Introduction

Azospirillum are prominent plant growth-promoting rhizobacteria (PGPR) used as inoculants for phytostimulation of several types of crops (mainly cereals) under different climatic conditions, and they may lead to improved crop yields (Charyulu et al. 1985; Okon and Labandera-Gonzalez 1994; Dobbelaere et al. 2001; Pedraza et al. 2009). This environmental biotechnology is also receiving attention as a mean to reduce chemical fertilizer doses without affecting crop yield, and can thus be evaluated as a component of integrated management strategies in agriculture (El Zemrany et al. 2006; Fuentes-Ramirez and Caballero-Mellado 2006; Adesemoye et al. 2009). Several modes of action have been documented in Azospirillum PGPR, noticeably nitrogen fixation (James 2000), nitric oxide production (Creus et al. 2005), and 1-aminocyclopropane-1-decarboxylate deaminase activity (Prigent-Combaret et al. 2008), but production of phytohormones such as auxins is often proposed as the main phytobeneficial mechanism (Dobbelaere et al. 2003).

Interactions of Azospirillum PGPR inoculants with other rhizosphere microorganisms have been considered by studying the ecological impact of inoculation on non-target, resident microorganisms and the effect on plant of mixed inocula involving Azospirillum strains. On one hand, studies on the ecological impact of Azospirillum inoculation did not evidence any positive interaction with indigenous microorganisms (Basaglia et al. 2003; Russo et al. 2005; Herschkovitz et al. 2005a, b; Lerner et al. 2006; Naiman et al. 2009; Baudoin et al. 2009b). On the one hand, the potential use of Azospirillum PGPR strains in mixed inocula is promising (Bashan 1998; Cassan et al. 2009; Singh et al. 2010; Combes-Meynet et al. 2011), but it remains poorly documented. By combining microorganisms with different metabolic capacities (N2 fixation, P mobilization, production of phytohormones, and antimicrobials, etc.), we could expect additive or synergistic effects resulting from the combination of different phytobeneficial capacities, which may be important to enhance performance consistency.

Several studies have focused on Azospirillum dual-inoculation with (1) other Azospirillum strains (Han and New 1998; Bashan et al. 2000), (2) Bacillus (El-Komy 2005), (3) Bradyrhizobium (Steinberg et al. 1989; Cassan et al. 2009), (4) phosphate-solubilizing bacteria (Arthrobacter or Agrobacterium; Belimov et al. 1995), (5) Rhizobium (Raverkar and Konde 1988; Er et al. 2004; Remans et al. 2008), (6) Pseudomonas (Joe and Sivakumar 2010; Combes-Meynet et al. 2011; Couillerot et al. 2011, Khorshidi et al. 2011,) and (7) Glomus (Mar Vázquez et al. 2000; Pulido et al. 2003), but only some of the combinations have shown enhanced plant growth stimulation compared to single inoculation (Belimov et al. 1995; El-Komy 2005; Remans et al. 2008; Combes-Meynet et al. 2011). To our knowledge, five studies have focused on interactions established in complex, Azospirillum-based mixed-inocula involving more than two other microorganisms, i.e., Rhizobium and arbuscular mycorrhizal fungi (AMF; Biró et al. 2000), Azotobacter and AMF (Singh et al. 2010), Burkholderia, Gluconacetobacter and Herbaspirillum (Oliveira et al. 2009), Pseudomonas and Rhizobium (Khan et al. 2009), or Pseudomonas and AMF (Walker et al. 2012). Depending on the strain combination, microbial interactions within these consortia had positive or negative effects on inoculant establishment on roots and resulted or not in enhanced plant growth in comparison with single inoculation, highlighting the importance of identifying synergistic strain combinations.

In the case of maize, the use of Azospirillum-Pseudomonas-Glomus consortia under low-fertilizer input conditions resulted in enhanced root system development compared with the non-inoculated control (Walker et al. 2012), but these positive effects might be enhanced if selection of the microbial partners was optimized. In this study, three prominent Azospirillum strains were compared when inoculated as part of a three-component consortium also including another PGPR inoculant (Pseudomonas fluorescens F113) and an AMF inoculant mix composed of three Glomus species. P. fluorescens F113 is another PGPR extensively studied as crop inoculant, and this strain has been shown to be a mycorrhiza helper bacteria (MHB; Garbaye 1994; Barea et al. 1998). In the case of maize, Azospirillum lipoferum CRT1 is one of the important PGPR strains used in Europe (Jacoud et al. 1998; Lucy et al. 2004; El Zemrany et al. 2006), whereas Azospirillum brasilense UAP-154 and CFN-535 inoculants are extensively used under agronomic conditions in Mexico, which is one of the leading countries in field inoculation (Dobbelaere et al. 2001; Fuentes-Ramirez and Caballero-Mellado 2006). Phytostimulation by Azospirillum strains has been shown to be cell-density dependant (Jacoud et al. 1999), and effective early root colonization is often required for effective stimulation (Dobbelaere et al. 2002).

Thus, the objective of this study was to compare A. lipoferum CRT1, A. brasilense UAP-154 and CFN-535 in Azospirillum–Pseudomonas–Glomus consortia for early promotion of maize growth. The experiment was performed under greenhouse conditions. Monitoring focused on rhizosphere survival of inoculated PGPR, root mycorrhization (noticeably to check lack of negative impact on indigenous AMF taxa), as well as on the effects on plant growth and on plant metabolic markers indicative of established plant–microbe interactions.

Materials and methods

Microorganisms

Bacteria used in this study are deposited at the MIAE strain collection in Dijon, France (http://www2.dijon.inra.fr/umrmse/spip.php?rubrique47), except for commercial strains A. brasilense UAP-154 and CFN-535 available at UNAM's Centro de Ciencias Genómicas in Cuernavaca, Mexico (http://www.ccg.unam.mx/en). A. brasilense strains UAP-154 and CFN-535 were both isolated from maize in Mexico (Dobbelaere et al. 2001), A. lipoferum CRT1 (MIAE 00337) was recovered from maize in France (Fages and Mulard 1988) and P. fluorescens F113 (MIAE 00794) was isolated from sugarbeet in Ireland (Fenton et al. 1992). To obtain Azospirillum inocula, strains were grown in NFb liquid medium (Döbereiner et al. 1976) supplemented with NH4Cl (0.2 g L−1) for 2 days at 30 °C with shaking (200 rpm). The cells were then washed two times with MgSO4 10 mM and the suspension adjusted to an optical density (OD600) of 0.6, giving 5 × 107 (for A. lipoferum CRT1), 2 × 109 (for A. brasilense UAP-154) and 7 × 109 (for A. brasilense CFN-535) CFU/mL. P. fluorescens F113 was grown in Luria Bertani medium supplemented with 0.25 g L−1 MgSO4⋅7H2O (LB-Mg) for 8 h at 30 °C and 200 rpm. The cells were washed two times with MgSO4 10 mM and adjusted to an OD600 of 0.2, giving 2 × 107 CFU/mL. Colony counts of inocula were obtained on RC agar (Rodriguez Caceres 1982) for Azospirillum strains and LB-Mg agar for P. fluorescens F113, after a 72-h incubation of plates at 30 °C.

The mycorrhizal inoculum consisted of a mixture of the Swiss isolates Glomus intraradices JJ291 (BEG accession 158 at the International Bank for the Glomeromycota; www.kent.ac.uk/bio/beg/), Glomus claroideum JJ360 (BEG 155), and Glomus mosseae JJ964 (BEG 161) (Jansa et al. 2005). Each was produced in plant cultures following commercial procedures by Symbio-M (Lanškroun, Czech Republic). Mycorrhized roots were chopped, mixed together and with zeolite carrier. The inoculum product contained 5.3 × 107 (G. intraradices JJ291), 2.9 × 107 (G. claroideum JJ360), and 2.5 × 106 (G. mosseae JJ964) gene copies of the nuclear Large ribosomal Sub-Unit (nLSU) of the respective AMF species per gram, as determined using the method of Thonar et al. (2011).

Greenhouse experiment

A greenhouse experiment was performed with sieved (4 mm) non-sterile soil taken from the loamy-sandy surface horizon of a Mexican oxisol from a field at Zacatepec near Cuernavaca, Morelos (clay 4.8 %, silt 7.9 %, sand 87.3 %, organic matter 4.3 %, pH 7.5). Seeds of maize (Zea mays) var. Costeño Mejorado (PROSASOL, Huitchila Morelos, Mexico) were surface-disinfected by stirring in sodium dichloroisocyanurate-containing Bayrochlor Mini solution (Bayrol, Dardilly, France) for 15 min, and washed several times with sterile distilled water (Couillerot et al. 2010a). The seeds were then germinated on water agar (8.5 g L−1) for 24 h in the dark at 30 °C.

Treatments included (1) a non-inoculated control, (2) inoculation with a two-component consortium composed of P. fluorescens F113 and Glomus mix, and (3) inoculation with a three-component consortium containing the two-component consortium and either A. lipoferum CRT1, A. brasilense UAP-154 or CFN-535. For each bacterial strain, inoculation was done by adding 1 mL of cell suspension (described above) to each germinated seed. In addition, 65 g of zeolite-formulated Glomus inoculum was placed approximately 3 cm below each germinating seed. Sterile water (2 mL) and non-inoculated zeolite (65 g) were used in the non-inoculated control and 1 mL sterile water in the Pseudomonas–Glomus treatment.

For the 10-day sampling, four maize plants were grown in 1-dm3 pots containing 1.5 kg soil previously supplemented with 270 mL sterile nutrient solution (described in Rodriguez-Salazar et al. 2009). For the later samplings (i.e., at 21 and 35 days), two maize plants were grown in 2-dm3 pots containing 2.3 kg soil previously supplemented with 340 mL of sterile nutrient solution. Five pots were used per treatment at each sampling, and the 75 pots were placed in a greenhouse (randomized block design) located at Cuernavaca (Mexico), with controlled temperature (26 ± 4 °C) and natural light. Watering was done by adding 270 and 340 mL of nutrient solution each day in 1 and 2 dm3 pots, respectively.

Sampling

Watering of the pots was reduced 48 h before the second and third samplings and stopped 24 h before each sampling. At each sampling, all shoots were cut off and dried for 2–4 days at 70 °C for biomass determination; one root system per pot was used for ipdC RT PCR analysis, and another root system per pot for real-time PCR quantification of PGPR inoculants (and AMF genotypes at the 35-day sampling, after splitting the root system in two parts). In addition, two other root systems per pot were used for plant metabolomic analysis at the first sampling (10 days).

DNA preparation

For PGPR inoculant monitoring, each root system was shaken vigorously to discard soil loosely adhering to the roots. Roots and tightly adhering soil were transferred in a 50-mL Falcon tube and flash-frozen in liquid nitrogen. Samples were then lyophilized for 48 h in Falcon tubes and homogenized by crushing in the tubes using a spatula, and 250–300 mg of lyophilized sample (rhizosphere soil + roots) were transferred in Lysing Matrix E tubes from the FastDNA® SPIN® kit (BIO 101 Inc., Carlsbad, CA). DNA was then extracted and eluted in 50 μL of sterile ultra-pure water, according to the manufacturer's instructions. DNA concentrations were assessed by OD measurements at 260 nm using NanoDrop (Nanodrop technologies, Wilmington, DE).

For AMF monitoring, roots from the third sampling were cut in 5-cm pieces. They were washed in ice-cold tap water, flash-frozen in liquid nitrogen, and lyophilized for 48 h in Eppendorf tubes. Lyophilized roots samples (25–35 mg) were homogenized by dry bead-beating three times 45 s with glass balls (1 mm diameter) in Biospec Beadbeater-8 (Fisher Scientific AG, Wohlen, Switzerland). DNA was then extracted with Plant DNeasy kit (Qiagen, Courtaboeuf, France) following manufacturer's recommendations.

Real-time PCR assessments

Bacterial root colonization was assessed by real-time PCR, as described in Couillerot et al. (2010a; for A. brasilense inoculants), Couillerot et al. (2010b; for A. lipoferum inoculant) and von Felten et al. (2010; for P. fluorescens inoculant). Briefly, real-time PCR for Azospirillum strains was done using the LightCycler® FastStart DNA Master SYBR® Green I kit and a LC-480 LightCycler (Roche Applied Science, Indianapolis, IN), and that for P. fluorescens F113 using the Fast SYBR® Green I kit and a 7500 Fast Real-Time PCR System (Applied Biosystems, Foster City, CA). Primers are described in Table S1.

AMF root colonization was assessed by real-time PCR targeting the nLSU RNA genes of G. claroideum, G. mosseae, G. intraradices, Gigaspora, Scutellospora (as described by Thonar et al. 2011) and Diversispora (as described by Wagg et al. 2011). Primers are described in Table S1. Additionally, two TaqMan markers targeting the mitochondrial Large ribosomal Sub-Unit (mtLSU) RNA gene of the G. intraradices species (i.e., mt5 marker) or the strain G. intraradices JJ291 (i.e., mt4 marker) were designed and validated (Table S2). Briefly, the inoculant strain G. intraradices JJ291, as well as the species G. intraradices, G. claroideum, and G. mosseae (to which the inoculants belonged) and the AMF genera Gigaspora, Scutellospora, and Diversispora were assessed using LightCycler® TaqMan® chemistry and a LightCycler 2.0 (Roche Applied Science), as described previously (Thonar et al. 2011; Wagg et al. 2011) or in the current supplementary information.

Melting curve calculation and determination of Tm values were performed using the polynomial algorithm function of LightCycler Software v.1 (Roche Applied Science) or of the Sequence detection Software v.1.4 (Applied Biosystems).

Normalization of the data

Plasmid APA9 (i.e., pUC19 with cassava virus insert; Genbank accession number AJ427910) was used as internal standard to normalize C T values. Real-time analyzes for Azospirillum, Pseudomonas, and Glomus AMF were carried out in different laboratories with different LightCyclers, so normalization was done separately. The protocol for A. brasilense inoculants is described in Couillerot et al. (2010a), for A. lipoferum inoculant in Couillerot et al. (2010b), for P. fluorescens inoculant in von Felten et al. (2010), and for AMF in Thonar et al. (2011). Briefly, known quantities of purified plasmid APA9 were added at the first step of each DNA extraction protocol, and real-time PCR analyzes with specific markers targeting the internal standard sequence were then performed on each DNA extract to estimate recovery rates of the internal standard. C T values thus obtained for the internal standard were used to normalize DNA extraction efficiency (Park and Crowley 2005).

Generation of standard curves for real-time PCR assessments

Real-time PCR quantification of PGPR inoculants in the rhizosphere required development of standard curves. Briefly, Lysing-Matrix E tubes (BIO 101 Inc.) containing 250–300 mg lyophilized sample (i.e., rhizosphere soil + roots) from the non-inoculated control (obtained as described above) were inoculated with one of the four PGPR strains. DNA extraction was performed using the FastDNA® SPIN® kit (BIO 101 Inc.) and real-time PCR was done as described above. A standard curve for each strain was generated by plotting the C T number against the logarithm of CFU added per g of soil, for the three independent replicates. Amplification efficiency was calculated from the slope of the standard curve using the formula \( E = {10^{{ - 1/{\text{slope}}}}} - 1 \) and standard curves were then used to estimate inoculant cell number in the rhizosphere of seed-inoculated maize plants. Real-time PCR quantification data were expressed as log cell equivalents per g of lyophilized rhizosphere sample (i.e., roots + tightly adhering soil).

Real-time PCR assays of the different AMF phylotypes was calibrated by using serially diluted cloned fragments (pGEM-T Easy vector, Promega) of the corresponding mtLSU, as described by Jansa et al. (2008) and Thonar et al. (2011). Real-time PCR quantification data were converted to gene copy number per gram of lyophilized root.

ipdC reverse transcriptase PCR analysis

Primers ipdCF3 (5´-CTTGCCCTTCTTCAAGGTGG-3´) and ipdCR3 (5´-GGGGGATTTCCAGATAGACC-3´) were designed for transcription analysis of auxin synthesis gene ipdC in Azospirillum spp., after alignment of all published ipdC sequences. Primer ipdCF3 displays 0 or 1 mismatch with Azospirillum ipdC sequences (n = 8) and at least 4 mismatches with the other ipdC sequences (n = 11), whereas primer ipdCR3 displays 0 mismatch with Azospirillum ipdC sequences and at least two mismatches with the other ipdC sequences (Fig. S2). Primer validation was performed by PCR (as described below) using genomic DNA from 33 Azospirillum strains, which showed that primers amplified ipdC in all 15 A. brasilense strains (including the two inoculants), in 2 of 11 A. lipoferum strains (including strain CRT1) and in Azospirillum doebereinerae GSF71, but not in six strains from Azospirillum canadense, Azospirillum halopraeferens, Azospirillum irakense, Azospirillum oryzae, or Azospirillum zeae (confirmed by sequencing).

Transcription of ipdC by Azospirillum spp. was measured by reverse-transcriptase (RT) PCR. Each studied root system (with tightly adhering soil) was placed in a 50-mL Falcon tube and flash frozen in liquid nitrogen. Samples were then washed with 35 mL solution of 100 mM CaCl2 and 50 mM Tris–HCl pH 7.0 (prepared with 0.5 % v/v DEPC treated water) supplemented with β-mercaptoethanol (5 %). This solution was centrifuged for 5 min at 6×g. Soil particles and root debris were discarded and the supernatant was centrifuged for 5 min at 2,250×g. The resulting cell pellet was resuspended in 1 mL of TRIZOL reagent (Invitrogen, Carlsbad, CA). DNAse treatment was then performed and cDNA synthesis was done with RevertAid™ H minus cDNA synthesis kit (Fermentas, Ontario, Canada). Low amounts of RNA were recovered, so only 600 ng of RNA were used per reaction. Amplification was done with 1.0 U of Taq DNA polymerase (Invitrogen), 2 μL of synthesized cDNA as template, 1× reaction buffer, 10 % DMSO, 1.5 mM MgCl2, 200 μM deoxynucleosides triphosphates, and 0.4 pmol of primers ipdCF3/ipdCR3. Because amplification was not observed directly in PCR assays, it was necessary to repeat it. For the second amplification, 4 μL of PCR product were diluted in 100 μL of water and 2 μL were used. Both PCR amplifications consisted of an initial denaturation for 5 min at 95 °C, 35 PCR cycles (30 s denaturation at 95 °C, 30 s annealing at 58 °C, and 30 s elongation at 72 °C), a final elongation for 10 min at 72 °C, and a cooling step at 5 °C.

Metabolomic analysis

At 10 days, the root systems from two plants per pot were washed with ice-cold distilled water and placed in aluminum envelops before being flash-frozen in liquid nitrogen. Samples were then lyophilized for 72 h and stored at −80 °C until analysis. Freeze-dried roots were introduced in Eppendorf tubes, to which liquid nitrogen was added. Roots were crushed using a ball mill (TissueLyser II, Qiagen), and extraction was performed using 2 mL methanol for 10 mg of dry sample, as described by Walker et al. (2011). Extraction was done twice and extracts were dried using Speedvac-assisted evaporation. Each sample was then resuspended in methanol to reach 10 mg dry extract per milliliter.

Chromatographic analysis of the extracts was achieved with an Agilent 1200 series HPLC (Agilent Technologies, Santa Clara, CA) equipped with a degasser (G132A), a quaternary pump module (G1311A), an automatic sampler (G1329A) and a Diode Array Detector (DAD G1315B), as described by Walker et al. (2011). 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). A step by step gradient was used with an increase of proportion of solvent B until 15 % during 5 min, then an isocratic level from 30 min, with a flux of 1 mL per minute. Chromatograms were recorded and processed at 254, 280, 310, and 366 nm. The Chemstation Agilent software was used for integration and comparison of chromatograms. Each chromatogram was integrated after standardization of integration parameters. Background peaks present on chromatograms were not integrated. Individual compounds were identified based on the results of Walker et al. (2011).

Statistics

Statistical analyses of real-time quantification data, ipdC expression data, and shoot biomass were performed at P < 0.05, using S plus software (TIBCO Software Inc., Palo Alto, CA). Chromatographic data obtained from root extracts, i.e., retention time and relative area of each integrated peak, were compiled in a matrix for discriminant principal component analysis (PCA), as described by Walker et al. (2011). Treatments were studied by ANOVA followed with Fisher's LSD tests (P < 0.05).

Results

Effect of microbial consortia on maize growth

By comparison with the non-inoculated control, inoculation of maize with the Pseudomonas–Glomus two-component consortium (i.e., P. fluorescens F113 and a mix of three Glomus isolates) resulted in higher shoot biomass at the first two samplings (Table 1). Shoot biomass was also higher in all three-component consortia than in the non-inoculated control at the first two samplings, but results were not influenced by the identity of the Azospirillum strain (Table 1). In addition, Azospirillum addition did not give any positive effect by comparison with the two-component consortium at the first sampling, and indeed shoot biomass in the two-component control was higher than those in the three-component consortia. The effect of inoculation on shoot biomass was not statistically significant at the third sampling, but this result was of limited significance. Indeed, roots had entirely colonized the whole soil volume in the pots by then, regardless of whether seeds were inoculated or not, and they could not expand further, thereby limiting the possibility of phytostimulation effects.

Table 1 Effecta of consortiumb inoculation on shoot dry weight (gram per plant) of maize (mean ± standard error; n = 5)
Full size table

Inoculant colonization of maize roots

When the Pseudomonas–Glomus two-component consortium was used, P. fluorescens F113 was enumerated by real-time PCR at about 108 cell equivalents per g of rhizosphere at the first sampling and at 2 log units lower by the third sampling (Fig. 1a). The presence of an Azospirillum inoculant resulted in higher F113 population levels at one of the three samplings (with A. lipoferum CRT1 or A. brasilense UAP-154) or had no effect on the pseudomonad (with A. brasilense CFN-535). As expected, strain F113 was not found in the non-inoculated treatment.

Fig. 1
figure 1

Root colonization of the PGPR strains P. fluorescens F113 (a) and Azospirillum (b) used as part of two-component (F113 + AMF mix; empty diamond) or three-component consortia (with A. lipoferum CRT1 (filled circle), A. brasilense UAP-154 (filled square), or CFN-535 (filled triangle)). Data represent means ± standard error (n = 5) of log cell equivalents per gram of rhizosphere. The detection limit (4 × 103 cell equivalents per gram of rhizosphere) is shown by dotted lines and symbols appear in white for Azospirillum inoculants below detection limit. Statistical differences between treatments at each sampling time are indicated with letters (ANOVA and Fisher's LSD tests; P < 0.05)

Full size image

The population size of A. brasilense CFN-535 dropped from 1.5 × 107 to 2.7 × 105 cell equivalents per gram of rhizosphere from the first to the third sampling based on by real-time PCR analysis (Fig. 1b). In comparison, A. brasilense UAP-154 and especially A. lipoferum CRT1 were recovered at lower levels, which fell below detection limit (4 × 103 cell equivalents per g of rhizosphere) by the second (for strain CRT1) or third sampling (for strain UAP-154). None of the Azospirillum inoculants (i.e., even the two Mexican isolates) was found in the non-inoculated treatment.

Among AMF inoculants, a real-time PCR quantification method was only available for G. intraradices JJ291. At the third sampling, this strain was not found in the non-inoculated treatment, but was detected in two of five replicate samples (at 41 and 128 mtLSU gene copies per milligram of dry root) when the Pseudomonas–Glomus two-component consortium was used. With the three-component consortia, G. intraradices JJ291 was only found in one of five replicates (at 60 mtLSU gene copies per milligram of dry root) in the treatment where A. brasilense CFN-535 was included.

Effect of inoculation on root-associated AMF populations

In the non-inoculated treatment, at the third sampling, the G. intraradices species was enumerated at 1.8 × 105 (nLSU method) and 2.0 × 105 (mtLSU method) gene copies, the G. claroideum species at 3.8 × 104 nLSU gene copies, and the Diversispora genus at 5.4 × 102 nLSU gene copies per mg of dry root (Table 2). When the two-component consortium was used, G. intraradices was recovered at higher level in comparison with the control, regardless of the method. With the three-component consortia, the population size of G. intraradices was comparable to that in the non-inoculated control and lower than that in the two-component control was used. Inoculation had no effect on the size of the G. claroideum species or the Diversispora genus (Table 2). The G. mosseae species and the AMF genera Scutellospora and Gigaspora were not found in any of the treatments. Thus, there was no negative effect of inoculation on root mycorrhization by indigenous AMF taxa.

Table 2 Effecta of consortium inoculation on AMF root colonizationb (expressed as thousand gene copies per mg of dry root) at 35 days after inoculation (mean ± standard error; n = 5)
Full size table

ipdC transcription in Azospirillum

The RT-PCR approach developed in this work targets ipdC from Azospirillum strains (mostly A. brasilense), but not in the other auxin-producing bacteria tested, including pseudomonads (not shown). Successful RT-PCR amplification of Azospirillum's ipdC mRNAs from rhizosphere samples was observed even in the absence of Azospirillum inoculation, i.e., in the non-inoculated control (at the first two samplings) and when the two-component consortium was used (at the first sampling) (Table 3). When three-component consortia were applied, transcription of Azospirillum ipdC genes (1) was found in 3–4 of 5 replicates at the first two samplings (versus only two replicates or less in the other treatments) and (2) was also detected at the third sampling.

Table 3 RT-PCR detection of Azospirillum ipdC gene, as indicated by the number of plants (out of 5) for which amplification was successful
Full size table

Effect of inoculation on secondary metabolite profiles of maize roots

Chromatograms at 280 nm for root methanolic extract gave 18 major integrated peaks, 11 of them corresponding to benzoxazinoid derivatives based on UV spectra (Walker et al. 2011). Polar compounds (based on water elution) were cyclic hydroxamic acids, whereas two more apolar compounds were benzoxazinone derivatives.

Discriminant PCA indicated that all inoculation treatments resulted in changes in the secondary metabolite profile of maize (Fig. 2). The inoculation impact varied according to the consortium, except that presence of A. brasilense UAP-154 within the Pseudomonas–Glomus two-component consortium had no effect. When assessing individual compounds responsible for treatment discrimination, it appeared that the prevalence of five PCA-discriminant secondary metabolites (including three benzoxazinoid derivatives and one cinnamic acid) differed significantly between treatments based on ANOVA and Fisher's test (Fig. 3).

Fig. 2
figure 2

Discriminant PCA performed on chromatographic data obtained for each methanolic extract of maize. Analyses were based on peak areas and retention times. Each point represents two pooled extracts (i.e., two plants) of the same treatment

Full size image
Fig. 3
figure 3

Effect of maize seed inoculation on root content in individual secondary metabolites that distinguished treatments in discriminant PCA. Compound identification was based on UV spectra. For each compound, statistical differences between treatments are indicated with letters (ANOVA and Fisher's tests; P < 0.05; n = 5)

Full size image

Discussion

Azospirillum PGPR strains have been extensively studied as phytostimulatory inoculants of cereal crops (Okon and Labandera-Gonzalez 1994; Dobbelaere et al. 2001), and to a much lesser extent in mixed inocula combining phytobeneficial microorganisms with different metabolic capacities (Bashan 1998). Indeed, most studies on mixed inocula containing diazotrophic bacteria have been performed with bacteria other than Azospirillum (Biró et al. 2000; Remans et al. 2008; Cassan et al. 2009; Oliveira et al. 2009). Several attempts have been made to combine Azospirillum with a microorganism such as Pseudomonas (Corich et al. 1995; Joe and Sivakumar 2010; Combes-Meynet et al. 2011; Couillerot et al. 2011; Khorshidi et al. 2011), which may function also as a biocontrol agent. Combining Pseudomonas antagonistic biocontrol agents and Azospirillum requires special attention regarding potential inhibitory effects of Pseudomonas antimicrobial metabolites, such as 2,4-diacetylphloroglucinol (DAPG), against Azospirillum (Couillerot et al. 2011). We verified that the three Azospirillum strains used in this study were rather resistant to DAPG, as growth inhibition required as much as 500 μM of synthetic DAPG. In fact, synergistic effects might even be expected since DAPG can act a signal enhancing the expression of phytostimulation-relevant genes in Azospirillum (Combes-Meynet et al. 2011).

In all inoculation treatments, P. fluorescens F113 colonized maize roots extensively. Its population level was significantly enhanced when the pseudomonad was in presence of certain Azospirillum inoculants, which here was only found at one sampling time but confirms a previous observation under field conditions (Walker et al. 2012). The three Azospirillum strains showed very different root colonization abilities. Only A. brasilense CFN-535 managed to colonize roots significantly (i.e., at levels above 105 cell equivalents per gram of rhizosphere) and durably (i.e., till the last sampling). A. lipoferum CRT1 declined as fast (Walker et al. 2012) or faster (El Zemrany et al. 2006; Baudoin et al. 2009a) than in maize trials done in Europe, and perhaps soil type and/or maize variety play an important part in this phenomenon. Auxin synthesis is often considered the main mode of action in Azospirillum PGPR (Dobbelaere et al. 2003), but here the ipdC gene proved too conserved within the Azospirillum genus to enable strain-specific PCR monitoring of ipdC expression. In this work, the RT-PCR approach developed for A. brasilense and related strains proved operational with rhizosphere samples, even though amplification was not successful with all plants studied. The fact that plants not inoculated with Azospirillum yielded ipdC RT-PCR signals points to the contribution of resident Azospirillum strains present in the soil used. Further work will be necessary to understand why, at the third sampling, Azospirillum ipdC transcription was found in one third of Azospirillum-inoculated plants (including in treatments where the Azospirillum inoculant was below detection limit) versus none of the plants not inoculated with Azospirillum.

Little is known about the genetic diversity of root-associated AMF communities, despite their importance for plant growth. Here, recent PCR methods for analysis of AMF community composition were used to probe AMF taxa colonizing maize roots in Mexican soils, and results showed that several AMF taxa were well established. The prevalence of G. intraradices was twice as high when the two-component consortium was used, regardless of whether a mtLSU or nLSU approach was used (Table 2), in accordance with the good correspondence between both approaches (Fig. S1). It is tempting to speculate that this was due to the inoculation of G. intraradices JJ291, but PCR monitoring of the latter fell below expectations. Indeed, G. intraradices JJ291 was not present in the non-inoculated control (as expected), but its detection was poorly effective in this and other inoculation treatments, thereby limiting the usefulness of this assessment in the current experiment. Strong competition with indigenous AMF can be expected (Biró et al. 2000). Results indicate also that presence of DAPG-producing P. fluorescens F113 had no apparent deleterious impact on root-associated AMF, despite antifungal properties of DAPG (Barea et al. 1998; Mar Vázquez et al. 2000; Gaur et al. 2004).

AMF establishment in the three-component consortia was comparable to that in the non-inoculated control. When compared to the two-component consortium, this suggests that presence of Azospirillum prevented enhanced establishment of G. intraradices. Previous analyses failed to evidence any effect of Azospirillum on mycorrhization (Mar Vázquez et al. 2000; Russo et al. 2005), but methodology differed.

Stimulation of maize shoot growth was significant when seeds were inoculated with any of the three-component consortia, i.e., whatever the Azospirillum strain involved. It is interesting to note that this took place despite (1) contrasted survival dynamics for different Azospirillum inoculants, and (2) maize secondary metabolite profiles that varied between most treatments. Maize elaborated specific metabolic patterns according to the Azospirillum strain present, showing thus a specific impact of Azospirillum presence within the three-component consortia. The variation induced by microbial inoculation concerned several types of secondary compounds, including some already identified (Walker et al. 2011, 2012). In addition, it was rather unexpected that presence of Azospirillum in the inoculum resulted in lower maize stimulation in comparison with the Pseudomonas–Glomus two-component consortium, but this effect was transient. Its molecular basis remains unknown.

In conclusion, this study indicated that Azospirillum–Pseudomonas–Glomus three-component consortia may be useful for early stimulation of maize growth. Despite evidence for distinct interaction functioning according to the Azospirillum strain included, the identity of the Azospirillum strain was not a significant factor determining phytostimulation efficiency. This is an important finding in biotechnological terms, as it will facilitate development of microbial consortia for crop inoculation.