Spatiotemporal Monitoring of the Antibiome Secreted by Bacillus

Apr 8, 2014 - Some soil Bacilli living in association with plant roots can protect their host from infection by pathogenic microbes and are therefore ...
2 downloads 17 Views 4MB Size
Download PDF
Article pubs.acs.org/ac

Spatiotemporal Monitoring of the Antibiome Secreted by Bacillus Biofilms on Plant Roots Using MALDI Mass Spectrometry Imaging Delphine Debois,*,†,∥ Emmanuel Jourdan,‡,∥ Nicolas Smargiasso,† Philippe Thonart,§ Edwin De Pauw,† and Marc Ongena*,§ †

Mass Spectrometry Laboratory, Chemistry Department, University of Liege, 4000 Liege, Belgium Centre Wallon de Biologie Industrielle, University of Liege, 4000 Liege, Belgium § Gembloux Agro-Bio Tech, University of Liege, 5030 Gembloux, Belgium ‡

S Supporting Information *

ABSTRACT: Some soil Bacilli living in association with plant roots can protect their host from infection by pathogenic microbes and are therefore being developed as biological agents to control plant diseases. The plant-protective activity of these bacteria has been correlated with the potential to secrete a wide array of antibiotic compounds upon growth as planktonic cells in isolated cultures under laboratory conditions. However, in situ expression of these antibiotics in the rhizosphere where bacterial cells naturally colonize root tissues is still poorly understood. In this work, we used matrixassisted laser desorption/ionization mass spectrometry imaging (MALDI MSI) to examine spatiotemporal changes in the secreted antibiome of Bacillus amyloliquefaciens developing as biofilms on roots. Nonribosomal lipopeptides such as the plant immunity elicitor surfactin or the highly fungitoxic iturins and fengycins were readily produced albeit in different time frames and quantities in the surrounding medium. Interestingly, tandem mass spectrometry (MS/MS) experiments performed directly from the gelified culture medium also allowed us to identify a new variant of surfactins released at later time points. However, no other bioactive compounds such as polyketides were detected at any time, strongly suggesting that the antibiome expressed in planta by B. amyloliquefaciens does not reflect the vast genetic arsenal devoted to the formation of such compounds. This first dynamic study reveals the power of MALDI MSI as tool to identify and map antibiotics synthesized by root-associated bacteria and, more generally, to investigate plant−microbe interactions at the molecular level.

M

embers of the Bacillus genus devote a large part of their genome to the formation of multiple antibiotics of various biochemical origins.1−3 This applies particularly to some isolates of the Bacillus amyloliquefaciens species living in association with plants in the rhizosphere which corresponds to the thin layer of soil surrounding root tissues and hosts most of the underground microbial life. These strains may dedicate more than 8% of their genetic equipment to nonribosomal synthesis of lipopeptide (LP)- and polyketide (PK)-type antibiotics.4,5 These NRPS/PKS antibiotics are tightly involved in the multitrophic interactions occurring between the producing bacterium, the host plant, and the other organisms sharing the same ecosystem. PKs such as macrolactin, bacillaene, and difficidin display strong antibacterial activities6,7 while LPs of the iturin and fengycin families are mostly fungitoxic and thus tightly involved in direct inhibition of fungal plant pathogens. Surfactins, representing the third main class of LPs, are involved in root colonization by the producing cells but were also identified as elicitors of host plant immunity leading to increased resistance toward further pathogen ingress.8,9 These NRPS/PKS compounds thus play pivotal © XXXX American Chemical Society

roles not only for ecological competence in terms of population establishment in the rhizosphere but also for protection of the host plant against infection by various pathogens. This antibiotic arsenal makes some B. amyloliquefaciens isolates among the most efficient microbial biopesticides developed to control plant diseases.10,11 However, the success of these biocontrol products globally suffers from some inconsistency in their efficacy upon field or greenhouse applications. It is notably due to the fact that environmental conditions prevailing in the phyllosphere are quite specific and may strongly impact the production of bioactive compounds by the bacterium. Indeed, the physiological status of bacterial cells is driven by global nutrient limitation, slow growth rate, and evolvement in biofilm-related structures on the surface of plant tissues as roots. These conditions are far from in vitro liquid cultures in rich media that are commonly used to study antibiotic Received: January 22, 2014 Accepted: April 8, 2014

A

dx.doi.org/10.1021/ac500290s | Anal. Chem. XXXX, XXX, XXX−XXX

Analytical Chemistry

Article

Determination of Bacillus Populations on Roots. At each time point in imaging experiments, rhizosphere samples (root and surrounding gelified medium) were collected and vortexed during 3 min in peptone water (peptone 1 g/L, NaCl 5 g/L, Tween 80 1 g/L) supplemented with glass beads for efficient recovery of cells adhering to the root tissues. Roots were then removed and weighed, and the resulting cell suspensions were diluted and plate-counted on LB medium (tryptone 10 g/L, yeast extract 5 g/L, NaCl 5 g/L, agar 1.75%). Bacillus populations on plants cultivated in perlite were established as already described.19 Structural Analysis of the New Variant. Identification of the new variant of surfactin was based on mass accuracy and fragmentation data, as described in ref 18 and the Supporting Information. Briefly, a Fourier Transform Ion Cyclotron Resonance (FTICR) mass spectrometer (9.4T Solarix; Bruker Daltonics, Bremen, Germany) was used to perform exact mass measurements. These data allowed for elemental composition searches which, allied to tandem mass spectrometry (MS/MS) results, lead to the identification of the new variant of surfactin. Extraction of Antibiotics for UPLC−MS Analyses. In samples used for MALDI MS imaging, antibiotics secreted by the bacterium were recovered from the gelified medium surrounding roots (approximately 1 g) by adding 5 mL of ACN/0.1% FA 50:50 v/v. The resulting sample was ultrasonicated for 1 min, and metabolites were extracted under stirring for 2 h, room temperature. Extraction was performed twice on the same sample. Both extracts were pooled, and the resulting solution was evaporated to dryness in a speed-vac. The material was solubilized in 200 μL of the same solvent and centrifuged at 10 000 g for 10 min before ultrahigh-performance liquid chromatography−mass spectrometry (UPLC−MS) analysis as described in the Supporting Information. Antibiotics produced on plants cultivated in perlite under gnotobiotic conditions were extracted as previously reported19 and measured as described in the Supporting Information.

synthesis. It is thus crucial to demonstrate that antibiotics are actually secreted and may accumulate in the microenvironment in biologically relevant quantities. However, the number of reports on antibiotic in situ production is still very limited, reflecting the inherent difficulties in detecting these small-size compounds in complex matrices.9,12,13 The present work was thus initiated with the global aim to provide further insights into the variety of antibiotics that can be secreted by Bacillus cells evolving on plant roots. Matrixassisted laser desorption/ionization mass spectrometry imaging (MALDI MSI) was first applied to map specific metabolites in animal or human tissue sections, but this technique is now also often used to study microbial colonies, 14−16 bacterial competition/inhibition,17 or to identify lipopeptides involved in the antagonism of fungal plant pathogens.18 We developed a gnotobiotic system in order to exploit MALDI MSI to monitor the spatiotemporal accumulation of NRPS/PKS compounds produced in the rhizosphere by the plant-beneficial B. amyloliquefaciens strain S499. By combining various experimental approaches, our results show that the Bacillus antibiotic-related secretome is restricted to lipopeptides with surfactins as sole ingredients that accumulate in biologically relevant quantities in the medium surrounding roots.

■

EXPERIMENTAL SECTION

Sample Preparation for MALDI MS Imaging. Tomato seeds (Solanum lycopersicum Merveille des marchés) were disinfected in ethanol 75% during 2 min followed by a 1.25% commercial bleach solution containing 0.01% of Tween 80. Seeds were then rinsed 5 times in sterile water and germinated on solid nutrient solution (5 mM Ca(NO3)2, 5 mM KNO3, 2 mM MgSO4, 1 mM KH2PO4, and micronutrients, supplemented to 1.75% agar), diluted twice, and directly poured on indium−tin oxide (ITO) coated glass slide in Petri dishes. Petri dishes were then incubated vertically in a greenhouse at 27 °C with a 16 h photoperiod. For bacterial inoculation, 7 days old tomato seedling roots were treated with 10 μL drops of B. amyloliquefaciens S499 cell suspension at OD600nm = 0.02. The bacterial suspension was prepared from a 16 h old culture centrifuged, washed, and resuspended in 10 mM MgSO4. After incubation, root-colonized samples were collected and processed for MALDI MS imaging. Leaves were cut, and the ITO-coated glass slide was removed from the Petri dish. A maximum of agar medium was discarded, and the slide was placed in a vacuum desiccator until complete drying. The pressure was reduced to 5−15 mbar. CHCA (Sigma-Aldrich, Steinheim, Germany) solution was prepared at a concentration of 5 mg/mL in ACN/0.2% TFA 70:30 v/v. The application of the matrix was performed with an ImagePrep automated device (Bruker Daltonics, Bremen, Germany). Details about method parameters are given in the Supporting Information. MALDI MSI Acquisitions. Ions images were recorded with an UltraFlex II TOF/TOF mass spectrometer (Bruker Daltonics, Bremen, Germany) using FlexImaging 2.1 software, with a pixel step size for the surface raster set to 150 μm. FlexAnalysis 3.0 (Bruker Daltonics, Bremen, Germany) was used to submit the whole data set to a processing method including smoothing (Gauss, 0.05 m/z, 1 cycle), baseline correction (TopHat), and external calibration. The data set was then submitted to normalization in FlexImaging 2.1. Ion density maps were created with a mass filter width fixed at 0.08 Da.

■

RESULTS AND DISCUSSION Potential of B. amyloliquefaciens S499 for Production of NRPS/PKS Antibiotics. Whole-genome sequencing of strain S499 using the Illumina next-generation sequencing platform (for details, see the Supplemental Experimental Procedures section in the Supporting Information) revealed B. amyloliquefaciens strain FZB42 as the closest relative, with more than 98% similarity at the genomic level. S499 contigs containing predicted coding sequences CDS for NRPS and PKS genes were therefore blasted against the genome sequence of FZB42 considered as the type strain for the plant-associated B. amyloliquefaciens subsp. plantarum clade.20,21 It allowed further assembly leading to the identification of large parts of the operons coding for synthetases responsible for the formation of the lipopeptide (LP) antibiotics iturin, surfactin, and fengycin, and for synthesis of the polyketides macrolactin, bacillaene, and (oxy)difficidin (Supporting Information Figure S1). Genes directing the nonribosomal synthesis of bacillibactin (dhb A-F) and bacilysin (bac(ywf)A-E) were also identified in the genome of S499. Both compounds may be involved in biocontrol of plant diseases due to their siderophore (iron deprivation for pathogens) and antibacterial activities, respectively.6,22 In addition, the sequence corresponding to the acnA gene coding for the precursor of amylocyclicin was detected in one of the S499 contigs. However, genes of other Bacillus antibiotics such as mersacidin and plantazocidin (not NRPS/ B

dx.doi.org/10.1021/ac500290s | Anal. Chem. XXXX, XXX, XXX−XXX

Analytical Chemistry

Article

Figure 1. Left panel: description of the methodology developed in this work. Roots are grown sole or together with B. amyloliquefaciens cells in a Petri dish. Photographs show roots after X days of growth. On the left, these are untreated roots, while on the right, biofilm-covered lateral hair roots (X dpi) are presented. Right panel: (A) average MALDI mass spectrum recorded on an A. thaliana root aged of 28 days, incubated with B. amyloliquefaciens S499 during 12 days, (B) microscope picture of the analyzed area on the root, (C) MALDI image of the [C14 − Surf + Na]+ ion, (D) MALDI image of the [C14 − ItuA + Na]+ ion, (E) MALDI image of the [C14 − FengB + Na]+, (F) average MALDI mass spectrum recorded on tomato root aged of 21 days, incubated with B. amyloliquefaciens S499 during 14 days, (G) microscope picture of the analyzed area on the root, (H) MALDI image of the [C14 − Surf + Na]+ ion, (I) MALDI image of the [C14 − ItuA + Na]+ ion, (J) MALDI image of the [C14 − FengB + Na]+, (K) average MALDI mass spectrum recorded on tobacco root aged of 29 days, incubated with B. amyloliquefaciens S499 during 13 days, (L) microscope picture of the analyzed area on the root, (M) MALDI image of the [C14 − Surf + Na]+ ion, (N) MALDI image of the [C14 − ItuA + Na]+ ion, (O) MALDI image of the [C14 − FengB + Na]+. Peaks corresponding to surfactins, iturins, or fengycins are labeled with S, I, or F, respectively. Color scale represents relative intensity for each signal: 5%−100%. Scale bar: 2 mm.

PKS) are seemingly not present in S499 even as fragmentary parts of the clusters.23,24 The genetic ability of S499 to synthesize all these metabolites was counterchecked by UPLC− ESI-MS analyses of supernatant extracts after in vitro growth of the bacterium in synthetic rich media. All three lipopeptide families are readily secreted by the strain (Supporting Information Figure S2A). The presence of multiple peaks for each LP family is due to the fact that the lipoinitiation reaction at the entry of the NRPS assembly line may use various fatty acids to bind the first amino acid, leading to the coproduction of several homologues differing in the length and isomery of the acyl chain.25,26 Extracted ion chromatograms also allowed readily detecting the PKS products macrolactin A and D, bacillaene, and difficidin (Supporting Information Figure S2B) together with the NRPS-formed bacilysin and the siderophore bacillibactin (data not shown). However, we could not detect the amylocyclicin product in any extract prepared from liquid cultures in various media. MS Imaging of the S499 Antibiome Produced on Different Plants. In order to decipher antibiotic production in

planta, we developed a gnotobiotic system in which the plantlet and the associated bacterium were grown on a gelified medium covering the MALDI target plate. The design of the methodology developed in this study is described in Figure 1, left panel. Imaging experiments were first conducted on roots inoculated with strain S499 approximately 2 weeks before analysis. These experiments were performed in parallel on tomato, tobacco, and Arabidopsis thaliana plantlets. At that time point, root systems of all plants were readily colonized by S499 forming consistent biofilm (Figure 1, left panel; see also Figure 3). Imaging results revealed quite similar antibiotic patterns, as demonstrated by average MALDI mass spectra (Figure 1, right panel, parts A, F, and K). In all cases, the main molecular ions detected in the vicinity of the roots (as [M + Na]+ or [M + K]+ ions) correspond to cyclic LPs, albeit in very different proportions for the three families (Figure 1, parts C−E, H−J, and M−O). Surfactin homologues (C12- to C15-acyl chains) together represent more than 90% of the whole LP production. Ion species corresponding to C13-, C14-, and C15-iturin A homologues were also detected as well as fengycins but in C

dx.doi.org/10.1021/ac500290s | Anal. Chem. XXXX, XXX, XXX−XXX

Analytical Chemistry

Article

antibiotic that could be clearly measured among all the bacterial products upon UPLC−ESI-MS analyses of rhizosphere extracts from the three plant species (Figure 2C). The compound was recovered at concentrations of 1.5 ± 1.1 μM in the root environment of these perlite-grown plantlets (mean and SD calculated from two plants coming from two independent experiments, n = 4). Again, only minute amounts of the other LPs (below the limit for an accurate quantification) and no PKS compounds were detected. These data show that the LP pattern observed in the gnotobiotic conditions used for MSI is not due to the specific experimental conditions used or influenced by the plant type or developmental stage. In order to evaluate strain specificity regarding this in planta antibiotic signature, we tested in parallel another isolate, strain GA1, of the same B. amyloliquefaciens species. This strain also synthesizes a range of polyketides27 but naturally secretes higher amounts of iturins and fengycins compared to S499 when cultivated under optimized in vitro conditions (Supporting Information Figure S3). Both in the gnotobiotic and perlite systems, slightly higher signals were observed in the case of GA1 for iturin- and fengycin-specific molecular ions, but it again corresponds to minor quantities compared to surfactins (Supporting Information Figure S3, parts B and C). By testing different host plant species and different B. amyloliquefaciens isolates, our data thus demonstrated that, in all instances, surfactins were detected in the root environment in much higher relative amounts compared to iturins and fengycins. Some factors inherent to the development of Bacillus in the rhizosphere may impact LP production and explain changes in the secreted pattern compared to planktonic cells grown in rich lab media. For instance, the nutritional status imposed by host plant exudation may strongly influence LP signature. It has been notably reported that surfactin but not iturin nor fengycin production by S499 is favored in the presence of organic acids, the main components of tomato exudates.12 Moreover, developmental factors may also impact LP production. Surfactin synthesis by S499 cells is very effective compared to iturins and fengycins during early biofilm formation and upon growth of cells as colonies on gelified media compared to cells living freely in liquid cultures.28,29 Population-driven responses may also play an important regulatory role regarding the synthesis of surfactin which is under the control of a complex network that governs cellular differentiation and quorum sensing via ComX and other pheromones of the Phr family.30,31 As a matter of fact, no ions corresponding to PKS were detected during the imaging runs with any of the three plant species. All those compounds are in general readily detectable by MALDI MS of culture supernatant extracts.5,6,20 On the basis of pictures of plant roots colonized by S499 (Figures 1 and 3), bacteria do form biofilms. However, the absence of PKS in our extracts is not expected to be related to these specific growth conditions. Indeed we did not observe major changes in the amounts of PKS produced upon cultivating the strain in a medium containing the main carbohydrates present in tomato exudates or in artificial medium. There was also no drastic influence of growing cells as pellicle on gelified medium compared to planktonic cells (data not shown). Fan and coworkers showed that the dif, bae, and mln polyketide genes are slightly up-regulated upon growth of strain FZB42 in maize root exudates rich in organic acids,32 suggesting that these antibiotics are readily formed under these conditions. In our study, PKS were neither detected in perlite-grown plantlets nor

much lower amounts. Ion images (Figure 1, parts C−E, H−J, and M−O) show that cyclic LPs all accumulated in the medium surrounding roots but did not necessarily exhibit the same distribution (Figure 1, parts C and D). A similar antibiotic profile was observed on replicate samples via UPLC−ESI-MS analyses of compounds extracted with ACN from the surface of the gelified medium surrounding tomato roots. Surfactin ions were abundant, while very little amounts of iturin and fengycin were detected (Figure 2B). It

Figure 2. Extracted ion chromatograms of each LP family from extracts prepared from different cultivations of B. amyloliquefaciens S499: (A) supernatant of liquid cultures, (B) surface of the gelified medium surrounding tomato roots, and (C) root environment of tomato roots grown on perlite substrate.

corresponded to concentrations of 8.3 ± 1.3, 0.8 ± 0.3, and 0.07 ± 0.04 μM for surfactins, iturins, and fengycins, respectively (mean and SD calculated from three plants coming from two independent experiments, n = 6). These concentrations were calculated based on a 2 mL volume of gelified medium surrounding roots in the diffusion zone of the LPs. In order to verify that our observations were not dependent on the experimental setup, we developed a third approach involving more developed plants, in which S499 was evaluated for the production of antibiotics upon colonization of three week old plants grown aseptically on perlite substrate. The bacterium readily established in the rhizosphere to reach quite high populations of 8.1 ± 3.5 × 107, 1.6 ± 1.9 × 108, and 1.4 ± 1.2 × 109 CFU/g root for tomato, tobacco, and Arabidopsis, respectively. Also in this system, surfactin was the sole D

dx.doi.org/10.1021/ac500290s | Anal. Chem. XXXX, XXX, XXX−XXX

Analytical Chemistry

Article

production of other antibiotics, but no other known compounds except LPs could be identified in any extracts. In silico analyses of some B. amyloliquefaciens genomes have revealed a number of so-called orphan gene clusters coding for other unknown antibiotics.5,20,21 Further investigation is necessary to identify such orphan clusters in the S499 genome and to determine whether rhizosphere conditions are conducive or not for the expression of these cryptic genes. Time-Course Imaging of LPs Pattern on Tomato Roots. Time-course MSI was performed at 3, 7, 14, and 32 days post inoculation (dpi) to determine possible changes in the antibiome secreted by S499 in the early and later phases of the colonization process. Control, untreated tomato roots have also been analyzed (Figure S4 of the Supporting Information). Bacterial cells developed quite rapidly at the root surface as a thin biofilm along the main root after 3 days (Figure 3A). This population keeps increasing in the next 11 days to establish strong biofilm and reach a concentration of approximately 109 CFU/g root including 75% of spores (Figure 3B). After a short decline, it then stabilizes over the remaining experimental time (until 35 dpi) reflecting the equilibrium imposed by the scarcity of space and nutriments that limits the surviving cell population. Time-course monitoring of LPs revealed surfactin accumulation in the root zone as early as 3 dpi, while production of iturins (20% of total LPs) and fengycins (3%) is delayed to 7 and 14 dpi, respectively (Figure 3C). Our time-course analysis thus revealed sequential secretion of the three LP families, but these compounds were almost the sole antibiotic products that could be detected in the medium surrounding root-associated biofilms. Surfactin is quite rapidly synthesized and accumulates to significant amounts within 3 dpi, while iturin and fengycin productions are delayed to the end of the aggressive phase of colonization. This could be related to the exponential phase-dependent expression of srf genes and stationary phase-dependent synthesis of fengycin and iturin observed in batch liquid cultures.29,34 On another hand, early surfactin secretion to reach micromolar concentrations in

Figure 3. (A) Pictures of biofilms formed by strain S499 on tomato roots at 3 and 14 dpi. (B) Population of S499 cells colonizing tomato roots over time. (C) Relative abundances of the secreted LPs families over time. Abundances are mean values, calculated from intensities measured on MALDI average mass spectra.

in planta. It thus agrees with results from Fan and co-workers who could not recover PKS from rhizosphere of FZB42colonized Lemna plantlets.33 We also looked for in planta

Figure 4. (A, F, K, P) Microscope images of tomato roots colonized by S499 for 3, 7, 14, and 32 days, respectively. (B, G, L, Q) MALDI images of [M + Na]+ of C14 − surfactin (m/z 1044.66). (C, H, M, R) MALDI images of [M + Na]+ of C15 − surfactin (m/z 1058.67). (D, I, N, S) MALDI images of [M + K]+ of C14 − surfactin (m/z 1060.64). (E, J, O, T) MALDI images of [M + K]+ of C15 − surfactin (m/z 1074.65). Color scale represents relative intensity for each signal: 5%−100%. Scale bar: 2 mm. E

dx.doi.org/10.1021/ac500290s | Anal. Chem. XXXX, XXX, XXX−XXX

Analytical Chemistry

Article

Figure 5. (A and B) MS/MS spectra recorded on the environment of a 39 days old tomato root, 32 dpi with S499 cells, with m/z 1000.6795 and m/ z 1014.6953 as precursor ion, respectively.

medium. The origin of such discrimination may rely on changes in ion fluxes in/out of root cells over time but it is still unclear and needs to be further investigated. However, it shows how powerful the methodology may be, allowing one to possibly point out complex, unexpected mechanisms. These data also show that some particular events occur at late time points (loss of precise distribution for surfactin homologues and detection of new variant) and pave the way for future investigation. Detection and Structural Analysis of a New Variant of Surfactin. Interestingly, MALDI MSI also revealed, at later time points of the interaction, the accumulation of previously unidentified compounds with m/z ratios 986.66, 1000.68, and 1014.69. The mass range of these ions, the mass difference between two consecutive peaks (14 Da), as well as their localizations on MALDI images (Supporting Information Figure S6C−F) suggest that they could correspond to LP homologues. FTICR MS was used to measure exact masses and determine the most probable elemental compositions (Supporting Information Table S1). It yielded molecular formulas very similar to surfactin sodium adducts based on the number of nitrogen and oxygen atoms. Further information was provided by tandem MS analyses, and the corresponding spectra (Figure 5, parts A and B) display some striking similarities with those obtained for known surfactin homologues (not shown). y-type ion series43,44 with m/z 437.2754, m/z 550.3608, and m/z 663.4466 were detected in the MS/MS spectra of each novel ion, indicating a similar peptide moiety with partial sequence Leu6-Leu7. b-type ions are formed on the N-terminus side of the peptide and logically differ by 14 Da from one homologue to the other as they contain fatty acid

the environment surrounding roots could be of biological relevance since this lipopeptide could play a role in root colonization35,36 by helping the bacterium to move on tissues37−39 and to start forming biofilm structures.41,42 Also of biological relevance in the context of biocontrol, surfactin has been demonstrated to act as elicitor of immune-related responses in the host plant when applied at 2−10 μM concentrations (induced systemic resistance, ISR).40−42 By contrast, a significant contribution of antifungal iturins and fengycins in direct antibiosis toward phytopathogens is questionable since we show here that these compounds are released at submicromolar quantities by S499 on roots. Such amounts are far from minimal inhibitory concentrations (typically in the 10−30 μM range)26 and are therefore probably not sufficient per se to provide some fungitoxic effect. MSI also showed that Na and K adducts of LP homologues localize differentially. Over the first 3 weeks, the sodium adducts of C14- and C15-surfactins are detected with a strong signal far from the root tissue (Figure 4, parts B, G, and L for C14 and Figure 5, parts C, H, and M for C15; see also Figure S5 of the Supporting Information). However, a very different distribution is observed for potassium adducts as the highest signals were detected close to the roots with, for example, [M + K]+ of C14 and C15 homologues accumulating in an area smaller than 500 μm wide both sides of the root (Figure 4, parts N and O). At late time point, a far less precise localization of both homologues is observed and ions are detected quite homogeneously over the whole area (Figure 4Q−T). Actually, the differential detection of these surfactin forms reflects the local distribution of sodium and potassium ions present in the F

dx.doi.org/10.1021/ac500290s | Anal. Chem. XXXX, XXX, XXX−XXX

Analytical Chemistry

■

Article

CONCLUSIONS Multitrophic interactions occurring in the rhizosphere between pathogenic microbes, beneficial bacteria, and the host plant are complex and still globally poorly understood. This is notably due to technological stumbling blocks and lack of suitable tools to deeply investigate this peculiar ecosystem and to determine its content in key substances released by the different partners. In that context, we report here the first application of MALDI MSI to investigate plant−microbe interactions at the molecular level. MS imaging has revealed as powerful tool to identify antibiotic compounds released by root colonizing Bacillus cells and to spatiotemporally resolve their dynamics of production. Globally, data strongly suggest that the antibiome expressed in planta by B. amyloliquefaciens isolates does not reflect the vast genetic arsenal devoted to the formation of bioactive compounds by these bacteria. Some of these Bacillus antibiotics may reach threshold concentrations for activity within the rhizosphere, while others would remain below this level or simply not produced. This new methodology can be obviously adapted for monitoring spatiotemporal antibiotic production patterns from other bacterial genera/species of interest as biocontrol agents. MSI could also be exploited as tool for selecting isolates particularly efficient at expressing their antibiome in planta among a first set of strains primarily screened for huge antibiotic potential in high-throughput and well-defined in vitro conditions.

chains with different lengths. For each homologue, the mass differences between b-type ions also revealed a Leu2-Leu3 sequence at the other extremity of the peptide, indicating a conserved Glu-Leu-Leu-X′-X-Leu-Leu sequence compared to known surfactins (Supporting Information Figure S7). However, there is an average mass difference between b- and y-type ions of known and new surfactins of 43.991 Da (Supporting Information Table S2). b5 and y4 ions are massshifted in the new products, meaning that the modification in peptide sequence is located between Leu3 and Leu6, so either on the Val or the Asp residue. Such a mass difference was interpreted as a loss of CO2 (exact mass: 43.98983, Δm = 1.17 mDa) that could only occur on the aspartic acid side chain. Moreover, peaks with a mass difference of 18 Da are observed for y4 and b5 ions on MS/MS spectra of known surfactins (data not shown). These are due to loss of a water molecule on acidic residues (Glu and Asp). This loss of water is preserved on the b5 ion of unknown compounds but is no longer observed for the y4, strongly suggesting the absence of aspartic acid residue. Some flexibility of specific adenylation domains in NRPS megaenzymes such as the surfactin synthetase has been reported leading to substitutions at precise locations in the peptide sequence, but it usually involves two closely related amino acids of the same type.26,45 Moreover, the type of amino acid residue that can be incorporated and the place in the peptide sequence where modifications may occur are quite fixed and addition of various amounts of Ala (and other amino acids) during S499 growth did not yield any changes in the pattern of surfactin homologues secreted by the strain compared to unsupplemented medium.40 This strongly suggests that the protein domain responsible for the activation/incorporation of the fifth residue in the surfactin assembly line is very specific to aspartate. Using the AntiSmash software,46 we identified the coding sequence corresponding to this domain in the srfAB gene of S499 (Supporting Information Figure S1). NCBI blast revealed a very high identity of 98−100% with other B. amyloliquefaciens strains showing that no specific mutations occurred in the S499 srfAB gene that would have impacted the functionality of the module responsible for incorporation of the fifth residue in the peptide. Therefore, the new surfactin derivatives with m/z 986, 1000, and 1014 may result from some postsynthesis modification of de novo produced surfactin such as decarboxylation of the aspartate lateral chain. Various tailoring enzymes may act at different levels during the formation of NRPS products in Bacillus.47 They are responsible for epimerization, cyclization, oxidation, and methylation reactions to shape the final product, but to our knowledge, decarboxylation activities have not been reported so far. MSI showed that these modified LPs localize similarly to de novo synthesized C14 and C15 homologues (Supporting Information Figure S6). However, their occurrence in the root surrounding medium is delayed suggesting that some plant factors such as decarboxylase activity may be involved in their formation. Whatever the origin of the modification, such substitution of Asp by Ala may impact the global structure of the molecule. Native surfactin typically displays a “horse saddle” topology, and both the nature and place of the various amino acid residues in the cyclic peptide moiety strongly contribute to the stability of this conformation.45,48 Asp/Ala exchange may therefore modify the tridimensional structure of the molecule and consequently its biological properties.

■

ASSOCIATED CONTENT

S Supporting Information *

Additional information as noted in text. This material is available free of charge via the Internet at http://pubs.acs.org.

■

AUTHOR INFORMATION

Corresponding Authors

*E-mail: [email protected]. *E-mail: [email protected]. Author Contributions ∥

D.D. and J.E. contributed equally to this work.

Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS D.D., E.J., N.S., and M.O. are employed by the FRS-FNRS (Fonds National de la Recherche Scientifique, Belgium). The authors thank Laurent Frazil for technical assistance. This work received financial support from the program Fonds de la Recherche Fondamentale Collective (FRFC) no. 2.4567.12 (FRS-FNRS, Belgium).

■

REFERENCES

(1) Hamdache, A.; Lamarti, A.; Aleu, J.; Collado, I. G. J. Nat. Prod. 2011, 74 (4), 893−899. (2) Sansinenea, E.; Ortiz, A. Biotechnol. Lett. 2011, 33 (8), 1523− 1538. (3) Stein, T. Mol. Microbiol. 2005, 56 (4), 845−857. (4) Chen, X. H.; Koumoutsi, A.; Scholz, R.; Eisenreich, A.; Schneider, K.; Heinemeyer, I.; Morgenstern, B.; Voss, B.; Hess, W. R.; Reva, O.; Junge, H.; Voigt, B.; Jungblut, P. R.; Vater, J.; Süssmuth, R.; Liesegang, H.; Strittmatter, A.; Gottschalk, G.; Borriss, R. Nat. Biotechnol. 2007, 25 (9), 1007−1014. (5) Rückert, C.; Blom, J.; Chen, X.; Reva, O.; Borriss, R. J. Biotechnol. 2011, 155 (1), 78−85. G

dx.doi.org/10.1021/ac500290s | Anal. Chem. XXXX, XXX, XXX−XXX

Analytical Chemistry

Article

(35) Bais, H. P.; Fall, R.; Vivanco, J. M. Plant Physiol. 2004, 134 (1), 307−319. (36) Dietel, K.; Beator, B.; Budiharjo, A.; Fan, B.; Borriss, R. Plant Pathol. J. 2013, 29 (1), 59−66. (37) Debois, D.; Hamze, K.; Guérineau, V.; Le Caër, J. P.; Holland, I. B.; Lopes, P.; Ouazzani, J.; Séror, S. J.; Brunelle, A.; Laprévote, O. Proteomics 2008, 8, 3682−3691. (38) Hamze, K.; Julkowska, D.; Autret, S.; Hinc, K.; Nagorska, K.; Sekowska, A.; Holland, I. B.; Séror, S. J. Microbiology 2009, 155 (2), 398−412. (39) Kinsinger, R. F.; Shirk, M. C.; Fall, R. J. Bacteriol. 2003, 185 (18), 5627−5631. (40) Henry, G.; Deleu, M.; Jourdan, E.; Thonart, P.; Ongena, M. Cell. Microbiol. 2011, 13 (11), 1824−1837. (41) Jourdan, E.; Henry, G.; Duby, F.; Dommes, J.; Barthelemy, J. P.; Thonart, P.; Ongena, M. Mol. Plant−Microbe Interact. 2009, 22 (4), 456−468. (42) Ongena, M.; Adam, A.; Jourdan, E.; Paquot, M.; Brans, A.; Joris, B.; Arpigny, J. L.; Thonart, P. Environ. Microbiol. 2007, 9, 1084−1090. (43) Biemann, K. Sequencing of peptides by tandem mass spectrometry and high-energy collision-induced dissociation. In Methods in Enzymology; James, A. M., Ed.; Academic Press: San Diego, CA, 1990; Vol. 193, pp 455−479. (44) Roepstorff, P.; Fohlman, J. Biomed. Mass Spectrom. 1984, 11 (11), 601−601. (45) Bonmatin, J. M.; Laprevote, O.; Peypoux, F. Comb. Chem. High Throughput Screening 2003, 6 (6), 541−556. (46) Blin, K.; Medema, M. H.; Kazempour, D.; Fischbach, M. A.; Breitling, R.; Takano, E.; Weber, T. Nucleic Acids Res. 2013, 41, W204−W212. (47) Giessen, T. W.; Marahiel, M. A. FEBS Lett. 2012, 586 (15), 2065−2075. (48) Shen, H. H.; Thomas, R. K.; Chen, C. Y.; Darton, R. C.; Baker, S. C.; Penfold, J. Langmuir 2009, 25 (7), 4211−4218.

(6) Chen, X. H.; Scholz, R.; Borriss, M.; Junge, H.; Mögel, G.; Kunz, S.; Borriss, R. J. Biotechnol. 2009, 140 (1−2), 38−44. (7) Moldenhauer, J.; Chen, X. H.; Borriss, R.; Piel, J. Angew. Chem., Int. Ed. 2007, 46 (43), 8195−8197. (8) Pérez-García, A.; Romero, D.; de Vicente, A. Curr. Opin. Biotechnol. 2011, 22 (2), 187−193. (9) Raaijmakers, J.; De Bruin, I.; Nybroe, O.; Ongena, M. FEMS Microbiol. Rev. 2010, 34, 1037−1062. (10) Cawoy, H.; Bettiol, W.; Fickers, P.; Ongena, M. Bacillus-based biological control of plant diseases. In Pesticides in the Modern World; Stoytcheva, M., Ed.; InTech: Rijeka, Croatia, 2011; pp 273−301. (11) McSpadden Gardener, B. B. Phytopathology 2004, 94 (11), 1252−1258. (12) Nihorimbere, V.; Cawoy, H.; Seyer, A.; Brunelle, A.; Thonart, P.; Ongena, M. FEMS Microbiol. Ecol. 2012, 79 (1), 176−191. (13) Raaijmakers, J. M.; Mazzola, M. Annu. Rev. Phytopathol. 2012, 50, 403−424. (14) Barger, S. R.; Hoefler, B. C.; Cubillos-Ruiz, A.; Russell, W. K.; Russell, D. H.; Straight, P. D. Antonie Van Leeuwenhoek 2012, 102 (3), 435−445. (15) Watrous, J. D.; Dorrestein, P. C. Nat. Rev. Microbiol. 2011, 9 (9), 683−694. (16) Yang, Y. L.; Xu, Y.; Straight, P.; Dorrestein, P. C. Nat. Chem. Biol. 2009, 5 (12), 885−887. (17) Hoefler, B. C.; Gorzelnik, K. V.; Yang, J. Y.; Hendricks, N.; Dorrestein, P. C.; Straight, P. D. Proc. Natl. Acad. Sci. U.S.A. 2012, 109 (32), 13082−13087. (18) Debois, D.; Ongena, M.; Cawoy, H.; De Pauw, E. J. Am. Soc. Mass Spectrom. 2013, 24 (8), 1202−1213. (19) Pertot, I.; Puopolo, G.; Hosni, T.; Pedrotti, L.; Jourdan, E.; Ongena, M. FEMS Microbiol. Ecol. 2013, 86 (3), 505−519. (20) Borriss, R.; Chen, X. H.; Rueckert, C.; Blom, J.; Becker, A.; Baumgarth, B.; Fan, B.; Pukall, R.; Schumann, P.; Spröer, C.; Junge, H.; Vater, J.; Pühler, A.; Klenk, H. P. Int. J. Syst. Evol. Microbiol. 2011, 61 (8), 1786−1801. (21) Chen, X. H.; Koumoutsi, A.; Scholz, R.; Schneider, K.; Vater, J.; Süssmuth, R.; Piel, J.; Borriss, R. J. Biotechnol. 2009, 140 (1−2), 27− 37. (22) Chen, X. H.; Koumoutsi, A.; Scholz, R.; Borriss, R. J. Mol. Microbiol. Biotechnol. 2009, 16 (1−2), 14−24. (23) Hao, K.; He, P.; Blom, J.; Rueckert, C.; Mao, Z.; Wu, Y.; He, Y.; Borriss, R. J. Bacteriol. 2012, 194 (12), 3264−3265. (24) He, P.; Hao, K.; Blom, J.; Rückert, C.; Vater, J.; Mao, Z.; Wu, Y.; Hou, M.; He, P.; He, Y.; Borriss, R. J. Biotechnol. 2013, 164 (2), 281− 291. (25) Kraas, F. I.; Helmetag, V.; Wittmann, M.; Strieker, M.; Marahiel, M. A. Chem. Biol. 2010, 17 (8), 872−880. (26) Ongena, M.; Jacques, P. Trends Microbiol. 2008, 16 (3), 115− 125. (27) Arguelles-Arias, A.; Ongena, M.; Halimi, B.; Lara, Y.; Brans, A.; Joris, B.; Fickers, P. Microb. Cell Fact. 2009, 8, 63. (28) Leenders, F.; Stein, T. H.; Kablitz, B.; Franke, P.; Vater, J. Rapid Commun. Mass Spectrom. 1999, 13 (10), 943−949. (29) Nihorimbere, V.; Ongena, M.; Cawoy, H.; Henry, G.; Brostaux, Y.; Kakana, P.; Thonart, P. Commun. Agric. Appl. Biol. Sci. 2009, 74 (3), 645−650. (30) Duitman, E. H.; Wyczawski, D.; Boven, L. G.; Venema, G.; Kuipers, O. P.; Hamoen, L. W. Appl. Environ. Microbiol. 2007, 73 (11), 3490−3496. (31) Espinosa-Urgel, M.; Kolter, R.; Ramos, J. L. Microbiology 2002, 148, 1−3. (32) Fan, B.; Carvalhais, L. C.; Becker, A.; Fedoseyenko, D.; Von Wirén, N.; Borriss, R. BMC Microbiol. 2012, 12, 116. (33) Fan, B.; Chen, X. H.; Budiharjo, A.; Bleiss, W.; Vater, J.; Borriss, R. J. Biotechnol. 2011, 151 (4), 303−311. (34) Koumoutsi, A.; Chen, X. H.; Henne, A.; Liesegang, H.; Hitzeroth, G.; Franke, P.; Vater, J.; Borriss, R. J. Bacteriol. 2004, 186 (4), 1084−1096. H

dx.doi.org/10.1021/ac500290s | Anal. Chem. XXXX, XXX, XXX−XXX