Introduction

Burkholderia is a versatile, widely distributed bacterial genus with complex taxonomy whose members occupy diverse ecological niches including soil, water, plants and animals1. Burkholderia species are currently classified into two groups, pathogenic species belonging to the Burkholderia cepacia clade (BCC) and plant-beneficial species in the environmental clade (PBE)2,3. PBE Burkholderia have a complex and diverse host niche and are considered as one of the most potent plant-growth-promoting rhizobacteria (PGPR)1,2 that are ubiquitous in soil and often present in the rhizosphere2,4, with heavy colonization reported for at least 30 plant species4.

PBE Burkholderia are common rhizosphere constituents of major agricultural crops including maize, rice, sugarcane, wheat, tomato and potato5,6,7,8,9,10,11,12,13,14. A valuable feature of some PGPR Burkholderia species is their capacity for biological nitrogen fixation (BNF)2,12, siderophore production, inorganic phosphates solubilization, indole-acetic acid production and phytopathogen inhibition15,16.

Overall, PGPR are recognized for their potential to contribute to sustainable cropping systems but the efficacy of PGPR differs widely, with benefits ranging from none to considerable across crops and environmental conditions. This variability is attributed to environmental factors and difficult-to-predict but specific host interactions17,18. Considering that the molecular mechanisms of the interaction of PGPR Burkholderia with their host plants are largely unknown, knowledge of these interactions will aid in advancing effective microbial inoculants for sustainable crop systems.

Previously, we isolated and characterized a new PGPR Burkholderia species (strain Q208 named after Australian sugarcane variety Q208) from the sugarcane rhizosphere and demonstrated its nitrogen-fixing activity in gnotobiotic culture19. The Burkholderia species heavily colonizes roots and stimulates growth of sugarcane plantlets19. Here, we advance our previous research to apply a dual transcriptomic analysis and investigate the interplay between Burkholderia Q208 and sugarcane plants. We demonstrate that the association consists of sophisticated mechanisms that enable the mutual adaptation characteristic of a beneficial association between plant and bacterium.

Materials and Methods

Plant growth conditions

Plantlets of the Australian commercial sugarcane cultivar Q208 were micro-propagated under sterile conditions on Petri dishes containing half-strength Murashige and Skoog (MS) medium20. Media were supplemented with 1% sucrose, adjusted to pH 5.5 and solidified with 0.3% phytagel (PhytoTechnologies, Kansas USA). Plant cultures were maintained at 28 °C and 16/8 h light/dark cycle with photon flux density 400 μmol m−2 s−1 during the light cycle. Uniform and fully developed plantlets with ≈2–3 cm roots were used for bacterial inoculation experiments.

Bacterial inoculation of plants

Burkholderia Q208 and Bacillus megaterium (used as a control in oxygen saturation and Confocal Raman microscopy measurements) were grown to mid-log phase in nutrient broth medium (Sigma-Aldrich), centrifuged at 4,500 × g for 15 min, washed twice and resuspended in sterile water. Bacterial solutions at an absorbance at 600 nm of ≈1 (approximately 109 cell per ml) were used for inoculations. Roots of plantlets were dipped in bacterial solution for 15 min before transfer into sterile 500-ml containers containing 200 ml of half-strength MS medium supplemented with 2% sucrose, 0.3% phytagel and pH adjusted to 5.7. Plants were incubated in a growth cabinet (28 °C, 16/8 h day/night, 400 μmol m−2 s−1) for 18 days (or as indicated within the text) after inoculation. Non-inoculated plants served as control. Each container contained one plantlet, and each treatment consisted of 10 replicate containers. The containers were randomly rearranged in the growth cabinet daily.

Biofilm formation assay

A 50-μl aliquot of overnight-grown Burkholderia Q208 was inoculated in 5 ml nutrient broth, and incubated at 28 °C with shaking at 200 rpm. After 1 h, 100-μl aliquots were placed in 16 wells of a polystyrene 96-well plate. The first 8 wells received an additional 50 μl sterile phosphate-buffered saline (PBS), while the other 8 wells received an additional 50 μl sterile sugarcane root extract. The root extract was prepared by macerating 2 g of axenic sugarcane root in 5 ml sterile PBS. The plate was incubated overnight at 28 °C with shaking. Assays for biofilm formation were performed using 0.1% crystal violet as previously described21.

Analysis of root morphology and element composition

The morphology of 15 roots (five plants, three primary roots per plant) was analyzed by hand cross-sectioning into ≈15 mm segments at the middle of the root using a vibratome (Leica VT 1200 S, Germany) and observed under confocal laser scanning microscopy. For elemental analysis plants were separated into roots and shoots, rinsed and cleaned three times in 0.5 mM CaCl2 to remove traces of growth medium from surfaces. Tissues were dried at 60 °C for five days, weighed, homogenized and analyzed for elemental content. Nitrogen and carbon were determined by combustion (TruSpec CHN analyzer, LECO, MI). Results represent averages of 10 plants. All growth experiments were repeated three times.

Statistical analyzes

Statistical analyses were performed using Student’s t test (GraphPad Prism4, San Diego) to test for significant differences in dry weight and elemental content of inoculated versus non-inoculated plants.

Oxygen (O2) saturation measurement

Oxygen saturation in the growth medium was determined using a Dissolved O2 meter SG6 (SevenGoTM Mettler Toledo, Schwerzenbach) following the manufacturer’s instructions.

Confocal laser scanning microscopy (CLSM)

Centre sections of primary roots (~15 mm long) were washed, embedded in 3% agarose and sectioned with a vibratome (Leica VT 1200 S, Germany). Roots were coated with agarose before processing to ensure that bacteria on the root surface were trapped in the agarose and not dislodged during cutting. Sections were transferred into curved slides, washed thoroughly with deionized water and analyzed by CLSM.

Transmission electron microscopy (TEM)

Five roots were fixed in 2.5% glutaraldehyde in 0.1 M phosphate buffer pH 6.8 overnight at 4 °C. After washing in 0.1 M phosphate buffer, roots were postfixed in 1% osmium tetroxide, dehydrated through a graded ethanol series, infiltrated with Epon and polymerized for 2 days at 60 °C. Thin sections were cut with a Leica Ultracut UC6 ultramicrotome, picked up on carbon-coated copper grids, stained with uranyl acetate and Reynold’s lead citrate22 and viewed in a JEOL 1010 TEM at 80 kV. Images were captured on an Olympus Soft Imaging Solutions Megaview III digital camera.

Scanning electron microscopy (SEM)

Plant roots fixed in 2.5% glutaraldehyde in 0.1 M phosphate buffer were dehydrated in a graded ethanol series and dried in a Tousimis Autosamdri-815 critical point dryer. Roots were then sputter-coated with platinum and viewed in a JEOL 7001 FEGSEM at 5 kV.

Confocal Raman microscopy

A loop of freshly grown bacterial cells was spread on optically polished CaF2 slides (Edmund Optics Singapore). Raman spectra were acquired using an Alpha 300 Raman/AFM (WITec GmbH, Ulm) equipped with a 100x N.A. 0.9 objective (Nikon)23. A frequency-doubled continuous-wave Nd:YAG laser stabilized at 532 nm was used for excitation. Laser power was kept below 5 mW at the sample. Raman signals were collected with a 100-μm optical fiber with resolution 4 cm−1. Spectral acquisition was performed at integration time 0.2 s, which proved to be sufficient for high-contrast resonance spectra for cytochromes. All images were constructed by collecting spectra on 80 points/line with 80 lines per/image at scan span 30 μm.

Quantification of bacteria by RT-qPCR

Gnotobiotic plantlets inoculated with Burkholderia Q208 (1.25 × 109 CFU/ml) were grown for 1, 5 or 15 days. Roots were washed twice with MgSol (10 mM MgSO4) and separated into non-sterilized or surface-sterilized roots. Surface sterilization was modified after Govindarajan et al.24. Briefly, roots were washed with MgSol and vortexed for 3 min under agitation in 2% sodium hypochlorite. Roots were rinsed five times for 5 min in sterile H2O, then rolled onto Luria-Bertani agar plates to verify surface sterility. Total Burkholderia Q208 dsDNA was extracted using MoBio PowerSoil DNA isolation kits following the manufacturer’s instructions (MoBio Laboratories, Carlsbad CA). Burkholderia 16S rRNA was amplified using Burkholderia-specific primers Cluster11F (5′-GAACAGAGGGTTGCCAAG-3′) and 1392R (5′-ACGGGCGGTGTGTAC-3′). Sugarcane 25S rRNA was amplified using the 25S rRNA primer pair 5′-ATAACCGCATCAGGTCTCCAAG-3′ and 5′-CCTCAGAGCCAATCCTTTTCC-3′ 25. RT-qPCR followed the protocol of Paungfoo-Lonhienne et al.26. After amplification, melting curve analysis was performed to verify uniqueness of the product. Measured Ct values were converted to relative copy-numbers by normalizing the Ct values of bacterial 16S rDNA by the Ct values of sugarcane 25S rDNA. All experiments were performed in three biological replicates, and all measurements in three technical replicates.

RNA extraction and sequencing

Sugarcane plantlets obtained by micropropagation (3 months) were grown in gnotobiotic culture inoculated with Burkholderia Q208 (termed ‘PBK’) for 16 days. Controls were Burkholderia Q208 grown in the absence of plantlet (bacteria only ‘BK’, inoculated into MS medium using sterile 1 mL pipette) and sugarcane plantlets cultivated in the absence of bacteria (plant only ‘P’) (Fig. 1). Each treatment consisted of three biological replicates. All samples were frozen in liquid N2 immediately after collection and stored at −80 °C. Total RNA of samples was extracted using MoBio PowerPlant RNA isolation kits following the manufacturer’s instructions (MoBio Laboratories). Samples were subsequently treated with RTS™ DNase (Applied Biosystems). RNA quantity and quality were verified using a Bioanalyzer (Agilent). Ribosomal RNA of plants and bacteria was removed using a Ribo-Zero Magnetic Kit (plant seeds/root) and a Ribo-Zero rRNA Removal Kit (meta-bacteria) respectively, following the manufacturer’s instructions (Epicentre, Madison WI). Construction of TruSeq RNA libraries and HiSeq2000 sequencing was conducted at the Ramaciotti Centre for Genomics (University of New South Wales, Australia). Libraries were prepared as described in the TruSeq RNAseq v2 Library Preparation guide (Illumina, September 2012) except that rRNA-depleted RNA was input from the “Elute, Prime, Frag” stage of the library preparation. A 12-cycle PCR was used to amplify the adapter ligated library. RNA-Seq libraries were sequenced (independently for each biological replicate) on a 100-bp paired-end lane using the Illumina platform (Illumina HiSeq 2000, San Diego).

Figure 1
figure 1

Schematic presentation of the “dual transcriptome analysis”.

Sugarcane plantlets (2) were inoculated with Burkholderia Q208 culture (1) and plants were grown for 16 days (3). The control, non-inoculated plants (4) were also grown for 16 days. Subsequently, bacterial culture (1), roots of the control plants (4; boxed area), and roots of the inoculated plants (3, boxed area) were subjected to transcriptomic analyses. In the case of (3) the analyses included the transcriptomes of both bacteria and plant roots (“dual transcriptomics”).

Full size image

Pre-processing of Illumina reads

All primary sequencing data were deposited in the European Nucleotide Archive (ENA) under accession number ERS568745. All reads were evaluated for quality using FASTQC (http://www.bioinformatics.babraham.ac.uk/projects/fastqc/).

Mapping of Illumina reads

All RNA reads of BK and PBK libraries were aligned against the Burkholderia Q208 genome (Accession number SRP029967) using the Subread V1.4.3 subread-align command. Read pairs aligned to the Burkholderia Q208 genome were then counted relative to their unpaired match position against Burkholderia Q208 genome features. Reads mapping to two features at the same position were ignored. Sugarcane reads in PBK libraries not mapping to Burkholderia Q208 or sugarcane libraries were independently mapped onto the sorghum genome using Bowtie2 version 2.06. Reads which mapped onto sorghum were counted using Subread featureCounts. The Sorghum bicolor genome used as reference was Release 21 from ftp.ensemblgenomes.org.

Quantification of gene expression

All reads from the pure BK vs PBK libraries, and from PBK vs P libraries, were feature-mapped onto the Burkholderia Q208 and S. bicolor genomes respectively using Subread featureCount, and the counts were input into DESeq for differential expression analysis. DESeq27 is available as a Bioconductor package from the Bioconductor repository28 and from http://www-huber.embl.de/users/anders/DESeq. Transcripts mapping to the Burkholderia Q208 genome, and those found to be significantly down- or up-regulated by more than twofold (dispersion-corrected mean of biological replicates) with an adjusted P value (P < 0.01) were analyzed further. Volcano plots were made in R (www.r-project.org). Functional and Gene Ontology annotations were mapped to the most differentially expressed genes using Blast2GO29 available at https://www.blast2go.com.

Results

RNA-seq analyses of Burkholderia Q208-sugarcane association

In our previous research (Paungfoo-Lonhienne et al.19), we demonstrated that Burkholderia Q208 is a PGPR of Australian sugarcane that abundantly colonizes the roots of host plants, which suggested that the beneficial association occurs through biofilm formation. Thus here, to understand the features underlying the association, we investigated the molecular interplay between the Burkholderia Q208 biofilm and the sugarcane roots. To simultaneously capture the molecular mechanisms involved in the establishment of the beneficial association between Burkholderia Q208 and sugarcane, we carried out a “dual transcriptomic analysis” to assess the changes in gene expression profiles in both roots and root-associated bacteria (Fig. 1 and Supplementary Figure S1.).

Approximately 6 million read pairs were sequenced from each biological replicate of axenically cultured Burkholderia Q208 (‘BK’), and 30 million read pairs from each biological replicate of axenic sugarcane roots (‘P’) and Burkholderia Q208 inoculated roots (‘PBK’). Reads were generally of high quality (FASTQC quality scores >28) and there were no concerns regarding the utility of any library. A screen against typical sequencing xenocontaminant DNA revealed no significant contamination for any library.

A full genome sequence of sugarcane is not available, so RNA-seq reads of sugarcane were compared to chromosomal and expressed sequence tags of Sorghum bicolor, sugarcane’s closest relative for which the entire genome is publicly available. It has been reported that sorghum and sugarcane genomes are mostly collinear and share 95.2% of sequence identity30. Of ≈215 million sugarcane reads (‘P’), approximately 65 million mapped to annotated sorghum features. Of ≈200 million mixed sugarcane + Burkholderia RNA reads (‘PBK’) after Burkholderia Q208 genomic subtraction, approximately 63 million mapped to sorghum (Table 1).

Table 1 Mapping results of RNA-Seq reads.
Full size table

Volcano scatter plots (Fig. 2a,b) summarize the fold-change and statistical significance distributions for transcripts of sugarcane and Burkholderia Q208. The plots confirm that the RNA libraries were of high quality. We noticed that many up- or down-regulated genes localize adjacent to each other in the genome of Burkholderia Q208. Gene Ontology (Biological Process) annotations show that for both organisms, over half of the most differentially induced genes are involved in metabolic processes: 57% of the bacterial genes (Fig. 2c) and 56% of the plant genes (Fig. 2d) have Level 3 annotation29 as Single-organism, Primary, Organic substance, Cellular or Nitrogen metabolic processes. By contrast, genes annotated as stress- or defense-related represent a minor proportion of induced genes with 1.4% and 3.5% for Burkholderia and sugarcane, respectively.

Figure 2
figure 2

Results of the “dual transcriptome analyses”.

(a) Volcano scatter plot of fold change (log2) Vs DESeq adjusted P-values (−log10) of Burkholderia vs Burkholderia colonized sugarcane roots (a) or of sugarcane vs sugarcane colonized by Burkholderia (b). Red circles represent adjusted P values < 0.01 and >2 fold change. Level 3 Gene Ontology (GO) terms distribution of 1,628 unigenes induced in Burkholderia colonized on sugarcane roots (c) or of 2,606 unigenes induced in sugarcane roots colonized by Burkholderia (d).

Full size image

Genes involved in formation of the biofilm are upregulated in Burkholderia

A common feature of biofilms is a developed bacterial layer31, which was observed when Burkholderia Q208 cells are grown in the presence of roots (Fig. 3a,2,3), but not when grown without plants (Fig. 3a,1). Scanning electron microscopy (SEM) analyses established that most of the bacteria are associated with the root surface rather than growing endophytically (Fig. 3a,2). This observation was confirmed by qRT-PCR (Fig. 3a,4), when we determined the abundance of Burkholderia Q208 16S rRNA relative to 25S rRNA of sugarcane roots grown with or without microbes. The presence of Burkholderia Q208 was considerably higher (5.3 fold (5 dpi) to 43.3 fold (13 dpi) (P < 0.001)) in the rhizosphere than in roots, implying that the association is primarily rhizospheric.

Figure 3
figure 3

Association of Burkholderia Q208 with sugarcane roots.

(a) Electron microscopy of the sugarcane roots inoculated with Burkholderia Q208 and the abundance of Burkholderia on the root surface, compared to the endophytic bacteria. a1 – TEM of the bacterial culture used for the inoculation; a2 – SEM of the sugarcane roots with a box indicating the magnified area shown in a3. The plants were grown in gnotobiotic conditions for 5 days. Arrows indicate bacteria, white arrowheads-plant cell wall, black arrowheads-blanket-like material covering Burkholderia cells. a4 –quantitative real-time PCR showing relative abundance of Burkholderia outside of the roots (rhizospheric) and inside of the roots (endophytic). Black bars display the results of the qRT PCR carried out on the roots with bacteria, and white bars the roots only, with root-associated bacteria stripped off. Abundance of 16S rRNA genes of Burkholderia was quantified relatively to 25S rRNA genes of sugarcane. Data represent averages of three independent replicates. *** indicates significant difference at P < 0.001 (Student’s t test). (b) Biofilm formation assay for the conditions when Burkholderia were grown in the nutrient medium without (1) or with (2) addition of sugarcane root extract. Blue color indicates formation of the biofilm, when crystal violet staining is applied to the media (top panel); the absorbance quantification of 7 biological replicates are shown (bottom panel). The bars represent SD. ** indicates significant difference at P < 0.01 (Student’s t test).

Full size image
Figure 4
figure 4

Burkholderia Q208 induces the plant root thickening and expansion of the aerenchyma.

Sugarcane roots of the plants not inoculated with Burkholderia (a), and those of the plants inoculated with Burkholderia Q208 (b). The right panels in (a) and (b) show CLSM images of the sections representing the average thicknesses of the main roots.

Full size image

We also tested the influence of sugarcane root extract on the ability of Burkholderia Q208 to form a biofilm, using the crystal violet assay21. Crystal violet is a basic dye which binds to negatively charged molecules, including cell surface and exopolysaccharides (EPS). Bacteria grown in nutrient medium with addition of sugarcane root extract exhibited dense staining with crystal violet (Fig. 3b), implying that sugarcane extract induces formation of the biofilm.

Bacterial biofilms consist of a number of components, among which are EPS, DNA and proteins32. The transcriptomic data reveal that the gene clusters bce-I and bce-II (Burkholderia cepacia complex exopolysaccharide-I and –II) directing EPS “cepacian” biosynthesis in Burkholderia species33 are over-expressed (on average 9-fold, Supplementary Table S1) in Burkholderia Q208 associated with sugarcane (Plant-associated Burkholderia (PBK) gene expression compared to Burkholderia alone (BK), Fig. 1). Similar to Burkholderia xenovorans33, the genes of clusters bce-I and bce-II of Burkholderia Q208 are adjacent on the genome, forming a cluster of 18 genes involved in nucleotide sugar biosynthesis, glycosyl and acyltransferase activities, polymerization and export of saccharides.

Biofilm formation is generally controlled by quorum sensing (QS), a form of chemical communication that coordinates gene expression when the concentration of bacterial cells reaches a certain threshold34. Nitrogen-fixing plant-associated Burkholderia display a highly conserved quorum sensing system, designated BraI/R, that is encoded by a cluster of three genes (braI, rsaL and braR), the products of which generate and respond to N-dodecanoyl-3-oxo-homoserine lactone35; this cluster is also present in the genome of Burkholderia Q208 (Supplementary Table S1). In the biofilm Burkholderia braI is up-regulated (2-fold) whereas, curiously, rsaL and braR are down-regulated (to 0.35- and 0.45-fold of reference levels, respectively). However, the absolute counts of raw reads of rsaL (16,000) and braR (15,500) are greater than the mean (700) read number over all expressed genes, suggesting that although these genes are down-regulated, BraI/R quorum sensing by Burkholderia Q208 remains active in the sugarcane rhizosphere.

Lipo-polysaccharides and flagella biosynthesis genes in Burkholderia Q208 are down-regulated

When Burkholderia Q208 is associated with sugarcane roots, two bacterial gene clusters involved in lipo-polysaccharides (LPS) production are down-regulated (Supplementary Table S2). Cluster 1 is composed of 17 genes including the O-antigen polymerase (0.21-fold; in other words a 5 fold decrease) involved in O-antigen assembly, and a member of the GtrA protein family (0.18-fold). In Shigella flexneri, GtrA encoded by bacteriophage is involved in O-antigen modification36. The second cluster contains 12 genes, including UTP-glucose-1-phosphate uridyltransferase (0.32-fold) involved in biosynthesis of bacterial polysaccharides (LPS and EPS) and often associated with pathogenesis: in Streptococcus pneumoniae it is considered the main virulence factor37. Independently of these two clusters, the lipid A-core-O-antigen ligase involved in ligation of O-antigen onto the core region of the lipid A-core block is also down-regulated (0.37-fold). These results suggest that LPS production by Burkholderia Q208 is reduced in the biofilm. Accordingly, the immune response to LPS is apparently not induced in roots of sugarcane; nitric oxide (NO) synthase NOA1 is down-regulated (0.56-fold, Supplementary Table S3), with NO production being a hallmark of the innate immune system induced by PAMP LPS in plants38.

Flagella serve as organs of locomotion facilitating bacterial chemotaxis. In biofilms, bacteria are non-motile and this is achieved by functional regulation or inhibition of flagellar gene transcription39. Associated with sugarcane roots, the Burkholderia Q208 gene cluster containing 50 genes related to chemotaxis and flagellar synthesis is down-regulated (mean 0.3-fold, Supplementary Table S4), suggesting that motility is inhibited through loss of flagella. In sugarcane roots, expression of the genes involved in the signalling pathway of plant immune response to flagellin, which includes receptors FLS2 and BAK1 and the mitogen-activated protein kinase (MAPK) cascade40, is not significantly altered (Supplementary Table S3). Thus important immunogenic factors are repressed in Burkholderia Q208 when associated with sugarcane, in line with the establishment of a mutualistic association. Remarkably, Cluster 2 involved in LPS biosynthesis (Supplementary Table S2) also contains the flagellar transcription activator flhD which is strongly down-regulated (0.07-fold), indicating that regulation of flagella and of LPS production are linked, could be a sign of adaptation to the mutualistic association that prevents activation of the immune defense of sugarcane.

Upregulation of cytochromes in Burkholderia Q208

Burkholderia species are obligate aerobes and lack fermentation pathways, but they can live in microaerobic conditions. Analysis of the Burkholderia Q208 genome shows that its respiratory chain terminates in two branches, with bo- and bd-type cytochromes as terminal oxidases41. Burkholderia Q208 genes encoding cytochrome bd occur in duplicate, similar to that of Escherichia coli, and encode cytochrome bd-I (cydA and cydB) and bd-II (cbdA and cbdB) complexes42. These genes are highly induced in the sugarcane rhizosphere (Supplementary Table S5), cydA 34-fold, cydB 27-fold, cbdA 112-fold, cbdB 49-fold. CydX, a protein essential for assembly and stability of the cytochrome complex43, is upregulated 21-fold. Expression of cytochrome bo is, however, unchanged. A high level of induction of cytochrome bd was confirmed by confocal Raman microscopy (Supplementary Figure S2). Cytochromes produce strong, representative Raman spectra at 532 nm excitation due to resonance of the porphyrin ring44. In the sugarcane rhizosphere, Burkholderia cells present strong bands at 752, 1130, 1315 and 1585 cm−1, a signature of Geobacter sulfurreducens renowned for its high electron-transfer activity23. Normalised to CH stretching (2938 cm−1), the bands corresponding to cytochromes are more intense in Burkholderia Q208 than in Bacillus megaterium cells used as control (Supplementary Figure S2).

Upregulation of energy pathways in Burkholderia Q208

The high level of cytochrome bd production by Burkholderia Q208 suggests a high rate of energy generation, as these cytochromes are involved in oxidative phosphorylation in the electron transport chain. Transcriptomic data of Burkholderia Q208 were analyzed to explore the nature of energy pathways operating in the rhizosphere. Glycolysis converts C6 monosaccharides to pyruvate, producing ATP and NADH in both aerobic and anaerobic conditions. Full commitment to glycolysis by Burkholderia Q208 is indicated by over-expression of phosphoenolpyruvate synthase (pps, 204-fold) and fructose phosphokinase (fpk, 159-fold: Supplementary Table S5) representing the key regulatory point at which glycolysis enters irreversibility. Pyruvate dehydrogenase, which catalyses the rate-limiting step of the complex that produces acetyl-CoA from pyruvate, is over-expressed threefold. Citrate synthase is over-expressed fourfold, suggesting increased activity of the Krebs cycle, hence production of GTP and NADH. In aerobic conditions, NADH produced by the Krebs cycle is oxidized by cytochromes of the electron transport chain, generating a trans-membrane proton gradient. As the bd-type cytochrome is highly over-expressed, this should increase the trans-membrane proton gradient. Consistent with this picture, the F0/F1 ATP synthase complex responsible for converting the trans-membrane proton gradient into ATP is highly induced (9 genes, average over-expression 126-fold: Supplementary Table S5).

Activation of additional energy pathways in Burkholderia Q208

Two anaerobic pathways are induced: the oxalate catabolism and the arginine-deiminase pathways. The anaerobic oxalate catabolism pathway has been well-studied in Oxalobacter formigenes45. Oxalate in O. formigenes is imported into the cell and activated to oxalyl-CoA by ligation with coenzyme A transferred from formyl-CoA and catalyzed by formyl-CoA transferase (Frc). Oxalyl-CoA is then decarboxylated by oxalate decarboxylase (Oxc) to CO2 and formyl-CoA. After transfer of the CoA moiety onto oxalate, formate is exported from the cell, coupled with the import of oxalate through the oxalate/formate antiporter (OxlT), generating a proton-motive force used by F0/F1 ATP synthase to produce energy45. In the Burkholderia Q208 genome, the oxlT gene is physically separated from frc and oxc, and appears to belong to a cluster of five genes including the three following genes: an ortholog of the E. coli acetyl-CoA:oxalate-CoA transferase yfdE, a hypothetical protein with 3D similarity to a proton symporter, and the PHB depolymerase. This cluster is highly overexpressed in Burkholderia Q208 associated with sugarcane (mean 328-fold: Supplementary Table S5). However, we observed almost no reads mapping to the oxlT transcript, implying that oxalate is not imported in a conventional way. The NAD-dependent formate dehydrogenase complex that catalyzes formate oxidation in aerobic conditions is down-regulated (Supplementary Table S5), suggesting that oxalate catabolism occurs under microaerobic conditions. The other upregulated pathway (anaerobic arginine-deiminase pathway) involves catabolism of arginine to ornithine with production of ATP. Similar to Rhizobium etli46, the three genes involved in the pathway are clustered (arcABC): arginine deiminase (arcA), ornithine carbamoyltransferase (arcB) and carbamate kinase (arcC). Export of residual ornithine from the cell is coupled with import of arginine through the arginine/ornithine symporter (arcD). All four genes are highly over-expressed (mean 320-fold: Supplementary Table S5), suggesting that Burkholderia Q208 uses this pathway for ATP generation under microaerobic conditions.

Genes for PHB accumulation and C4-dicarboxylate transport are upregulated in Burkholderia Q208

Burkholderia in the biofilm accumulate the carbon storage polymer poly-β-hydroxybutyrate (PHB) as evidenced under TEM as electron-transparent droplets in the cytoplasm (Supplementary Figure S3). PHB production commences from acetoacetyl-CoA and involves reduction of acetoacetyl-CoA to hydroxybutyryl-CoA by PhaB or PhbB enzymes that cause oxidation of one molecule of NADH, followed by polymerization by PhaC or PhbC. PHB mobilization occurs in three steps: depolymerization by a PHB depolymerase to hydroxybutyrate, oxidation of hydroxybutyrate by 3-hydroxybutyrate dehydrogenase yielding acetoacetate, and transfer of CoA to acetoacetate by acetoacetate-succinyl-CoA transferase. The genes involved in all PHB cycling steps47 are highly upregulated in Burkholderia Q208, most strongly the orthologs of phbB (720-fold) and phbC (236-fold) (Supplementary Table S5).

dctA encodes the structural component necessary for C4-dicarboxylate transport, while dctB and dctC encode positive regulatory elements48. Our data show that when Burkholderia Q208 are associated with sugarcane roots, the genes for C4-dicarboxylate transport dctA, dctB and dctD are over-expressed (5-, 1.6- and 1.4-fold respectively, Supplementary Table S6).

Morphological changes of plant roots in association with Burkholderia Q208 biofilm

Considerable structural changes are observed in roots when the association with Burkholderia is established, with increased root thickness brought about by additional layers of cortex cells and larger aerenchyma (Fig. 4). Aerenchyma facilitates gas flow through plant tissues49 and is an important adaptation to hypoxic conditions in the rhizosphere. Confirming this notion, the concentration of oxygen in the growth medium of sugarcane inoculated with Burkholderia was significantly lower than in the control plants (P < 0.001 at top and middle, and <0.0001 at the bottom, of the growth vessels) or when inoculated with Bacillus megaterium (P < 0.05 at the top-middle, <0.001 at the bottom) (Supplementary Figure S4).

Hypoxia-related genes are upregulated in sugarcane

Plants subjected to hypoxia increased ethylene production, which is responsible for formation of aerenchyma in the root cortex via the action of aminocyclopropane-1-carboxylate synthase (ACC synthase) and ACC oxygenase (ACC oxidase 1)50. Analysis of the sugarcane transcriptome suggests a similar mechanism: in sugarcane inoculated with Burkholderia Q208, ACC synthase ACS8 and ACC oxidase 1 are over-expressed (4.6 and 10.8-fold respectively: Supplementary Table S7). Hypoxia increases expression of the haemoglobin ortholog AHB1/GLB1 in sugarcane (16-fold). Similarly, in Arabidopsis, hypoxia-induced expression of AHB1/GLB1 occurs predominantly in roots, and its function is linked to its ability to bind oxygen51. Adaptation to hypoxia is further confirmed by increased expression of the APETALA2/ethylene response factors RAP2 (3-fold) and especially of prolyl 4-hydroxylase alpha (53-fold). In Arabidopsis, RAP2.2 is involved in post-translational modification of hypoxia-induced proteins52 and prolyl 4-hydroxylase alpha regulates responses to oxygen deficiency53.

Availability of photosynthetic carbon is required for establishment of effective plant-Burkholderia association

Sucrose is the main carbohydrate derived from photosynthesis, and its degradation to hexoses in legume nodules is important for delivering photosynthates to symbionts. Sucrose synthases fulfilling this function are also essential for BNF in Pisum sativum (RUG4)54 and Phaseolus vulgaris (PvSSn)55. In sugarcane inoculated with Burkholderia Q208, sucrose synthases 1 and 2, orthologs of PvSSn and RUG4, were over-expressed (1.5- and 3-fold respectively, Supplementary Table S8), suggesting that hexose production is increased in sugarcane roots.

To evaluate the dependence of the Burkholderia Q208-sugarcane association on plant carbon supply, we truncated sugarcane plantlets before inoculation with the bacterium to reduce the ability of plantlets to provide photosynthates. In truncated plantlets, the effect of Burkholderia Q208 on plant biomass was strongly reduced (Fig. 5a), compared to non-truncated plantlets, suggesting that availability of photosynthates is a main driver of the successful relationship. Nevertheless, the role of plant stress in adversely affecting the bacterium-sugarcane association should not be ruled out.

Figure 5
figure 5

Effect of Burkholderia Q208 on growth of sugarcane plants.

(a) Dry weight of the sugarcane plants, with shoots either not (left panel) or truncated (right panel) at the time of inoculation. Data represent the average of 10 individual plantlets. Asterisks above bars indicate significant differences from the control: *P < 0.05; ***P < 0.001; ****P < 0.0001 (Student’s t test). (b) Inoculation with Burkholderia Q208 significantly increases nitrogen concentration in the sugarcane plants but does not affect carbon concentration.

Full size image

Discussion

Our morphological analyses reveal that the association between Burkholderia Q208 and sugarcane roots is supported by a number of transformations in both partners. We detected significant changes in gene expression in sugarcane and Burkholderia Q208 by simultaneously analyzing transcriptomes of both partners which match the observed morphological changes. The bacteria form a biofilm on the root surface, lose motility and immunogenicity, activate hexose intake, and generate energy in both aerobic and microaerobic conditions. Plants in turn do not activate their immune response, expand root aerenchyma to conceivably counteract oxygen limitation induced by the bacterial biofilm, and supply the bacteria with photosynthates. In its entirety, the partnership has the hallmarks of a mutually beneficial association that resembles the symbiotic association between rhizobia and legumes at molecular and morphological and molecular levels, resulting in the establishment of a stable bacterial population, and stimulation of plant growth (Fig. 6b).

Figure 6
figure 6

Schematic presentations of Burkholderia-sugarcane interactions revealed via “dual transcriptomics” analysis.

(a) Energy pathways of Burkholderia Q208 induced upon association with sugarcane. Arrows and frames indicate energy pathways in aerobic (red arrows) and microaerobic (blue) conditions. All the designated pathways are activated in the biofilm. Stars highlight the energy carriers generated by each pathway and the up-regulation of the particular genes is indicated in the boxes. (b) A proposed model for the crosstalk between plant roots and Burkholderia. The biofilm is formed by Burkholderia cells on the surface of sugarcane roots. Microaerobic conditions are created by the bacteria through activation of the aerobic energy generation (cytochrome bd and F0/F1 ATP synthase are over-expressed); this microaeroby presumably supports BNF. The plant provides oxygen to the roots and to the biofilm through enlarged aerenchyma in order to support its own metabolism and that of the bacterial cells. The bacterial energy demand is supported by the plant’s photosynthates, such as malate, succinate and fumarate.

Full size image

The apparent first step in the establishment of the association is the formation of a robust bacterial biofilm on roots. Transcriptomic analyses reveal that the flagellar biosynthesis genes in Burkholderia Q208 are down-regulated, indicative of a transition of Burkholderia from a motile state to non-motile multicellular aggregates, typical for bacterial cells within biofilms39.

A PGPR should not be recognized as virulent by the plant and accordingly several key pathways are down-regulated. Two main factors of virulence of pathogenic BCC-clade Burkholderia are lipopolysaccharides (LPS) and flagella56,57. More generally, LPS and flagella are pathogen-associated molecular patterns (PAMPs), responsible for activating the innate immune system of plants38,40. LPS is the major component of the outer membrane of Gram-negative bacteria58, and are formed by three chemically linked components: a hydrophobic glycolipid integral to the membrane (lipid A), a core polysaccharide located at the surface, and a polysaccharide consisting of repeating units of sugars extending from the membrane surface into the environment, known as the O-antigen, as the immuno-dominant portion of the molecule59. O-antigen plays also a crucial role in root colonization, as the O-antigen mutants of Pseudomonas fluorescens lose this ability60,61. Among the strongly downregulated genes implicated in LPS biosynthesis in Burkholderia Q208 are those involved in O-antigen assembly (Supplementary Table S2), suggesting that there is no apparent necessity to produce them once the biofilm is established, due to the potential immune response by the plant. Accordingly, genes involved in plant immune response are not activated (Supplementary Table S3). A similar situation is observed during formation of the legume plant-rhizobium interaction when LPS of rhizobia undergo structural modifications, and one of the explanations for this event is minimization of the plant immune response to bacterial invasion62,63,64.

In response to biofilm formation by Burkholderia Q208, roots expand aerenchyma as a likely response to low oxygen concentrations due to the combined effects of the physical impediment of the biofilm to oxygen diffusion into roots and high oxygen consumption by bacteria. In legume nodules, oxygen is absorbed with high affinity by leghaemoglobin to create the microaerobic environment conducive to BNF via the nitrogenase complex65. Effective compartmentalization in root nodules restricts low concentration of free oxygen to nodules and maintains the aerobic environment for roots tissues of the legumes. It may be less straightforward to achieve this situation in sugarcane, where the bacteria occur in a less-compartmentalised (although microaerobic) biofilm thus reducing oxygen concentration in the roots. Increased thickness and volume may allow sugarcane roots to be less affected by the microaerobic conditions at their surface, while the expanded aerenchyma facilitates increased air diffusion into roots. An elevated level of haemoglobin as deduced by the over-expression of the haemoglobin ortholog AHB1/GLB1 in sugarcane roots presumably also contributes to increased oxygen delivery to roots. Together, these adaptations allow roots to remain aerobic while also enabling the conditions essential for BNF.

The association between Burkholderia Q208 and plants requires a significant amount of energy to support bacterial growth and function, including the energy-demanding process of BNF. Indeed, in the rhizosphere, Burkholderia Q208 activate several pathways that favor energy production under both aerobic and microaerobic conditions (Fig. 6a). As a result, the bacteria experience an energy-rich, oxygen-depleted environment, two features required for BNF (Fig. 6b). Our results suggest that over-expression of cytochrome bd and F0/F1 ATP synthase contributes to both aspects: firstly, it is responsible for increased production energy in aerobic and probably sub-aerobic conditions. Indeed, cytochrome bd has a very high affinity for O2, which implies that it is able to function at low oxygen concentration. Secondly, it is the factor that triggers the depletion of oxygen in Burkholderia cells, creating a microaerobic environment essential for BNF, as shown in Azotobacter vinelandii41 and Klebsiella pneumonia66.

However, by inducing anoxic conditions the Burkholderia Q208 potentially create adverse conditions for aerobic respiration: under low oxygen conditions, aerobic metabolism leads to accumulation of NAD(P)H which inhibits citrate synthase and isocitrate dehydrogenase, thereby impeding the Krebs cycle67. Burkholderia Q208 is an obligate aerobe and therefore lacks fermentation pathways that would restore regular cell redox potential, thus it is crucial for the bacterium to have an alternative system for oxidizing of NADH. The adaptation of Burkholderia consists in storing the surplus of carbon and energy in the form of PHB granules (Supplementary Figure S3). PHB metabolism is tightly linked to the redox state of the cell, and its accumulation in diazotrophs such as Azotobacter and Rhizobium is induced by oxygen limitation67. In Azotobacter, PHB serves as an alternative electron acceptor, relieving inhibition of the Krebs cycle by increasing the cells’ redox potential67.

Besides regulating the redox state of the cell, PHB provides a competitive advantage by allowing storage and mobilization of carbon in response to environmental changes. Upregulation of the genes involved in PHB degradation in Burkholderia Q208 indicates that PHB granules are used. The data suggest that under microaerobic conditions, PHB mobilization serves the oxalate catabolism, as the up-regulated oxalate catabolic gene cluster includes the PHB depolymerase and the acetyl-CoA:oxalate-CoA transferase (Supplementary Table S5) which activates oxalate via transfer of CoA. Oxalotrophy is widespread in Burkholderia of the environmental clade68 which includes Burkholderia Q208, although uncommon elsewhere, and may provide additional selective advantage for establishing the association with sugarcane. Notably, Burkholderia Q208 possesses a further adaptation to microaerobic conditions in the form of the arginine-deiminase pathway which is also up-regulated in the biofilm.

Another important feature of Burkholderia Q208 is the activation of the C4-dicarboxylate transport (Dct) system. In the Rhizobium-legume symbiosis, bacteria rely strongly on carbon sources generated by the host69. Dct involves transport into bacterial cells of macromolecules such as succinate, malate, fumarate, and aspartate, which are metabolized under aerobic or anaerobic conditions70. Importantly, this transport system is essential in rhizobia-legume symbiosis as it is involved in transferring the dicarboxylates produced by the host plant to the bacteroids as a source of carbon and energy71.

Our data show that plant nitrogen acquisition is stimulated in the association, with significantly higher nitrogen concentration in root and shoot tissues and increased plant growth (Fig. 5b). In contrast, concentration of carbon in shoots remains steady and this suggests that carbon is not specifically supplied by bacteria, but rather is the result of photosynthesis. Strongly upregulated dct transport and PHB accumulation systems suggest that bacteria receive in return plentiful and perhaps excessive amounts of plant photosynthates.

In summary, we provide multiple lines of evidence that the mutually beneficial association between Burkholderia Q208 and sugarcane brings about substantial molecular and morphological adaptations in both partners. This study reveals new aspects of the interactions between roots and bacteria occurring in the plant rhizosphere.

Additional Information

How to cite this article: Paungfoo-Lonhienne, C. et al. Crosstalk between sugarcane and a plant-growth promoting Burkholderia species. Sci. Rep. 6, 37389; doi: 10.1038/srep37389 (2016).

Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.