The bacterium Paenibacillus validus stimulates growth of the arbuscular mycorrhizal fungus Glomus intraradices up to the formation of fertile spores


  • Editor: Holger Deising

Correspondence: Hermann Bothe, Botanical Institute, University of Cologne, D-50923 Köln, Germany. Tel.: +49 221 470 2760; fax: +49 221 470 5039; e-mail:


Two isolates of Paenibacillus validus (DSM ID617 and ID618) stimulated growth of the arbuscular mycorrhizal fungus Glomus intraradices Sy167 up to the formation of fertile spores, which recolonize carrot roots. Thus, the fungus was capable of completing its life cycle in the absence of plant roots, but relied instead on the simultaneous growth of bacteria. The supernatant of a mixed batch culture of the two P. validus isolates contained raffinose and another, unidentified trisaccharide. Among the oligosaccharides tested, raffinose was most effective in stimulating hyphal mass formation on plates but could not promote growth to produce fertile spores. A suppressive subtractive hybridization library followed by reverse Northern analyses indicated that several genes with products involved in signal transduction are differentially expressed in G. intraradices SY 167 when grown in coculture with P. validus (DSM 3037). The present investigation, while likely representing a significant step forward in understanding the arbuscular mycorrhizal fungus symbioses, also confirms that its optimal establishing and functioning might rely on many, as yet unidentified factors.


The importance of arbuscular mycorrhizal symbiosis has been fully recognized only in recent years (Smith & Read, 1997; Varma & Hock, 1998). The fine fungal hyphae effectively penetrate into the soil particles and more effectively exploit water and minerals than the plant roots. Exchange of these nutrients is believed to occur mainly at the arbuscules. Fungi in turn receive c. 20% of the organic carbon fixed by the plants, probably as glucose. Colonization of roots helps plants to cope with adverse environmental conditions such as nutrient deficiency, injury by pathogens, stress caused by strong soil acidity or pollution by heavy metals and salts. Because of these broad implications mycorrhizal symbiosis is, ecologically, the most important symbiosis. In nature, more than 80% of higher plants are colonized by arbuscular mycorrhizal fungi (AMF). However, by comparison with the best-studied Rhizobium–legume symbiosis, investigations on AMF are still at infancy and are hindered by the inability to grow the fungi independently of the plant hosts. Interactions between both symbiotic partners are complex and apparently do not rely on the exchange of only one or few, easily identifiable molecules.

There are repeated reports that specific bacteria promote the interactions between AMF and plants and therefore could possibly serve as a third partner in this symbiosis (Garbaye, 1994; Azcón-Aguilar et al., 1998; Rillig, 2004). A bacterium related to the genus Burkholderia was identified in spores and hyphae of the AM fungus Gigaspora margarita. This bacterium possesses the nitrogenase genes nifHDK and expresses the nifH gene in spores (Bonfante, 2003). Recently, an isolate from Gigaspora, termed Candidatus Glomeribacter gigasporum, was obtained, which, however, could not be grown independently of the host and apparently has lost the nitrogenase genes (Jargeat et al., 2004). Paenibacillus ssp. isolated from the roots of Sorghum bicolor stimulated mycorrhizal colonization of this plant and suppressed in vitro mycelium growth of Phytopthora parasitica (Budi et al., 1999). Bacteria of the genus Paenibacillus were also shown to live intracellularly in the ectomycorrhizal fungus Laccaria bicolor (Bertaux et al., 2003) and could have a fungal growth promoting effect also in this symbiosis.

This laboratory reported that Paenibacillus validus, isolated as a contaminant from plates with germinating spores of Glomus intraradices, promoted growth of this fungus (Hildebrandt et al., 2002). Independently of any plant impact, G. intraradices was grown on plates until the formation of new spores. In comparison with the yield obtained by growth of G. intraradices in coculture with Ri-T-DNA transformed carrot roots (Mosse & Hepper, 1975; Bécard & Fortin, 1988), the spores obtained by growth with P. validus did not reach the size of those originally used for the inoculation (Hildebrandt et al., 2002). The newly formed spores contained plenty of nuclei, which apparently had multiplied during the growth with P. validus. However, it was not thoroughly examined whether these newly formed spores were effective in germinating and recolonizing plant roots.

This article describes the interactions between Paenibacillus and G. intraradices in more detail. Conditions were established where a mixture of Paenibacillus isolates stimulated growth of G. intraradices until the formation of newly colonizing spores. The supernatant of batch cultures of P. validus was screened for substances that promote growth of G. intraradices. Finally, a suppression subtractive hybridization (SSH) library from plates of G. intraradices germinating either in the presence or the absence of P. validus has been constructed. This SSH library revealed genes that were differentially expressed by G. intraradices in the presence of P. validus. The data presented here indicate that growth of the fungus, although relying on complex factors, can be achieved, under certain conditions, independently of the obligate biotrophy with plants.

Materials and methods

Organisms used

Bacterial strains were obtained from the German collection of Microorganisms in Braunschweig (DSMZ) where the own isolates of Paenibacillus validus ID617 and ID 618 were also deposited. The AMF strains were Glomus intraradices from Premier Tech Mycorise ASP (Rivière-du-Loup, QC, Canada), Glomus lamellosum MUCL 43195, Glomus geosporum BEG 11, Glomus mosseae BEG12 and G. margarita BEG34. G. intraradices Sy167, originally obtained from the late Professor H. Marschner, D-Hohenheim, was maintained in our laboratory for years.

Growth of the organisms

For the analysis of the supernatant by gas chromatography/mass spectrometry (GC/MS), the mixture of P. validus DSM ID617 and ID618 was grown in modified M medium (Bécard & Fortin, 1988) until late logarithmic growth and harvested after 2 days (OD 560 nm about 1.0). Cells (10 L) were centrifuged (3000 g, 10 min), and the supernatant was analysed by GC/MS. After freeze drying the supernatant, 10 mg of the residue were derivatized in a two-step procedure as follows: 40 μL of methoxyamine hydrochloride solution (20 mg mL−1 in pyridine) were added and the solution was heated to 40°C for 90 min. Subsequently, 40 μL N-methyl-N-trimethylsilyltrifluoracetamide (MSTFA) were added and the mixture was heated to 60°C for 90 min. Finally, 20 μL of an alkane mixture (C10, C12, C15, C19, C22, C28, C32, C36; 0.25 mg mL−1 each in cyclohexane) were added for the determination of the retention indices (RI) of the eluted compounds.

Analysis of the derivatized growth medium was performed on a TraceMS GC/MS system (Thermo, San Jose, CA) equipped with Trace GC and AS 2000 autosampler. The system was operated under Xcalibur (version 1.2). A programmed temperature vaporizer (PTV) injector supplied with 120 × 2 mm glass liner was used. Upon injection of 2 μL of the sample at 70°C the solvent was evaporated at this temperature for 2 min. Subsequently, the injector was heated to 280°C at a rate of 14°C s−1, remaining at 280°C for 10 min. Separation was achieved on a DB5-MS column (30 m × 0.25 mm, 0.25 μm film, J&W Scientific, Folsom, CA) with He (1 mL min−1, split 25 mL min−1) as carrier gas. The temperature of the GC oven was raised from 70°C (1 min) to 76°C with 1°C min−1 and then with 6°C min−1 to 325°C, followed by an isothermal period of 10 min. Mass spectra were obtained in EI mode at 70 eV and a source temperature of 200°C. The detector was operated at 500 V and the emission current was set to 150 μA. Full scan mass spectra were acquired from m/z 39 to 782 at a rate of 1 scan s−1.

For the monoxenic culture with P. validus, G. intraradices or the other isolates mentioned in Table 1 were grown on plates in modified M-medium (Bécard & Fortin, 1988) and 0.4% (weight in volume, w/v) gellan gum (GelGro, ICN, D-Eschwege, Germany), which allowed subsequent solubilization, as described (Hildebrandt et al., 2002). The inoculum consisted of about 1000 AMF spores distributed on the surface of the plate. The spores were then covered by a conventional dialysis membrane (molecular weight cut-off 12 kDa) followed by 10 mL of liquid 0.8% Bacto agar dissolved in M-medium at 55°C. This procedure did not affect the germination rate of the spores. A small portion of P. validus was carefully removed from a preculture growing on modified-M-medium agar with an inoculation loop and spread on the layer of the Bacto agar/M-medium. This experimental set-up separated both microbial partners and therefore allowed the subsequent isolation of fungal material. Controls were grown similarly but without bacterial inoculation. After 3 months of the coculture at room temperature, the fungal material could be isolated by solubilization of the GelGro agar with citrate buffer (pH 6.0) at 30°C (Doner & Bécard, 1991) and stored in liquid nitrogen prior to use for RNA extraction. The dual culture of G. intraradices and carrot roots was maintained for 2 months at 27°C as described (Ouziad et al., 2005a).

Table 1.   Growth of mycorrhizal fungi in dependence of bacteria
 Stimulation of growthFormation of DPC**Formation of spores
  1. Groups of spores (50–100) were germinated on agar plates supplemented with M-medium. After 1 week of growth, the bacterial culture was added. The stimulation of growth and the formation of arbuscular mycorrhizal fungi structures were determined 12 weeks after the germination of spores.

  2. **, DPC, densely packed coils; +++, growth till the formation of mature spores with the same size as the mother spores; ++, till the formation of new spores which were; however, smaller and not mature also formation of densely packed coils, as described in Hildebrandt et al. (2002); +, small stimulation of the growth of the fungal hyphae and formation of small-sized vesicle like structures; −, no growth stimulation; *, not demonstrable, see Results.

(a) Growth of Glomus intraradices Sy167, bacterial strain varied
Paenibacillus polymyxa DSM36
Paenibacillus amylolyticus DSM 3034
Paenibacillus validus DSM 3037++++++
Paenibacillus validus (own isolate: DSMZ ID 617)++++++
Paenibacillus validus (own isolate: DSMZ ID 618)++++++
Paenibacillus validus (mix of own isolates: DSM ID 617 and 618)+++++++++
Bacillus mycoides DSM 2084
Escherichia coli K12
Azospirillum brasilense Sp7
Rhizobium etli G12+ (first 3 weeks of coculture)+ (remain small, ∅ about 40 μm)
Control (no bacterium added)−, however, in rare cases formed but remain small (∅ about 30 μm)
(b) Growth in the presence of the Paenibacillus validus mixture, fungal strain varied
Glomus intraradices Sy167+++++++++
Glomus intraradices“Premier Tech” Mycorise ASP+++++++++
Glomus lamellosum MUCL 43195+++++++++
Glomus geosporum BEG11Not demonstrable*Not demonstrable*Not demonstrable*
Glomus mosseae BEG12*Not demonstrableNot demonstrable 
Gigaspora margarita BEG34

Growth of Glomus intraradices in the presence of the different oligosaccharides

Glomus intraradices Sy l67 was grown on modified M-Medium with either 10 mM, 100 mM or 1 M raffinose (or the other oligosaccharose mentioned in Table 2) and 0.4% (w/v) GelGro. For this, the plates were supplemented with the different sterile filtered oligosaccharide solutions to reach the final concentrations in the plates. Then about 1000 spores obtained from the dual culture with Ri T-DNA transformed carrot roots were added. After 2.5 months, the fungal material was harvested by solubilization of the GelGro with citrate buffer (pH 6.0) at 30°C and thoroughly mortared in liquid nitrogen. Fungal growth was determined by measuring the protein content. Protein was extracted from the fungal material harvested from three plates with 40 μL of extraction buffer (10 mM Tris-HCl pH 8.0, 1 mM EDTA, 2% polyvinylpolypyrrolidone, 0.5% leupeptin, 0.5% phenylmethylsulfonylfluoride, w/v). After 30 min of centrifugation (30 min, 4°C), the protein content of the supernatant was determined by the Lowry method.

Table 2.   Stimulation of hyphal growth of Glomus intraradices Sy167 by different saccharides
SaccharideConcentration used
10 mM100 mM1 M
  1. Data are given in μg protein as well as total hyphal weight (in mg) per three plates. Standard deviations are given from three different trials.

Control without any additional saccharide255±233.0±0.585±6.8      

RNA extraction and SSH library construction

Total RNA was extracted using RNeasy spin columns (Qiagen, D-Hilden, Germany) by following the supplier's protocol. The obtained fungal RNA was freed from DNA using DNAseI (Invitrogen, D-Karlsruhe, Germany). RNA amount and quality were controlled by electrophoresis and measuring absorbance. For suppressive subtractive hybridization (SSH) (Diatchenko et al., 1996), doubled-stranded complementary DNA (cDNA) was obtained using the SMART-PCR cDNA Synthesis Kit (Clontech, D-Heidelberg, Germany) from 1 μg of total RNA of G. intraradices Sy167 spores and hyphae, grown either with Paenibacillus (tester) or without Paenibacillus (driver). The cDNA populations were then subtracted using the PCR-Select cDNA Subtraction Kit (Clontech) following the manufacturer's instructions. The remaining cDNA was amplified and 1 μL of the final PCR reaction products was cloned into the pGem-Teasy vector (Promega, D-Mannheim, Germany).

Reverse northern analyses with the RNA from clones of the SSH library

Inserts of the recombinant clones were amplified by PCR using the primers Nested PCR primer1 and Nested PCR primer2R (sequences available at that flank the borders of every insert from the subtractive library. PCR was performed using Taq polymerase (Promega) in a volume of 50 μL with 40 cycles (15 s at 95°C, 30 s at 65°C, 90 s at 72°C). Aliquots (15 μL) of the PCR products were separated in a 1.5% agarose gel and blotted in parallel onto two nylon membranes (Biodyne B, Pall, Pensocola, FL) using a vacuum blotting manifold (2016 VacuGeneXL Vacuum Blotting System, Amersham Biosciences, D-Freiburg, Germany) according to the manufacturer's protocol and fixed by baking at 120°C for 30 min. A digoxigenin (DIG)-labelled probe was prepared from each population (driver and tester) using SMART cDNA as the template, by incorporation of DIG-dUTP (Roche, Mannheim, Germany) in a PCR reaction, using the SMART PCR primer from the SMART cDNA synthesis kit. Hybridizations (using DIG-Easy Hyb from Roche, T=42°C) and filter washings were performed as described by the manufacturer (Roche). The DIG-labelled hybridized DNA was detected immunologically by alkaline phosphatase-conjugated antibodies using CSPD=disodium 3-(4 methoxyspiro {1,2-dioxetan-3,2′-(5′chloro) tricyclo [,7]decan}-4-yl) phenylphosphate as the substrate.


Glomus intraradices can be grown in culture only with Paenibacillus validus up to the formation of fertile spores

The stimulatory effect of bacteria on the growth of Glomus intraradices was now studied in more detail here than in the preceding publication (Hildebrandt et al., 2002). Of the bacteria tested, only Paenibacillus validus, and no other aerobic spore forming bacteria such as Paenibacillus polymyxa, Paenibacillus amylolyticus or Bacillus mycoides, was able to stimulate growth of hyphae, the formation of massive assemblies of thin hyphae, called densely packed coils (DPC, for their appearance see Figs 1a–c in Hildebrandt et al. (2002) and Fig. 1, this study) as well as the synthesis of new spores independently of any impact of a plant or plant-derived compound (Table 1a). Significantly, a 1 : 1 combination of the our two isolates, P. validus DSMZ ID 617 and 618, was more effective in stimulating growth of the fungus than one P. validus isolate alone. When the combination of these two P. validus isolates was applied, the newly formed spores reached full size with an average diameter of 70 μm after 3 months. Spores of the same size were also obtained when G. intraradices was grown in cocultures with Ri-T-DNA transformed carrot roots. It must be stressed that the spore number obtained with the P. validus combination was only 8±7 per mother spore and plate and thus not comparable with the large amount of spores formed in the dual system with carrot tissues which, however, had arisen from many (around 100) mother spores. There were no indications of the formation of arbuscules or vesicles.

Figure 1.

 Growth of Glomus intraradices with the mixture of the two Paenibacillus validus isolates DSM ID 617 and ID 618. After 14 weeks of growth, densely packed coils (DPC) (a, b) and spores with the same size and appearance as the mother spores (c, d) are formed. After solubilization from the GroGel agar these spores were able to recolonize Ri-T-DNA carrot roots as shown in Fig. 1(e)–(h). Sixteen weeks later, G. intraradices had formed as many spores of the normal size as obtained with any other spore material taken from the biotrophic growth with plants. Coculture with carrot roots (e), spores and hyphae growing into the root-free plate compartment (f–h). Bar=100 μm in each case.

Among the other bacteria tested, only Rhizobium etli G12 showed some growth stimulatory effect on G. intraradices Sy167 (Table 1a). The formation of hyphal mass was enhanced within the first three weeks of growth but did not continue. In addition, spore-like vesicles with a diameter of approximately 40 μm were detectable in small numbers. These structures were slightly larger than the spore-like vesicles occasionally seen after spores were germinated on plates in the absence of any bacterium or plant structure.

With regard to the fungal side, the P. validus combination stimulated the growth of two G. intraradices isolates and G. lamellosum but not of Gigaspora margarita (Table 1b). In the case of G. geosporum and G. mosseae difficulties had to be encountered with the germination of spores under aseptic conditions. In both cases, only about 10% of the approximately 100 spores obtained from pot cultures germinated on the plates. In addition, plates with germinating spores were often contaminated by other bacteria, indicating that the surface sterilization of spores of these two fungi was inadequate. Thus, any effect of P. validus on the growth of these two fungi on plates could not be assessed with certainty.

Spores obtained from the coculture of the mixture of the two P. validus isolates and G. intraradices Sy 167 were fertile (Fig. 1). The mean spore diameter in μm was 73±15 (n=30) after coculture with either Ri-T-DNA transformed carrot roots or P. validus. Spore and hyphal material taken from the plates grown in coculture with P. validus recolonized Ri-T-DNA transformed carrot roots as effectively as spores obtained from coculture with Ri-T-DNA transformed carrot roots or after growth with Tagetes. Owing to the stickiness of the spore and hyphal material on the plates after growth with P. validus single spores could not be isolated for recolonization. Therefore, the spores and hyphae were isolated from the plates by GelGro solubilization and used as inocula for the growth in coculture with carrot roots. Visual inspection showed that most of the spores formed fresh germ tubes upon growth on the plates in coculture with carrot roots. After 3 months, the fungi then had formed as many new, mature spores as any other G. intraradices inoculum obtained from pot cultures with tomato or Tagetes. The mean spore diameter in μm was also the same in three independent experiments: 62±10 (n=30) for those originated only from the carrot root system and 60±8 (n=30) for those obtained after the coculture with the two P. validus isolates. Thus, Koch's postulate was fulfilled, as the spores were isolated and reinfective.

Raffinose stimulated hyphal growth of Glomus intraradices

In order to identify molecules that are specifically formed and excreted by P. validus and which may stimulate growth of the fungal hyphae, the supernatant of a batch culture of both Paenibacillus DSM ID617 and ID618 was analysed by GC/MS. Besides a heavily overloaded peak of oktakis-(trimethylsilyl)-sucrose (37.96 min) as the growth substrate the gas chromatogram (Fig. 2) showed four signals of the pentakis-(trimethylsilyl) methoximes (cis/trans isomers) of fructose (26.92 and 27.08 min) and glucose (27.37 and 27.62 min) and hexakis-(trimethylsilyl)-myo-inositol (30.33 min, internal standard). Additionally, a small signal was present at 46.12 min, which on more detailed analysis (Fig. 2, inset) proved to consist of two peaks, which were identified as persilylated trisaccharides from their retention behaviour and mass spectral fragmentation pattern. The smaller peak eluting at 46.05 min matches with silylated raffinose, as shown by comparison of RI and mass spectra with an authentic standard. The compound providing the larger peak could not be identified as yet, but might be a compound closely related to raffinose. Any attempt for a preparative separation of these trisaccharides failed thus far. The two substances were apparently synthesized only in small amounts, and we had to encounter difficulties to separate these substances from the large concentration of sucrose in the medium thus far. A peak in the gas chromatogram typical for trisaccharides was not detected in the supernatant of batch cultures of P. polymyxa DSMZ 36.

Figure 2.

 Oligosaccharides found in the supernatant of a batch culture of a mixture of Paenibacillus validus DSM ID 617 and ID 618 as determined by gas chromatography and mass spectrometry.

To assess any effects of saccharides on fungal growth, G. intraradices Sy167 was germinated on plates with either raffinose, melobiose, trehalose, glucose, fructose or in addition to sucrose as carbon source in the growth medium (Table 2). Indeed, 10 mM raffinose significantly stimulated hyphal growth by almost 50%, as determined by the hyphal protein content and total weight of the hyphae after a culture period of 2.5 months. The effect was even more distinct at higher raffinose concentrations (100 mM and 1 M). However, raffinose did not stimulate fungal growth as far as the formation of new, fertile spores. DPC and branched absorbing structures (BAS) were also not detectable. A slight increase in the protein content was also measured when trehalose (10 and 100 mM) was the carbon source. Melibiose and glucose and fructose were ineffective in stimulating hyphal growth or showed, at best, only a small increment in total protein and weight above the control value (Table 2).

Genes differentially expressed by Glomus intraradices upon colonization by Paenibacillus validus

To detect such genes, an SSH library was constructed using total RNA of 3-month-old spores and hyphae of G. intraradices. For this, G. intraradices was germinated either with P. validus DSM 3037 (providing the tester RNA) or without (for the driver RNA). Using the SMART-PCR cDNA Synthesis Kit and the PCR-Select cDNA Subtraction Kit, 480 cDNA clones were obtained from the tester RNA. Their sequence list revealed some interesting genes: a putative nitrate transport protein, a high affinity ammonium transporter and, as the only protein possibly involved in saccharide metabolism, a UTP-glucose-1-phosphate uridyltransferase. However, the sequence homology was weak in some cases, e.g. for the nitrate transporter. To verify which of the genes were, indeed, differentially expressed in G. intraradices upon coculture with P. validus, reverse Northern analyses were performed with the SSH library. The aforementioned genes were not differentially expressed. Altogether, 43 genes showed an at least twofold more intense hybridization signal in the reverse Northern blots with total cDNA from the fungus grown with P. validus compared with the controls with RNA from the noninoculated G. intraradices (Fig. 3). Among these, the sequence of four clones did not score in the databanks and 20 matched with the 28S rDNA. Among the remaining 19, several genes coded for products possibly involved in signal transduction induced by P. validus such as Ca2+/calmodulin-dependent protein kinase, GTP-binding protein Rab11b and others (Table 3). As discussed below, an O-linked N-acetylglucosamine transferase may be particularly interesting, although its sequence homology to the corresponding gene of an archaea was only around 50% (Table 3).

Figure 3.

 Reverse Northern analyses with the RNA from clones of the suppressive subtractive hybridization library. The genes marked in bold (no. 1–7) are differentially expressed. 1, Ca2+/calmodulin-dependent protein kinase; 2, 28S ribosomal RNA; 3, serine/threonine-protein kinase; 4, GTP-binding protein Rab11b; 5, polyubiquitin; 6, O-linked N-acetylglucosamine transferase; 7, putative senescence-associated protein; 8, probable nitrate transport protein; 9, zinc transporter protein ZIP1; 10, high affinity ammonium transporter; 11, glutamine synthetase; 12, glutathione methyl-S-transferase; 13, UTP-glucose-1-phosphate uridyltransferase; 14, formate/nitrite family of transporter.

Table 3.   Fungal genes of the suppressive subtractive hybridization library that were differentially expressed as revealed by reverse northern analysis
Size (bp)Best hitOrganismAccession numberE-valueHomology (%)
370Ca2+/calmodulin-dependent protein kinasePhyscomitrella patensAAO06899.18e−0853
532Serine/threonine-protein kinaseNeurospora crassaCAD79666.12e−1447
418Protein kinaseArabidopsis thalianaAAG51328.17e−1456
344GTP-binding protein Rab11bMus musculus (mouse)A550053e−5798
489Putative senescence-associated proteinPisum sativumBAB33421.12e−4571
330RloECampylobacter jejuniAAM00871.12.052
611PolyubiquitinPhanerochaete chrysosporiumS34655e−10297 proteinHaemonchus contortusAAR99585.16e−0566
570Heat-shock protein 90Cryptococcus bacillisporusAAN76524.14e−6386
422Immunoglobulin heavy chain variable regionHomo sapiensAAM87871.13.450
460O-linked N-acetylglucosamine transferaseMethanosarcina acetivorans str C2AAAM04780.15e−1353
477Putative RNA helicaseSaccharomyces cerevisiaeAAB17005.12e−2158
433Leucine aminopeptidaseBos taurus1LAM2e−4679
334HMG box mitochondrial transcription factor AXenopus laevisAAA91456.10.00453
368Putative Zn-proteaseSchizosaccharomyces pombeT409721e−0757
254Hypothetical proteinGibberella zeae PH-1FG06369.18e−0860
415Hypothetical proteinGibberrella zeae PH-1FG08494.11e−2175
518Hypothetical proteinAspergillus nidulansGi 490928047e−0753
497Hyaluronate lyaseStreptococcus pneumoniaeF979072.641


It had been speculated that the fungi, during the long evolution in symbiosis, have lost genes, which makes them indispensable to the host plant (Gadkar et al., 2001). Such a speculation is no longer valid, as the present study showed that the AMF Glomus intraradices can be grown until the formation of fertile spores independently of any plant structure. It had also been described that intraradical hyphae take up carbon from the plants and synthesize lipids whereas extraradical hyphae are unable or have a limited ability to do this (Douds et al., 2000). As shown in the present study, in the coculture with Paenibacillus validus, hyphae on the plates must be able to take up organic carbon to meet the full demand of the fungus during its life cycle and therefore must functionally behave like intraradical hyphae of the symbiosis. It should be emphasized that P. validus was not in physical contact with the hyphae but separated from them by a dialysis membrane (Hildebrandt et al., 2002). Thus, chemical compounds or enzymes excreted by P. validus, which convert constituents of the medium to nutrients required by the fungus, are the cause for the monaxenic growth of G. intraradices. It should only be a matter of time for the nutrients or signals required for axenic growth of AMF, independently of a second partner, to be identified.

It has also been argued that only active arbuscules are able to absorb nutrients, possibly because of their structure and permeability properties (Aczón & Ocampo, 1984). Typical arbuscules (and also vesicles) are not formed by G. intraradices in coculture with P. validus. It is therefore tempting to speculate that the DPCs formed on the plates inoculated with the bacterium (Hildebrandt et al., 2002) serve as feeding structures. Likewise, arbuscules are not synthesized by AMF grown in coculture with carrot roots outside the roots, and arbuscule-like structures (Chabot et al., 1992) or BASs (Bago et al., 1998) were postulated to take up nutrients from the outside in such a coculture.

In contrast to the preceding publication (Hildebrandt et al., 2002), where one P. validus isolate stimulated growth of G. intraradices until the formation of spores which, however, were small, the combination of two P. validus isolates now promoted the formation of new and fertile spores of the same size as the parental ones. This specific interaction of the two P. validus isolates is somewhat surprising. Beneficial effects of bacteria of different taxonomic affinities have repeatedly been reported since the time of Mosse (1959). Effects of bacteria of the Bacillus group, in particular of members of the more recently recognized genus Paenibacillus (Ash et al., 1993), seem to be more distinct. Members of the genus Paenibacillus were more frequently encountered in the washed extract of cucumber plants than in any other treatment (Mansfeld-Giese et al., 2002). Only Gram-positive bacteria (mostly Paenibacillus ssp. and Bacillus ssp.) were observed to be associated with fungal hyphae, and one of them, Bacillus cereus strain VA1, was even attached to AMF hyphae (Artursson & Jansson, 2003). Paenibacillus may be involved in aggregation of root-adhering soil (Bezzate et al., 2000), which may facilitate mineral acquisition by the hyphae, in the suppression of pathogens (Bezzate et al., 2000) or in the production of phytohormones such as cytokinins (Timmusk et al., 1999). Biodiversity among the bacteria of the genus Paenibacillus is, however, complex. For example, even with P. polymyxa, 67 strains were isolated from the rhizosphere of maize planted in the Brazilian Cerrado soil (Von der Weid et al., 2000). Thus, Paenibacillus might play a predominant role in plant growth promotion among the plant growth promotion rhizobacteria, but it will be difficult to assign each bacterial strain its specific role under natural conditions. Laboratory conditions may favour growth of Paenibabacilli and therefore may exaggerate the effects. Spores of mycorrhizal fungi are generally surface sterilized prior to the start of an experiment with AMF. Since Paenibacilli are spore forming, they may survive surface sterilization better than other bacteria. Thus, we do not claim, at the present state of experimental findings, that Paenibacillus is a third partner in the AMF–plant symbiosis. The current surprising finding is, however, that the two specific P. validus strains promote fungal growth even to the complete life cycle by providing signal molecules and/or by converting medium compounds to substances that are consumed by the fungus.

It is not an easy task to identify these substances. The analysis of the supernatant revealed that P. validus forms raffinose-like trisaccharides. Plant root exudates have repeatedly been reported to stimulate hyphal growth but not spore germination (Aczón & Ocampo, 1984; Elias and R, 1987). Raffinose is synthesized by many plants, particulary under stress (Minorsky, 2003). It is also known to be a cell wall component and to be utilized by many bacteria, even by Paenibacilli associated with entomopathogenic nematodes (Enright et al., 2003). However, a constituent closely related to raffinose observed in the GC–MS spectrum (Fig. 2) might be more active than raffinose in stimulating hyphal growth. This is suggested from the observation that even the high concentration of 1 mM raffinose was not saturating for maximal growth stimulation (Table 2). We have not yet been able to separate the small amount of trisaccharide from the 20 mM sucrose initially present in the medium. Since raffinose was not able to induce the formation of new spores on the plates, other substances supporting the full life cycle of the fungus must be excreted by P. validus. Raffinose and the other oligosaccharides tested generally stimulated hyphal growth on plates, but not specifically hyphal branching as the recently identified hyphal branching factor (Akiyama et al., 2005).

As in the Rhizobium–legume symbiosis, many compounds such as flavonoids, carbohydrates, CO2, hormones, mono- and disaccharides, lipo-oligosaccharides and others have been discussed as possible factors in the AMF symbiosis (reviewed by Azcón-Aguilar et al., 1998; Bago & Bécard, 2002; Rillig, 2004). The SSH clone libraries, unfortunately, do not provide direct clues as to which factors from the plant and fungal sides are primarily responsible for the active symbiosis. Already in the first cDNA library of an AMF in the symbiotic state, three-fourth of the expression sequence tags showed no or low similarities to known genes (Sawaki & Saito, 2001). In our hands also, SSH libraries of G. intraradices grown with Ri T-DNA transformed carrot roots and affected by either heavy metal (Ouziad et al., 2005a) or salt stress (Ouziad et al., 2005b) did not reveal differentially expressed genes with a priori evident functions. The same was the case with the currently constructed SSH library with hyphae from G. intraradices incubated either with or without P. validus. The demonstration of the differentially expressed O-linked N-acetylglucosamine transferase gene may be taken as faintly indicating that compounds like lipo-oligosacharides may serve as signal molecules as in the Rhizobium symbiosis. Several genes with products involved in signal transduction were detected by the reverse Northern analyses, indicating that the interaction between G. intraradices and P. validus requires extensive changes in the metabolism of the fungal partner.

In summary, the present finding that mycorrhizal fungal growth is not strictly dependent on plants but that P. validus isolates can substitute the plant is a step forward to understand the interrelationship between the symbiotic partners. However, the complexity of the AMF-plant symbiosis with many factors apparently involved from both sides warrants further extensive studies.


This work was kindly supported by a grant from the Deutsche Forschungsgemeinschaft. The authors are indebted to Dr M. Geoffrey Yates of Lewes, Sussex, UK, for critically reading the manuscript and for improving the English.