sodiated adducts, large acyl DFO-Bs, as well as the macrocyclic desferrioxamine E (Fig. S1, ESI†). The remaining nodes that link within the desferrioxamine molecular families were unable to be identified using the annotation workflow described in Fig. 4.
MS/MS based structure elucidation was carried out in a similar fashion to studies conducted previously (Fig. 4).25–28 Early observations of desferrioxamine B and D1 fragmentation showed an initial loss of the N-terminus, corresponding to the loss of either 118.1 or 160.1. Using this as a foundation, structures were either started with a N-hydroxycadaverine or a N-hydroxy-N-acetylcadaverine, respectively. The remainder of structure elucidation was completed stepwise: (1) predicting the molecular formula from the accurate mass and collection of MS/MS data, (2) filling out the structures with appropriate desferrioxamine precursors (i.e. N-hydroxycadaverine, putrescine, succinate, acetate, etc.), and (3) taking the remainder of the molecular formula and placing an appropriate acid at the C-terminus that either has precedent from previous studies or could be a potential binding partner with an acyltransferase like coenzyme A.
Overall, 41 new DFO analogues were identified in the molecular network and through manual data processing (Table 1). These range from short acyl chain DFO-Bs, acyl desferrioxamine D-like (DFO-D) analogues which have not previously been described, tenacibactins, aryl DFOs, and long chain dicarboxylate DFOs (Table 1). Additionally, 5 macrocyclic (with uncertain structures) and 25 known hydroxamate siderophores were detected, including known acyl, aryl DFOs and bisucaberins, giving further credence to the new structure assignments. Their identification, observed presence in monoculture and/or coculture, and identification confidence based on Schymanski's rules39 are present in Table 1. Schymanski's rule distinguish various levels of confidence for mass spectral-based assignments, where the highest score is 1 (verifiable by MS, MS2, RT, and an internal standard) and the lowest score is 5 (exact mass only). The majority of the identified compounds rank at level 2 or level 3 on the Schymanski scale.
Name | Measured mass [M + H]+ | Theoretical mass [M + H]+ | Δ ppm | Sum formula [ + H]+ | ID confidence | Monoculture | Ref. | |
---|---|---|---|---|---|---|---|---|
Nomenclature that are italicized were identified in this study. | ||||||||
Cyclic desferrioxamines | Bisu-03 | 429.2344 | 429.2342 | −0.46 | C19H33N4O7 | 2 | Senges et al. | |
Bisu-01 | 443.2495 | 443.2500 | −1.13 | C20H35N4O7 | 2 | Senges et al. | ||
Uncertain structure | 457.2660 | 457.2657 | 0.65 | C21H37N4O7 | 3 | * | This study | |
Uncertain structure | 499.3110 | 499.3126 | −3.20 | C24H43N4O7 | 3 | * | This study | |
Uncertain structure | 513.3310 | 513.3283 | 5.25 | C25H45N4O7 | 3 | * | This study | |
Uncertain structure | 527.3450 | 527.3439 | 2.08 | C26H47N4O7 | 3 | * | This study | |
Uncertain structure | 583.4050 | 583.4065 | −2.57 | C30H55N4O7 | 3 | This study | ||
Desferrioxamine E | 601.3557 | 601.3556 | 0.17 | C27H49N6O9 | 2 | |||
Aryl desferrioxamines | Legonoxamine B | 337.1754 | 337.1758 | −1.19 | C17H25N2O5 | 2 | Maglangit et al. | |
Legonoxamine C | 351.1912 | 351.1914 | −0.57 | C18H37N4O5 | 3 | This study | ||
Legonoxamine D | 437.2755 | 437.2758 | −0.69 | C22H37N4O5 | 3 | This study | ||
Legonoxamine E | 537.2924 | 537.2919 | 0.93 | C26H41N4O8 | 3 | * | This study | |
Legonoxamine F | 551.3088 | 551.3075 | 2.36 | C27H43N4O8 | 3 | This study | ||
Legonoxamine G | 623.3764 | 623.3763 | 0.16 | C30H51N6O8 | 3 | * | This study | |
Acyl-ferrioxamine 1/legonoxamine A | 637.3917 | 637.3919 | −0.31 | C31H53N6O8 | 2 | * | D’Onofrio et al.; Maglangit et al. | |
Legonoxamine H | 665.3893 | 665.3869 | 3.61 | C32H53N6O9 | 3 | This study | ||
Acyl-ferrioxamine 2 | 679.4025 | 679.4025 | 0.00 | C33H55N6O9 | 2 | * | D’Onofrio et al. | |
Legonoxamine A glycoside | 799.4456 | 799.4448 | 1.00 | C37H63N6O13 | 3 | * | This study | |
DFO-Bs | Un-pre[5 + 5*] | 431.2500 | 431.2520 | −4.63 | C19H35N4O7 | 3 | This study | |
Pre[5 + 5*] | 433.2662 | 433.2657 | −1.15 | C19H37N4O7 | 2 | Rütschlin et al. | ||
Pre[5 + 5*] aldehyde | 447.2450 | 447.2449 | 0.22 | C19H35N4O8 | 3 | This study | ||
Pre[ha + 5 + 5*] | 517.3221 | 517.3232 | −2.12 | C24H45N4O8 | 2 | Rütschlin et al. | ||
Abloxime/IC202C | 517.3347 | 517.3344 | 0.58 | C23H45N6O7 | 2 | * | Iijima et al. | |
Proferrioxamine G1t | 519.3495 | 519.3501 | −1.16 | C23H47N6O7 | 2 | Feistner et al. | ||
DesuA1 | 545.3658 | 545.3657 | 0.18 | C25H49N6O7 | 2 | * | Senges et al. | |
DesA1 | 547.3449 | 547.3450 | −0.18 | C24H47N6O8 | 2 | * | Senges et al. | |
Des A2/IC202B | 557.2880 | 557.2881 | −0.18 | C23H43N6O8Al | 2 | Iijima et al. | ||
Desferrioxamine B | 561.3602 | 561.3606 | −0.71 | C25H49N6O8 | 2 | * | ||
Desferrioxamine N/desf-05 | 575.3762 | 575.3763 | −0.17 | C26H51N6O8 | 2 | * | Ejje et al.; Senges et al. | |
C3 acyl DFO-B | 589.3915 | 589.3919 | −0.68 | C27H53N6O8 | 2 | * | This study | |
C4 acyl DFO-B | 603.4073 | 603.4076 | −0.49 | C28H55N6O8 | 2 | * | This study | |
C5 acyl DFO-B | 617.4233 | 617.4232 | 0.16 | C29H57N6O8 | 2 | * | This study | |
C4 acyl hydroxylated DFO-B | 619.4024 | 619.4025 | −0.16 | C29H55N6O9 | 3 | * | This study | |
uC6 acyl DFO-B | 629.4238 | 629.4232 | 0.95 | C30H57N6O8 | 3 | This study | ||
C6 acyl DFO-B | 631.4388 | 631.4389 | −0.16 | C30H59N6O8 | 2 | * | This study | |
uC7 acyl DFO-B | 643.4389 | 643.4389 | 0.00 | C31H59N6O8 | 3 | This study | ||
C7 acyl-DFO | 645.4550 | 645.4545 | 0.77 | C31H61N6O8 | 2 | Traxler et al.; Sidebottom et al. | ||
C8 acyl-DFO | 659.4697 | 659.4702 | −0.76 | C32H63N6O8 | 2 | Traxler et al.; Sidebottom et al. | ||
uC10 acyl DFO-B | 685.4856 | 685.4858 | −0.29 | C34H65N6O8 | 3 | This study | ||
uC11 acyl DFO-B | 699.5007 | 699.5025 | −2.57 | C35H69N6O8 | 3 | This study | ||
C11 acyl-DFO | 701.5174 | 701.5171 | 0.43 | C35H69N6O8 | 2 | Traxler et al.; Sidebottom et al. | ||
C12 acyl-DFO | 715.5322 | 715.5328 | −0.84 | C36H71N6O8 | 2 | Traxler et al.; Sidebottom et al. | ||
C13 acyl-DFO | 729.5496 | 729.5484 | 1.64 | C37H73N6O8 | 2 | Traxler et al.; Sidebottom et al. | ||
uC12 acyl FO-B [M + Al] + | 737.4776 | 737.4752 | 3.25 | C36H66N6O8Al | 3 | This study | ||
Dodecanedioic DFO-B | 745.5086 | 745.5070 | 2.15 | C36H69N6O10 | 3 | This study | ||
Amphiphilic desferrioxamine 12 | 745.5438 | 745.5434 | 0.54 | C37H73N6O9 | 2 | Sidebottom et al. | ||
Tridecanedioic DFO-B | 759.5221 | 759.5226 | −0.66 | C37H71N6O10 | 3 | This study | ||
Amphiphilic desferrioxamine 15 | 759.5597 | 759.5590 | 0.92 | C38H75N6O9 | 2 | Sidebottom et al. | ||
DFO-Ds | Desferrioxamine H | 461.2602 | 461.2606 | −0.87 | C20H37N4O8 | 2 | Adapa et al. | |
Glutaric desferrioxamine H | 475.2760 | 475.2762 | −0.42 | C21H39N4O8 | 2 | This study | ||
C2 acyl glutaric desferrioxamine H | 489.2900 | 489.2919 | −3.88 | C22H41N4O8 | 2 | This study | ||
Desferrioxamine D4 | 559.3453 | 559.3450 | 0.50 | C25H47N6O8 | 2 | This study | ||
Deoxydesferrioxamine D3 | 587.3757 | 587.3763 | −1.02 | C27H51N6O8 | 3 | * | This study | |
Desferrioxamine D3 | 589.3915 | 589.3919 | −0.68 | C27H53N6O8 | 2 | * | This study | |
Desferrioxamine D1 | 603.3714 | 603.3712 | 0.33 | C27H51N6O9 | 2 | * | ||
C2 acyl DFO-D | 617.3864 | 617.3869 | −0.81 | C28H53N6O9 | 2 | This study | ||
C4 acyl DFO-D | 645.4185 | 645.4182 | 0.46 | C30H57N6O9 | 2 | * | This study | |
C5 acyl DFO-D | 659.4331 | 659.4338 | −1.06 | C31H59N6O9 | 2 | This study | ||
C6 acyl DFO-D | 673.4498 | 673.4495 | 0.45 | C32H61N6O9 | 2 | * | This study | |
C7 acyl DFO-D | 687.4655 | 687.4651 | 0.58 | C33H63N6O9 | 2 | * | This study | |
C11 acyl DFO-D | 743.5280 | 743.5277 | 0.40 | C37H71N6O9 | 2 | * | This study | |
C12 acyl DFO-D | 757.5436 | 757.5434 | 0.26 | C38H73N6O9 | 2 | * | This study | |
C13 acyl DFO-D | 771.5607 | 771.5590 | 2.20 | C39H75N6O9 | 2 | * | This study | |
C14 acyl DFO-D | 785.5749 | 785.5747 | 0.25 | C40H77N6O9 | 2 | * | This study | |
Tenacibactins | Tenacibactin C | 503.3070 | 503.3075 | 0.99 | C23H43N4O8 | 3 | Jang et al. | |
Tenacibactin E | 531.3380 | 531.3388 | −1.15 | C25H47N4O8 | 3 | * | This study | |
Tenacibactin F | 545.3545 | 545.3545 | 0.00 | C26H49N4O8 | 3 | * | This study | |
Tenacibactin G | 587.3990 | 587.4014 | 4.08 | C29H55N4O8 | 3 | This study | ||
Tenacibactin H | 601.4168 | 601.4171 | −0.49 | C30H57N4O8 | 3 | * | This study | |
Tenacibactin I | 615.4326 | 615.4327 | −0.16 | C31H59N4O8 | 3 | This study | ||
Tenacibactin J | 629.4484 | 629.4484 | 0.00 | C32H61N4O8 | 3 | This study |
We propose henceforth that the previously known ‘acyl DFOs’ should be renamed ‘acyl DFO-B’ due to the observation in this study of analogues that contain a N-acetyl-N-hydroxycadaverine terminus, similar to desferrioxamine D. The new acyl DFO-D analogues discovered here also range from short, medium and long acyl chained derivatives (Table 1). It is likely the acyl DFOs are branched analogues as previous studies noted that it was unlikely the chains were fully linear since Streptomyces spp. utilise branched starter units in fatty acid synthesis.26,27 This is also the case for the suite of tenacibactin analogues observed in the data. These branched acyl chain siderophores were originally observed in Tenacibaculum sp. A4K-17,38 and are essentially truncated versions of acyl DFO-Bs that lack the N-terminal N-hydroxycadaverine.
Streptomyces coelicolor has shown broad substrate specificity in regard to medium and long acyl chain incorporation into desferrioxamine structures,19 but not short acyl chained groups, where Streptomyces sp. S29 can facilitate production of short, medium, and long acyl chained DFO-B analogues. Traxler et al. showed the ability of other Actinobactera to react to the presence of S. coelicolor through the production of acyl DFO-B analogues.26 Interestingly, we see this same phenomenon for the coculture of Streptomyces sp. S29 and the fungi. Medium and long chained acyl DFO-Bs (including unsaturated derivatives) were almost exclusively produced in coculture whereas short, medium and long chained acyl DFO-Ds were produced under both mono- and coculture conditions. The tenacibactin analogues were distributed amongst both cultures (Table 1). It was hypothesized these long-chained derivatives might be a way for the uptake of these analogues to bypass the DesE uptake mechanism, instead they ‘hand-off’ the Fe(III) via membrane bound ferrioxamines.19,43
Legonoxamine F contains one 1,6-diaminohexane instead of the typical 1,5-diaminopentane. This is a well-established phenomenon in siderophore biosynthesis31,44 linking it to the promiscuity of the S29 des biosynthetic gene cluster. Legonoxamine G contains a N-hydroxyputrescine moiety due to the initial loss of 104 Da instead of the typical 118 Da associated to N-hydroxycadaverine. Putrebactin is the archetypal putrescine containing siderophore,45 and as its biosynthetic gene cluster has been shown to lack genes coding for production of putrescine, it is assumed that instead, this precursor is sequestered from metabolite pools.46 Interestingly, researchers have also found that under putrescine-depleted conditions, Shewanella putrefaciens shifts to producing desferrioxamine B when cadaverine precursors are available.47 In desferrioxamine biosynthesis, incorporation of the putrescine precursor in desferrioxamine A1 has been shown before in multiple species30,44,48 and now in this study with the discovery of legonoxamine G.
Legonoxamine A glycoside was observed using the predicted molecular formula (C37H62N6O13) and the loss of a sugar (−162 Da). An additional putative diglycosylated DFO (m/z 899.4827), with a predicted molecular formula of C38H70N6O18, was observed in the mass spectrum but upon selected ion monitoring (SIM), produced insufficient fragment ions for structural characterisation. The production of glycosylated DFOs has only been reported once before: nocardamin glucuronide, isolated from Streptomyces sp. 80H647.49 There are only a few reports of microbial enzymes that produce glucuronides,50 yet, the addition of glucose or other hexoses are more commonplace in bacterial natural products. We propose legonoxamine A glycoside undergoes a similar glycosylation mechanism to nocardamin glucuronide, occurring on a N-hydroxyl.
Microbial cytochrome p450s are responsible for adding complexity to fatty acids to be used as building blocks for more complex metabolites or as signalling molecules.52 Previous studies have shown the bacterial production of dicarboxylic acids might occur via cytochrome p450 monoxygenase.53 In order to determine the presence of enzymes that facilitate ω-oxidation in S29, protein sequences that belong to the long-chain fatty acid ω-monooxygenase enzyme class (EC 1.14.14.80) and from fatty acid ω-oxidation gene ontology (GO:0010430) were obtained from the Uniprot database. BLAST results of these sequences against the S29 draft genome reveals just one hit (Supplementary data, ESI†), comprising 3 orthologs with mid-term amino acid identity (40–50%), a methyl-branched lipid ω-hydroxylase (cyp124) from Mycobacterium spp. (∼43% identity) (Uniprot: P9WPP2, P0A517 & P9WPP3). Pfam sequence search confirms cytochrome P450 domain for this protein and further experimentation could confirm its ω-monooxygenase activity. Searching Uniprot under these search criteria returns only a handful of bacterial sequences with a couple belonging to Actinobactera, therefore, the possibilities to track this activity in the S29 genome are very narrow resulting in an inconclusive result regarding the presence of an ω-monooxygenase in S29.
The molecular network was also used to evaluate the production of the DFOs between the two cocultivations with A. niger and B. cinerea and the Streptomyces sp. S29 monoculture (Fig. S3, ESI†). The results showed that 32 of the 71 ions were present in the monoculture, whereas all of the remaining were found in coculture extracts only (Table 1). There was no qualitative significant difference in the siderophores produced under cocultivation with A. niger compared to those produced in cocultivation with B. cinerea, indicating the potential broad response to a fungal microbe in the environment. These results might point to DFO and siderophore production in a broader context as a common chemical response to other microbes, whether it be bacterial or fungal, in close environmental proximity to the primary organism.
In order to evaluate the accuracy of this analysis, a characterized DFO in this study (legonoxamine E, Fig. 5) and an unknown DFO observed using MS2LDA (m/z 650.37, [M + 2H]2+) were more closely analysed (Fig. S5, ESI†). The unknown desferrioxamine is a singleton (Fig. S5B, ESI†) in both the original network and the merged network, yet, MS2LDA identifies it contains DFO motifs and upon closer inspection and prediction of its molecular formula (C59H103N12O20). The potential of this putative DFO being a dimer was ruled out based on fragmentation analysis lacking the monomer. Elucidation of its full structure was not able to be completed with MS/MS data alone, therefore, further studies are underway to characterise its structure. Nonetheless, this example displays the utility of MS2LDA to evaluate the chemical potential of a bacterium and its potential use in these types of studies in the future.
![]() | ||
Fig. 5 Representation of desferrioxamine structural diversity of Streptomyces sp. S29. (A) and (B) new legonoxamine derivatives (purple), tenacibactins (orange), DFO-Ds (red), and a variety of DFO-Bs (blue) were all found. Full MS/MS data and annotations can be found in ESI† for all 41 new analogues discovered. |
![]() | ||
Fig. 6 Metal-complexed molecular families. The GNPS library was able to identify multiple metal ion complexes (hexagonal nodes). All other features observed in the molecular network were assignable as Fe(III) or Al(III) bound ferrioxamines. Table S1, ESI† contains accurate mass information on all complexed ions found in this figure. |
Region 7 (Fig. S53, ESI†) contains a siderophore BGC in Streptomyces sp. S29 with high homology to desA-D genes and to the ferric-siderophore uptake and utilization genes (desE-F). This des cluster follows the same structure as many Streptomyces spp., including the position of iron boxes upstream putative desE (s29_002291) and desA (s29_002293) genes (Fig. S31, ESI†). DesC of the des BGC from Streptomyces sp. S29 (s29_2295) exhibits high identity to DesC from S. coelicolor (72%) and to LgoC from Streptomyces sp. MA37 (69%). In S. coelicolor A3(2), transcription of des BGC is repressed by the divalent metal-dependent regulatory (DmdR) protein DmdR1 and derepressed by iron limitation.58 BLASTN/P analysis revealed a putative DmdR repressor in Streptomyces sp. S29 (s29_002782) with 89% nucleotide and 93% amino acid identity to DmdR1 (AJ271797.1/CAC28070.1). BLASTN analyses identified eight putative DmdR-binding sites (iron boxes) in Streptomyces sp. S29 by query of putative iron boxes found upstream in different ORFs in S. coelicolor A3(2) genome.19
Acyltransferases involved in hydroxamate biosynthesis normally exhibit narrow substrate tolerance (e.g. IucB, which shows a high degree of specificity towards acetyl-CoA).19 The array of DFOs produced by S. coelicolor (B, E, G1 and amphiphilic) and aryl DFOs (legonoxamine A-B) produced by Streptomyces sp. MA37, has been attributed to substrate tolerant acyltransferases DesC and LgoC, respectively.19,59 Senges et al. acknowledged in their evaluation of the Streptomyces chartreusis secretome that in iron depleted minimal medium they also observed additional desferrioxamines produced and that the overall biosynthetic potential may be much broader than what has been predicted.28 Cloning and more detailed analysis of the Streptomyces sp. S29 des gene cluster is ongoing to link the chemical potential shown in this study.
In conclusion, the chemistry observed in the Streptomyces sp. S29 secretome shows it potential as an agent to prevent fungal infection, evidenced by its locking away of Fe(III) via overproduction of desferrioxamines. New analogues were observed in both monoculture and cocultivation studies with fungal phytopathogens, A. niger and B. cinerea. Streptomyces sp. S29 seems to produce an arsenal of desferrioxamine analogues using any available precursor materials available whether that be putrescine, cadaverine, aryl, acyl, or long chained dicarboxylates. The role of bacterial communities in plant and crop protection is becoming more important due to the growing world population, with a strong emphasis on improving strategies at preventing phytopathogenic infection. Overall, the production of 46 new and 25 previously reported siderophores indicates this Streptomyces sp. should be studied further for its plant protective properties as a biocontrol agent.
Disc diffusion assays were adapted for the microbes tested using the Clinical and Laboratory Standards Institute M44 Method for Antifungal Disk Diffusion Susceptibility Testing of Yeasts.60 Extract concentrations were standardised to 200 μg mL−1 for all bioassays conducted.
In order to conduct post-fermentation streaks, 50 μL of fermentation was aseptically transferred onto a ISP2 agar plate. 24 and 48 hour incubation (28 °C) intervals were checked for fungal growth. Due to the faster growth rate of the fungi, we would expect any presence of fungal spores to lead to the observation of mycelial growth, yet no fungal growth was observed.
Each liquid fermentation included Diaion HP-20 resin (3 g/50 mL) that was sterilised with culture medium. After 7 days, the resin was vacuum filtered, subsequently washed five times with Milli-Q water and finally macerated exhaustively in methanol (3×). The methanolic extract was then dried on a rotary evaporator and subsequently underwent Kupchan partitioning.61 Finally, the sec-butanol fraction underwent C18 solid-phase extraction (Phenomenex), with each separation yielding 5 fractions of increasing methanol and these samples were subsequently prepared for LC-MS studies.
The 100% methanol fractions from the Botrytis and Aspergillus co-culture were subjected to further separation via reversed phase-HPLC with a Waters Sunfire C18 semi-preparative column (5 μM, 100 Å, 250 × 10 mm) wand a solvent system of A-95/5 MeOH/H2O and B-MeOH. The Agilent 1200 HPLC utilized a solvent gradient of 0% B to 100% B at 2.0 mL min−1 over 50 min.
For documentation on running MS2LDA, the user is referred to the original publication and guidelines, which are also present at ms2lda.org.62 All spectral images in the ESI† were created using Metabolomics Spectrum Resolver (https://metabolomics-usi.ucsd.edu/).63
All siderophores dereplicated and elucidated in this study can be found in Table 1. The structure elucidation of each new derivative and its corresponding MS/MS spectra were generated using Metabolomics Spectrum Resolver are displayed in the ESI† (Table S2 and Fig. S7–S52). All MS/MS spectra were also annotated and deposited in the GNPS library, as well as the original datasets (Table S2, ESI†).
All data files and workflows can be found at the links below: GNPS molecular networking job: https://gnps.ucsd.edu/ProteoSAFe/status.jsp?task=3c73305ef7ed48e19cb18e50b8b6b2bd. MS2LDA annotation experiment: http://ms2lda.org/basicviz/summary/1319/. MASSIVE: doi:10.25345/C5772V.
Footnotes |
† Electronic supplementary information (ESI) available: Tables of known and new desferrioxamine derivatives and their corresponding CCMS Library IDs and Metabolomics Resolver Images (Table S2), disc diffusion assay results, desferrioxamine containing subnetworks, new desferrioxamine analogues with annotated structures and MS2 spectra, ITS regions for fungal phytopathogens, cytochrome p450 enzyme sequence. See DOI: 10.1039/d0mo00084a |
‡ DLS and ED contributed equally to this work. |
This journal is © The Royal Society of Chemistry 2021 |