•Most Azospirillum plant growth-promoting rhizobacteria (PGPR) benefit plant growth through source effects related to free nitrogen fixation and/or phytohormone production, but little is known about their potential effects on plant physiology. These effects were assessed by comparing the early impacts of three Azospirillum inoculant strains on secondary metabolite profiles of two different maize (Zea mays) cultivars.
•After 10 d of growth in nonsterile soil, maize methanolic extracts were analyzed by reverse-phase high-performance liquid chromatography (RP-HPLC) and secondary metabolites identified by liquid chromatography/mass spectrometry (LC/MS) and nuclear magnetic resonance (NMR).
•Seed inoculation resulted in increased shoot biomass (and also root biomass with one strain) of hybrid PR37Y15 but had no stimulatory effect on hybrid DK315. In parallel, Azospirillum inoculation led to major qualitative and quantitative modifications of the contents of secondary metabolites, especially benzoxazinoids, in the maize plants. These modifications depended on the PGPR strain × plant cultivar combination.
•Thus, Azospirillum inoculation resulted in early, strain-dependent modifications in the biosynthetic pathways of benzoxazine derivatives in maize in compatible interactions. This is the first study documenting a PGPR effect on plant secondary metabolite profiles, and suggests the establishment of complex interactions between Azospirillum PGPR and maize.
Plant growth-promoting rhizobacteria (PGPR) colonize roots and engage in associative symbiosis with various host plants (Bashan et al., 2004). This interaction takes place at the root–soil interface (the rhizosphere), where PGPR are stimulated by plant root exudates and exert, in return, positive effects on plant growth (Richardson et al., 2009). For certain PGPR, these positive effects on plant growth are indirect and result in fact from biocontrol mechanisms involving antagonism towards phytopathogens (Raaijmakers et al., 2009) and/or the induction of systemic resistance pathways in the plant (Verhagen et al., 2004). By contrast, many other PGPR act mostly by phytostimulation, that is, direct positive effects on plant growth (Richardson et al., 2009).
In the case of phytostimulatory Azospirillum strains, the main modes of action that have been studied are associative nitrogen fixation (Bashan et al., 2004) and phytohormone production (Dobbelaere et al., 2003). The latter can lead to stimulation of root growth and ramification (Jacoud et al., 1999; El Zemrany et al., 2006), thereby enhancing water and mineral uptake by the plant (Fallik et al., 1994). These modes of action rely on source effects for supply of fixed nitrogen and phytohormone signals, respectively, which suggests that complex recognition interactions (between the two partners) and responses may not be needed for effective functioning of this associative symbiosis. However, the association between plant-beneficial microbes (including PGPR) and their host plant(s) is thought to be ancient (Lambers et al., 2009) and has probably been shaped by coevolutionary processes, which suggests that the microbial partners could have significant effects on host physiology. This may be expected especially in the case of microorganisms whose mode of action suggests phytohormonal effects, as the latter must be fine-tuned to the partners’ needs to achieve successful phytostimulation (Remans et al., 2008; Baudoin et al., 2010). If this hypothesis is true, one consequence is that PGPR strains differing in the degree of root colonization and/or the expression of their plant-beneficial traits would be expected to differ in phytostimulation efficacy, which is the case with Azospirillum (Jain & Patriquin, 1984; Fages & Mulard, 1988).
The objective of this work was to assess whether the associative symbiosis between phytostimulatory PGPR and plants involves complex, partner-dependent changes in plant physiology. The focus was on plant secondary metabolism because of its importance in plant–microbe interactions and in plant ecological adaptation (Manuwoto & Scriber, 1985; Hartman, 2007), and also because inoculation effects on plant primary metabolism were modest (unpublished preliminary results on organic acids (V. Walker et al.)). To this end, maize (Zea mays) plants were inoculated with nitrogen-fixing phytohormone-producing Azospirillum strains and grown in nonsterile soil, and the roots and shoots were analyzed by chromatographic profiling of secondary metabolites. In addition, metabolic plant markers of the Azospirillum–maize interaction were identified by chemical means.
Materials and Methods
The PGPR studied were the maize isolates Azospirillum lipoferum CRT1 from France (Fages & Mulard, 1988) and Azospirillum brasilense CFN-535 and UAP-154 from Mexico (Dobbelaere et al., 2001). All three fix nitrogen and harbor the auxin biosynthesis gene ipdC (coding for the indole-3-pyruvate decarboxylase), but not the 1-aminocyclopropane-1-carboxylate (ACC) deaminase gene acdS (C. Prigent-Combaret, unpublished). PR37Y15 is a semi-late maize hybrid (Pioneer Semences SAS, Aussonne, France) and DK315 a semi-early hybrid (Monsanto SAS/Dekalb, St Louis, MI, USA).
Zea mays L. seeds were surface-sterilized with a Bayrochlor mini solution (Bayrol 3 g l−1; Bayrol, Planegg, Germany) for 15 min, under magnetic stirring, and rinsed 4–6 times with sterile water (1 h in total). They were pre-germinated for 48 h at room temperature on 1% (v/v) water agar, which produced radicles c. 1 cm (DK315) and 1–2 mm (PR37Y15) in length.
Azospirillum was grown in Luria–Bertani medium (Sambrook et al., 1989) containing only 5 g l−1 of NaCl (i.e. LBm; Pothier et al., 2007), or Nfb (N-free broth) medium (Nelson & Knowles, 1978) supplemented with 1/40 (v/v) LBm medium. Liquid pre-cultures were prepared from log-phase colonies by incubation (at 28°C, with shaking at 200 rpm) for 16–18 h until an optical density (OD) of 0.7 (at 600 nm), and cultures for seed inoculation were obtained in the same way after transferring aliquots (CRT1 and CFN-535, 1% v/v; UAP-154, 2% v/v) into fresh broth. After centrifugation (at 8320 g for 10 min), the cells were washed twice in 10 mM magnesium sulfate buffer and the suspensions were adjusted to 3–5 × 108 CFU ml−1 based on OD measurements. Pre-germinated seeds were soaked for 30 min in a cell suspension (giving 1–4 × 107 CFU per seed in inoculation treatments) or in sterile water (noninoculated control).
Glasshouse experiment and plant extracts
Seedlings were transferred to 2-dm3 jars containing nonsterile, sieved (Ø 4 mm, sieve mesh size) loamy soil (16.2% clay, 43.9% silt and 39.9% sand, pH 7.0, in water; 2.1% organic matter; El Zemrany et al., 2006) from the surface horizon of a luvisol at the experimental farm of La Côte St André near Lyon (France). Each jar contained four seeds, and eight jars were used for each treatment. The jars were placed in a completely randomized design in a glasshouse at 20°C with a 16 : 8 h day : night photoperiod (relative humidity 45% during the day and 65% at night). Soil was watered to 20% (w/w) water content.
At 10 d after inoculation, plants were dug up and soil adhering to roots was discarded by washing with water. For each plant, roots and shoots were dipped into liquid nitrogen to avoid enzymatic reactions and freeze-dried for 72 h (−54°C), and dry weight was determined.
For chromatographic analysis, freeze-dried roots and shoots were each placed in an Eppendorf tube, to which liquid nitrogen was added. Roots and shoots were crushed using a ball mill (TissueLyser II; Qiagen, Courtaboeuf, France). Each plant part was extracted using 2 ml methanol for 10 mg of dry sample. The extraction was performed twice and extracts were pooled and dried using Speed-vac-assisted evaporation (Labconco, Kansas City, MO, USA). Each sample was then re-suspended in methanol to obtain 4 mg dry extract ml−1.
Chromatographic analysis of the extracts was performed with an Agilent 1200 series HPLC (Agilent Technologies, Santa Carla, CA, USA) equipped with a degasser (G132A), a quaternary pump module (G1311A), an automatic sampler (G1329A) and a diode array detector (DAD G1315B). The separation was carried out at room temperature using a Nucleodur Sphinx C18 column (250 × 4.6 mm; 5 μm; Macherey-Nagel, Düren, Germany). For each sample, 20 μl of extract was injected and the column was eluted at 1 ml min−1, with an optimized gradient established using solvents A (acetic acid 4‰ (v/v) in water) and B (acetic acid 4‰ (v/v) in acetonitrile) (Carloerba Reagents, Val de Reuil, France). For root extracts, a step-by-step gradient was used with an increase of proportion of solvent B until it reached 15% during 5 min, followed by an isocratic phase lasting 30 min, with a flux of 1 ml min−1. For shoot extracts, separation was carried out using the same column, with a linear gradient of acetonitrile in water in which the proportion of acetonitrile increased from 0 to 100% in 55 min (for PR37Y15) or from 10 to 100% in 50 min (for DK315). Chromatograms were recorded and processed at 280 nm, which enabled recovery of all metabolites recorded at other wavelengths (although flavonoids were more intense at 254 or 366 nm, and cinnamic acids at 310 nm). The chemstation Agilent software was used for integration and comparison of chromatograms. Each chromatogram was integrated after standardization of integration parameters. Back-ground peaks present on chromatograms were not integrated.
Liquid chromatography/mass spectrometry analysis
Separation of compounds for mass spectrometry analysis of extracts was achieved using an Agilent 1100 series HPLC equipped with a degasser (G1322A), a binary pump module (G1312A), an automatic sampler (G1313A) and a diode array detector (DAD G1314A). The separation was carried out at room temperature using a Nucleodur Sphinx C18 column (250 × 4.6 mm; 5 μm; Macherey-Nagel). HPLC was interfaced with a HP MSD 1100 series (Agilent Technologies, Santa Carla, CA, USA). Each chromatogram was automatically integrated using chemstation software and then reprocessed manually for better standard integration of minor peaks.
Mass spectrometry operating conditions were gas temperature 350°C at a flow rate of 10 l N2 min−1, nebulizer pressure 30 psi, quadripole temperature 30°C, capillary voltage 4000 V and fragmentor 100. Full scan spectra from m/z 100 to 800 in both positive and negative ion modes were recorded. These parameters allowed the use of analytical conditions similar to those used for HPLC-DAD analysis.
Purification and identification of a selected secondary metabolite
For purification of secondary metabolite 8, roots of 200 10-d-old PR37Y15 plantlets grown in perlite medium (Durantin SA, Charly, France) were harvested and pooled. The methanolic extract of the pooled sample was dried and 337 mg of dry extract was collected. The extract was adsorbed on Lichoprep RP18 silica (Lichoprep RP18, 25–40 μm, 460 × 25 mm; Merck, Darmstadt, Germany) in order to perform an MPLC (Medium Pressure Liquid Chromatography). Phyto-constituents were eluted using 2.5 ml min−1 of an optimized gradient prepared with water and acetonitrile (Carloerba Reagents). A step-by-step gradient from 0 to 50% acetonitrile was generated over 14 h in increments of 10%.
The fractions were analyzed by RP-TLC (Reverse Phase-Thin Layer Chromatography) (RP18 F254; Merck) in the following solvent system: 62.5% water, 37.5% acetonitrile and 0.4% acetic acid (Carloerba Reagents). Six homologous fractions, eluted with 30% acetonitrile and identified based on UV absorbance, were dried using Speedvac-assisted evaporation and pooled, giving 1.1 mg of dried extract. This was then submitted to a semi-preparative HPLC (C18 Nucleodur 260 × 10 mm; 5 μm; Macherey-Nagel), giving 0.8 mg of pure compound 8.
Pure compound 8 was dissolved in methanol-d4-D2O (1 : 1) and submitted to 1H and 13C monodimensional and bidimensional homonuclear nuclear magnetic resonance (NMR) 1H-1H COSY (correlation spectroscopy), heteronuclear 1H-13C HSQC (Heteronuclear Single Quantum Coherence) and 1H-13C HMBC (Heteronuclear Multiple Bond Coherence) (Bruker DRX 500, Bruker biospin, Madison, NI, USA) analysis with the solvent used as a reference. NMR spectra were reprocessed using MestReNova 5.3.0 software (Mestrelab Research, Santiago de Compostela, Spain).
Analysis of maize secondary metabolites in different experiments
To assess whether the same inoculation effects were found under different experimental conditions, root secondary metabolites of maize cv PR37Y15 inoculated or not with A. lipoferum CRT1 and grown in two maize soils comparable to the Côte St André soil were assessed, as already described. One inoculation experiment was a field trial (clay loam soil; sampling at 16 d) and the other a glasshouse trial (clay loam soil; sampling at 35 d).
In the main experiment, the effects of treatments were analyzed using ANOVA followed by Tukey’s tests (P <0.05) based on plant growth data and contents of individual secondary metabolites. The retention time of each peak in the chromatograms (at 280 nm) was aligned and its relative intensity recorded in a matrix to perform discriminant principal component analysis (PCA). Relationships between compounds (on the basis of relative peak areas) were examined using the Pearson correlation coefficient (P <0.05). Discriminant PCA was also used to compare secondary metabolic profiles with those obtained under different experimental conditions.
Effect of Azospirillum inoculation on early plant growth
In comparison with the noninoculated control, at 10 d, seed inoculation of maize cv PR37Y15 resulted in higher root biomass with A. brasilense UAP154 and higher shoot biomass (up to 70% higher) with all three PGPR (Fig. 1a). By contrast, inoculation had no effect on the root or shoot biomass of the other maize cv DK315, regardless of the bacteria used (Fig. 1b).
Identification of maize secondary metabolites
Mass spectra recorded in both negative and positive modes allowed us to perform identification of each major secondary metabolite found in maize tissues (Fig. 2, Table 1). For compound 8, additional NMR experiments confirmed that it was 6-methoxbenzoxazolin-2-one (MBOA) (Fig. 3). All compounds identified in one or both cultivars are summarized in Fig. 4.
Table 1. Spectral data for compounds 1–9
sh, spectral shoulder; Hex, hexose; nd, not determined, TFA, trifluoroacetate.
In shoot methanolic extracts of cv PR37Y15, glycoside and aglycone forms of benzoxazine derivatives (benzoxazinones and benzoxazolinones) were found. These were 2,4-dihydroxy-7-methoxy-1,4-benzoxazin-3-one (DIBOA-β-d-glycoside) (compound 1), 2,4-dihydroxy-7-methoxy-1,4-benzoxazin-3-one (DIMBOA) (compound 2), 2,4-dihydroxy-7-methoxy-1,4-benzoxazin-3-one (DIMBOA-β-d-glycoside) (compound 3), 2-hydroxy-7-methoxy-1,4-benzoxazin-3-one (HMBOA-β-d-glycoside) (compound 6), and MBOA (compound 8). The same compounds were also found in root extracts of cv PR37Y15, which also contained 2-(β-d-glucopyranosyloxy)-4,7-dimethoxy-1,4-benzoxazin-3-one (HDMBOA-β-d-glycoside) (compound 7).
In the case of cv DK315, only the aglycone forms DIMBOA (compound 2), 2,4-dihydroxy-6,7-dimethoxy-1,4-benzoxazin-3-one (DIM2BOA) (compound 4), 2-hydroxy-7-methoxy-1,4-benzoxazin-3-one (HMBOA) (compound 5), MBOA (compound 8) and 6,7-dimethoxy-2-benzoxazolinone (M2BOA) (compound 9) were identified. Furthermore, DIM2BOA and HMBOA were found only in root extracts.
Effect of Azospirillum inoculation on secondary metabolic profiles of maize cv PR37Y15
Chromatograms obtained for root methanolic extracts of maize cv PR37Y15 at 10 d showed a total of 45 integrated peaks. Using retention times and peak areas at 280 nm (Fig. 5), a data matrix was built in order to carry out discriminant PCA to compare inoculation treatments (Fig. 6a). The factorial plan defined by axes 1 and 2 explained 68.5% of data variability. Discriminant analysis separated all four treatments from one another. Unexpectedly, there was as much difference between the two A. brasilense strains as between A. brasilense CFN-535 and A. lipoferum CRT1.
Chromatograms for shoot methanolic extracts of maize cv PR37Y15 yielded a total of 28 integrated peaks. Discriminant analysis (73.2% variability along axes 1 and 2) gave three distinct groups of treatments, that is, no inoculation, inoculation with A. brasilense UAP-154, and inoculation with A. lipoferum CRT1 or A. brasilense CFN-535 (Fig. 6b).
Effect of Azospirillum inoculation on secondary metabolic profiles of maize cv DK315
Chromatograms obtained for root methanolic extracts of maize cv DK315 at 10 d showed a total of 56 integrated peaks. Discriminant analyses (70.1% of data variability with axes 1 and 2) allowed the separation of all four treatments, the main difference being between A. brasilense CFN-535 and the three other treatments (Fig. 6c).
Chromatograms for shoot methanolic extracts of maize cv DK315 gave 34 integrated peaks. Discriminant analyses (68.8% of data variability) yielded three distinct groups, which were no inoculation, inoculation with A. brasilense CFN-535, and inoculation with A. lipoferum CRT1 or A. brasilense UAP-154 (Fig. 6d).
Effect of Azospirillum inoculation on individual secondary metabolites of maize
At 10 d, the total amount of benzoxazinoid compounds (based on total absorbance data) did not differ between treatments, regardless of the maize hybrid (data not shown). For the PR37Y15 cultivar, the prevalence of nine PCA-discriminant secondary metabolites, that is, six for roots (Fig. 7a) and five for shoots (Fig. 7b), differed significantly between treatments based on ANOVA and Tukey’s tests. For root extracts, all were benzoxazine derivatives (Fig. 7a). Of these, DIMBOA-Glc, DIMBOA, HDMBOA-Glc and MBOA represented 62–81% of integrated peak areas on chromatograms. In contrast to PGPR strain A. brasilense UAP-154, the PGPR strains A. lipoferum CRT1 and A. brasilense CFN-535 resulted in higher HMBOA-Glc (except for CFN-535), DIMBOA-Glc and HDMBOA-Glc contents and lower DIMBOA (except for CFN-535) and MBOA contents in comparison with the noninoculated control. Shoot extracts did not display DIMBOA and HDMBOA-Glc, but two unknown cinnamic acids were detected (Fig. 7b). Treatment effects on DIMBOA-Glc and MBOA shoot contents were similar to those on roots. In addition, A. lipoferum CRT1 and A. brasilense CFN-535 resulted in lower shoot contents for cinnamic acid 2.
For the DK315 cultivar, 10 PCA-discriminant secondary metabolites, that is, five for roots (Fig. 7c) and five for shoots (Fig. 7d), differed in prevalence according to treatment. Two other root secondary metabolites were identified (HMBOA and DIM2BOA), but they were not discriminant (data not shown). For root extracts, three benzoxazine derivatives differed with treatment, namely M2BOA and two unidentified cyclic hydroxamic acids (Fig. 7c). Except for M2BOA, all compounds responsible for segregation between treatments were minor compounds (i.e. compounds representing < 5% of the total chromatogram). For shoot extracts, two other cyclic hydroxamic acids and several minor compounds, including a flavonol, differed in prevalence according to treatment (Fig. 7d). Inoculation with A. lipoferum CRT1 or A. brasilense UAP-154 gave similar results for several of these compounds.
Inoculation effects on maize secondary metabolites in different experiments
In the absence of inoculation, root secondary metabolites of maize cv PR37Y15 found in the glasshouse at 10 d (Fig. 5) were also recovered (although not in the same amounts) under field conditions at 16 d, that is, for a similar crop phenology (three to four leaves and four leaves, respectively). The same compounds (with cinnamic acids in larger amounts) as well as three additional secondary metabolites were recovered from older maize (seven to eight leaves) in a separate glasshouse experiment.
In the presence of seed inoculant A. lipoferum CRT1, modifications in the root secondary metabolic profile of glasshouse maize at 10 d were also found with field maize of similar phenology (Supporting Information Fig. S1). By contrast, there was no significant difference between control and inoculated maize plants at 35 d under glasshouse conditions. Therefore, inoculation effects of A. lipoferum CRT1 were also found under different experimental conditions, but were not observed with older maize.
Relationships between secondary metabolites
When relative area data were assessed by Pearson correlation analysis, significant positive or negative linear correlations were found between several secondary metabolites taken two by two (Fig. S2). Indeed, negative correlations were obtained between DIMBOA and DIMBOA-Glc, DIMBOA and HDMBOA-Glc, MBOA and M2BOA, MBOA and HDMBOA-Glc, and DIMBOA-Glc and DIBOA-Glc. In addition, a positive correlation was found between DIMBOA and MBOA, and between DIMBOA-Glc and HDMBOA-Glc.
Although the effects of Azospirillum inoculation on plant growth often take several weeks to materialize (Arsac et al., 1990; Jacoud et al., 1998; Ribaudo et al., 2001), changes in enzymatic activities related to primary metabolism (Fallik et al., 1988) and also sometimes changes in root growth, root hair growth and density, and coleoptyle and/or hypocotyl length (Fulchieri et al., 1993; Cassán et al., 2009) may start earlier. In view of this, and considering our preliminary data (unpublished), we hypothesized that changes in plant metabolism that follow Azospirillum inoculation may be detectable at a rather early stage, and we chose to sample maize 10 d after inoculation in the current work. Differences were observed between certain Azospirilllum inoculants in terms of phytostimulation level, as found previously (Fages & Mulard, 1988).
Cambier et al. (2000) showed that detection of certain benzoxazinoid-related compounds depended on maize variety and tissue. In this study, most benzoxazinoids were found in both roots and shoots. Nevertheless, in the PR37Y15 cultivar, we detected HDMBOA-Glc only in roots and DIMBOA only in shoots. A difference was also observed for the DK315 cultivar, as we found DIM2BOA and HMBOA only in roots (Table 1). This type of result is compatible with data from the literature. However, it must be kept in mind that biosynthesis sites of benzoxazinoids in plants and their translocation mechanisms between roots and shoots are essentially unknown, which limits interpretation of the data.
Roots displayed mainly glycosylated derivatives in the PR37Y15 cultivar, but only aglycone forms in the DK315 cultivar. As monocots elaborate secondary metabolites mainly in glycosidic forms, the sugar moiety being then removed by β-glucosidases in the vacuole membrane (Gruhnert et al., 1994), this difference between the two cultivars was probably attributable to a low β-glucosidase activity in the PR37Y15 cultivar. Nevertheless, this difference could also be the consequence of enhanced activity of the BX8 and/or BX9 enzymes (leading to the transformation of DIMBOA into DIMBOA-Glc) in PR37Y15 (Morant et al., 2008).
This work showed that the associative symbiosis between Azospirillum PGPR and maize had a significant impact on plant secondary metabolism. Indeed, secondary metabolite profiling proved to be effective in showing inoculation effects on plant physiology by day 10, even though the experiment was carried out in nonsterile soil, in which roots were also exposed to a large, diversified microbial community (Sanguin et al., 2006). These effects were not restricted to the particular conditions of the glasshouse experiment, as comparable effects were also recorded in the field at similar plant phenology for PR37Y15 exposed to the inoculant A. lipoferum CRT1 (Fig. S1). The addition of phytohormones such as IAA or -3-butyric acid (IBA) (Rout, 2006; Jeong et al., 2007) or changes in nitrogen availability (Baricevic et al., 1999; Cañas et al., 2009) may modify total plant contents in certain categories of secondary metabolites (e.g. total phenolics and alkaloids), but whether they would also lead to different profiles in individual secondary metabolites remains to be determined. Here, inoculation of maize seedlings changed the contents of several individual benzoxazinoids important for plant interactions. For instance, the cyclic hydroxamic acids DIMBOA and MBOA inhibit germination of wild oat (Avena fatua) (Pérez, 1990), and MBOA that of cress (Lepidium sativum) (Noguchi & Macías, 2006). In wheat (Triticum aestivum), DIMBOA and DIMBOA-Glc act as repellents to aphids such as Sitobion avenae (Leszczynski & Dixon, 1990; Nicol et al., 1992), and they inhibit the take-all fungus Gaeumannomyces graminis (Wilkes et al., 1999). Antibacterial properties have been reported for aglycone forms of hydroxamic acids such as DIMBOA and MBOA, which can inhibit Agrobacterium tumefaciens (Sahi et al., 1990) and Erwinia/Pectobacterium spp. (Corcuera et al., 1978). This can be ecologically relevant, because secondary metabolites may be exudated by roots and may be exudated in larger amounts in inoculated plants (Volpin et al., 1996; Raja et al., 2006). In addition to their role in plant defense, some benzoxazinoids can also interfere with plant hormone functioning. MBOA can impair binding of auxin to membrane receptors (Venis & Watson, 1978) and inhibit root growth at up to 0.1 mM (Hasegawa et al., 1992), while DIMBOA and auxin may have a synergistic effect on coleoptyle elongation (Park et al., 2001). Therefore, the complex impact of Azospirillum on maize benzoxazinoids might have, in turn, multiple effects on plant development, plant health, and/or rhizobacterial community composition.
The impact of Azospirillum inoculants on maize secondary metabolism was strain-dependent, which suggests that the Azospirillum inoculants were recognized by the plant host and recognition might be strain specific. In cv PR37Y15, for instance, PCA discriminated among the three strains, but results suggested that A. lipoferum CRT1 and A. brasilense CFN-535 favored the transformations of DIMBOA to DIMBOA-Glc and DIMBOA or HDMBOA to HDMBOA-Glc, at the expense of the formation of MBOA from DIMBOA. By contrast, A. brasilense UAP-154 inoculation seemed to enhance the transformation of glycoside derivatives into aglycone forms (increasing DIMBOA content) and to inhibit BX8/BX9 enzymes (decreasing contents of HDMBOA-Glc and DIMBOA-Glc). Interestingly, this coincided with the growth promotion pattern at 10 d, as A. brasilense UAP-154 inoculation was the only strain that also increased root biomass in that cultivar. The fact that some of the effects on maize PR37Y15 were similar for A. lipoferum CRT1 and A. brasilense CFN-535 but different for A. brasilense UAP-154 was unexpected, and suggests that strain differences were more important than differences between species (A. lipoferum vs A. brasilense) and geographic origins (France for CRT1 vs Mexico for CFN-535 and UAP-154).
Interestingly, the strain-dependent impact of Azospirillum inoculants on maize secondary metabolism was not the same on both cultivars. Azospirillum inoculants had no phytostimulation effect on cultivar DK315, although the latter displayed inoculant-dependent variation of secondary metabolite profiles. This suggests that these bacteria can have strain-dependent effects on the secondary metabolism of maize without influencing the physiological functions that are related to primary metabolism and involved directly in plant development and growth.
In conclusion, we have developed a novel approach to characterize the impact of PGPR bacteria on the physiology of maize grown in nonsterile soil, based on profiling of plant secondary metabolites. The findings indicate that inoculation resulted in strain-dependent effects on maize secondary metabolism, suggesting the presence of fine-tuned interaction mechanisms. Secondary metabolite profiling resulted in the identification of benzoxazinoid compounds that may serve as early markers of effective PGPR–maize interactions. Furthermore, this approach will also be relevant for exploring the functioning of other types of plant–microbe interaction.
This work was supported in part by the European Union (FW6 STREP project MicroMaize 036314). We are grateful to J. Caballero-Mellado (UNAM, Cuernavaca, México) for providing strains, J. Jansa and colleagues (ETH Zürich, Switzerland) for help with glasshouse trials, and P. Castillon (Arvalis, Baziège, France) for supplying seeds. We thank D. Desbouchages (IFR41, Université Lyon 1) for help with glasshouse experiments. This work made use of the platforms CESN (UMR CNRS 5557 Ecologie microbienne) and Serre (IFR 41) in Université Lyon 1.