Nod factors and a diffusible factor from arbuscular mycorrhizal fungi stimulate lateral root formation in Medicago truncatula via the DMI1/DMI2 signalling pathway


  • Boglárka Oláh,

    1. LIPM UMR 2594/441 CNRS/INRA – BP 52627, 31326 Castanet-Tolosan Cedex, France, and
    2. SCSV UMR 5546 CNRS/Université Paul Sabatier – BP 17, 31326 Castanet-Tolosan Cedex, France
    Search for more papers by this author
    • Present address: Institute of Genetics, Biological Research Centre HAS, PO Box 521, H-6701 Szeged, Hungary.

  • Christian Brière,

    1. SCSV UMR 5546 CNRS/Université Paul Sabatier – BP 17, 31326 Castanet-Tolosan Cedex, France
    Search for more papers by this author
  • Guillaume Bécard,

    1. SCSV UMR 5546 CNRS/Université Paul Sabatier – BP 17, 31326 Castanet-Tolosan Cedex, France
    Search for more papers by this author
  • Jean Dénarié,

    1. LIPM UMR 2594/441 CNRS/INRA – BP 52627, 31326 Castanet-Tolosan Cedex, France, and
    Search for more papers by this author
  • Clare Gough

    Corresponding author
    1. LIPM UMR 2594/441 CNRS/INRA – BP 52627, 31326 Castanet-Tolosan Cedex, France, and
      (fax +33 561285061; e-mail
    Search for more papers by this author

(fax +33 561285061; e-mail


Legumes form two different types of intracellular root symbioses, with fungi and bacteria, resulting in arbuscular mycorrhiza and nitrogen-fixing nodules, respectively. Rhizobial signalling molecules, called Nod factors, play a key role in establishing the rhizobium–legume association and genes have been identified in Medicago truncatula that control a Nod factor signalling pathway leading to nodulation. Three of these genes, the so-called DMI1, DMI2 and DMI3 genes, are also required for formation of mycorrhiza, indicating that the symbiotic pathways activated by both the bacterial and the fungal symbionts share common steps. To analyse possible cross-talk between these pathways we have studied the effect of treatment with Nod factors on mycorrhization in M. truncatula. We show that Nod factors increase mycorrhizal colonization and stimulate lateral root formation. The stimulation of lateral root formation by Nod factors requires both the same structural features of Nod factors and the same plant genes (NFP, DMI1, DMI2, DMI3 and NSP1) that are required for other Nod factor-induced symbiotic responses such as early nodulin gene induction and cortical cell division. A diffusible factor from arbuscular mycorrhizal fungi was also found to stimulate lateral root formation, while three root pathogens did not have the same effect. Lateral root formation induced by fungal signal(s) was found to require the DMI1 and DMI2 genes, but not DMI3. The idea that this diffusible fungal factor might correspond to a previously hypothesized mycorrhizal signal, the ‘Myc factor’, is discussed.


Legumes are able to establish two types of very different endosymbiotic associations. Like most terrestrial plants they form arbuscular mycorrhiza (AM) with fungi of the group Glomeromycota. Arbuscular mycorrhizal fungi are able to transfer rare or poorly soluble nutrients such as phosphorus, copper and zinc from the soil to the plant, which in turn provides carbohydrate to the fungus (Gianinazzi-Pearson, 1996; Harrison, 2005). In addition, legumes have the specific ability to form nodules, which are specific organs that host nitrogen-fixing bacteria, collectively called rhizobia. Rhizobia provide assimilatable nitrogen to the plant and the plant provides the rhizobia with reduced carbon. These nutrient exchanges may be of critical importance when soil fertility and water availability are low, conditions that severely limit agricultural production in many parts of the world (Vance, 2001).

The genetic and molecular dissections of the rhizobium–legume symbiosis have shown that during early stages of the interaction the two partners exchange molecular signals. The host plant secretes phenolic compounds, essentially flavonoids, that activate the transcription of bacterial symbiotic genes, the nodulation (nod) genes, which in turn specify the synthesis of lipochito-oligosaccharidic signals, the Nod factors (NFs) (Dénariéet al., 1996; Schultze and Kondorosi, 1998). Nod factors trigger a number of early symbiotic responses in the plant root such as ion fluxes, calcium spiking, root hair deformation, early nodulin gene expression and cortical cell division, leading to formation of a nodule primordium (Downie and Walker, 1999; Geurts and Bisseling, 2002). The use of the model legumes Medicago truncatula and Lotus japonicus has allowed the identification of a number of plant genes involved in NF perception and transduction (Ben Amor et al., 2003; Catoira et al., 2000; Endre et al., 2002; Radutoiu et al., 2003; Stracke et al., 2002).

Mycorrhization and nodulation are quite different processes, involving unrelated microbial symbionts, and giving rise to very different anatomical and physiological structures in the host plant root. However, it has been shown that legumes have genes that control both types of symbioses (Duc et al., 1989). The model legumes have been used recently to study these ‘common’ symbiotic genes, and three have been identified in M. truncatula, the so-called DMI genes (does not make infections) (Catoira et al., 2000). DMI genes control a NF signalling pathway leading to nodulation, but are also required for formation of mycorrhiza, indicating that the symbiotic signalling pathways activated by both the bacterial and the fungal symbionts share common steps. DMI1 and DMI2 are necessary for the induction of a NF-induced calcium spiking response in root hairs and DMI3 acts downstream of the calcium spiking response (Wais et al., 2000).

The addition of exogenous NFs stimulates mycorrhizal colonization in soybean and Lablab purpureus (Xie et al., 1995, 1997), suggesting the existence of cross-talk between the signalling pathways leading to formation of nodules and mycorrhiza. In alfalfa, the use of a split-root system has revealed that NFs provoke a negative autoregulation of both nodulation and mycorrhization, suggesting the existence of common regulatory mechanisms activated by NFs and AM fungal signals (Catford et al., 2003). To further analyse cross-talk between these two pathways we have studied the effect of NF treatment on mycorrhization in M. truncatula. We show that NFs increase mycorrhizal colonization and stimulate lateral root formation (LRF), and that a diffusible factor from AM fungi also elicits LRF. The common DMI1 and DMI2 genes are required for the induction of LRF induced by both bacterial and fungal signals.


Nod factors stimulate formation of mycorrhiza in M. truncatula

The effect of NFs on formation of mycorrhiza was investigated in the model legume M. truncatula. Wild-type M. truncatula seedlings were inoculated with individually transferred, surface-sterilized spores of the AM fungus Gigaspora margarita (Table 1), then grown in sterile conditions in test-tubes with or without NFs in the medium. The medium was sufficiently transparent to allow observation of spore germination and hyphal development along roots, and low in phosphate, to favour fungal colonization of roots, and in nitrogen, to avoid inhibition of the NF signalling pathway. NFs from Sinorhizobium meliloti (Figure 1), the rhizobial symbiont of M. truncatula, were used at 10−9 mol l−1. Six weeks after inoculation with spores, plant roots were stained and microscopically screened for mycorrhizal infection. The level of infection was more limited in this in vitro mycorrhization system compared with classical techniques using older plants grown in sand and vermiculite. Mycorrhized zones corresponded to only a small proportion of the root system and individual infection sites did not appear to have coalesced. Internal colonization was quantified by counting ‘infection units’, defined as zones containing arbuscules and internal hyphal networks.

Table 1.  Fungal isolates and plants
DesignationCharacteristicsPutative function of mutated geneReference/Source
Fungal isolates
(1) Mycorrhizal fungi
 Gigaspora rosea (DAOM 194757)Gigasporaceae Bécard and Fortin (1988)
 Gigaspora margarita (BEG 34)Gigasporaceae BIORIZE, Dijon, France
 Glomus intraradicesGlomaceae PREMIER TECH 2000 Ltée, Rivière-du-Loup, Québec, Canada
(2) Root pathogens
 Fusarium oxysporum f.sp. medicaginisAscomycete CBS 179.29
 Phoma medicaginisvar.pinodellaAscomycete CBS 110-32
 Aphanomyces euteichesDrechs.Oomycete ATCC 201684
 Medicago truncatula
  Jemalong (A17)Wild type, Nod+/Myc+ Penmetsa and Cook (1997)
  C31 (nfp)Nod/Myc+ mutantLysM-RLK: possible Nod factor receptorBen Amor et al. (2003)
  Y6 (dmi1)Nod/Myc mutantPossible ion-channelAnéet al. (2004)
  TR25 (dmi2)Nod/Myc mutantLRR-RLKEndre et al. (2002)
  TRV25 (dmi3)Nod/Myc mutantCa2+/calmodulin-dependent protein kinaseLévy et al. (2004);
Mitra et al. (2004)
  B85 (nsp1)Nod/Myc+ mutantPotential transcription factorSmit et al. (2005)
Figure 1.

Structural characteristics of the chitin oligomers and major Nod factors used in this study.

Infection units were significantly (P < 0.05) more numerous on the roots of seedlings grown in the presence of NFs (Table 2). Early time-course observation of root systems also revealed that lateral roots (LRs) were significantly (P < 0.05) more numerous on seedlings inoculated with G. margarita spores compared with uninoculated control seedlings, and that NFs further significantly (P < 0.05) increased the number of LRs (Figure 2). The root system of each plant was subsequently cut into fragments of 1 cm and infection units were counted in each fragment. This showed that the infection density was also significantly (P < 0.05) higher on the root systems of seedlings grown in the presence of NFs (Table 2).

Table 2.  Effect of Nod factors on root colonization of M. truncatula by G. margarita
 Control+Nod factors
  1. Plants were grown in the presence of G. margarita with or without NFs at 10−9 mol l−1 for 6 weeks. Both the number and density of infection units were significantly increased (P < 0.05) when NFs were present (*). Data represent mean ± SEM from one representative experiment of 15 plants/treatment.

Infection units/plant31.8 ± 7.955.7 ± 5.1*
Density of infection (infection units cm−1)0.801 ± 0.1591.24 ± 0.135*
Figure 2.

Effect of G. margarita and Nod factors on lateral root formation in M. truncatula.
Time scale of LRF during a 10-day period in control plants, in plants co-cultured with G. margarita (G.m.) and in the presence of 10−9 mol l−1S. meliloti NFs. Compared with control plants, the effects of G. margarita alone and of G. margarita plus NFs on LRF are significant from day 6 onwards (P < 0.05). Compared with the G. margarita treatment, LRF was significantly stimulated in plants with G. margarita plus NFs at days 6, 7 and 9 (P < 0.05). Bars show SEM, results of one representative experiment of 15–19 plants/treatment.

Specific Nod factors stimulate lateral root formation and root growth

To determine whether the NF stimulation of LRF in seedlings inoculated with G. margarita spores is due to an indirect effect via activation of the plant–mycorrhizal fungus interaction or to a direct effect on root development, we studied the effect of purified NFs on root systems of seedlings grown aseptically, in the absence of the fungal symbiont. To facilitate visualization of roots, M. truncatula seedlings were grown on plates. S. meliloti NFs were added to the medium at final concentrations of 10−7, 10−8, 10−9, 10−10 and 10−11 mol l−1.

S. meliloti NFs very significantly (P < 0.01) stimulated LRF at concentrations of 10−7, 10−8 and 10−9 mol l−1 (Figure 3A and Figure S1 for time-course results). The increase in LRs was in the range of 35–40%. Dry weight measurements at 18 days showed that the presence of NFs at 10−7 and 10−8 mol l−1 resulted in significant increases in root dry weight (Figure 3B).

Figure 3.

Effect of Nod factors on root development and growth of M. truncatula.
Histograms show the effect of different concentrations of S. meliloti NFs on LRF (A) and on the dry weight of root systems (B). (A) a, control; b, NF 10−11 mol l−1; c, NF 10−10 mol l−1; d, NF 10−9 mol l−1; e, NF 10−8 mol l−1; f, NF 10−7 mol l−1. (B) a, control; b, NF 10−9 mol l−1; c, NF 10−8 mol l−1; d, NF 10−7 mol l−1. For (A) and (B) data are from 30 plants/treatment (one representative experiment) observed at days 7, 8, 10 and 12. In (A) means are presented as percentages of the control. Statistical comparisons made by analysis of variance of all cumulated time-course data. (*) and (**) indicate significant (P < 0.05) and highly significant (P < 0.01) differences, respectively.
(C) Time-course experiment of the effect on LRF of NFs at 10−9 mol l−1 from three rhizobial species. One representative experiment of 45 plants/treatment, scored double blindly. Statistical analysis of time points 5–10 days, showed that only the S. meliloti NF treatment was significantly different (P < 0.001) from the control. Bars show SEM. S. meliloti, Sinorhizobium meliloti; S. fredii, Sinorhizobium fredii; R.leg.viciae, Rhizobium leguminosarum bv. viciae.

To investigate the structural features required for eliciting this root-branching response, three differently substituted chitin tetramers (Figure 1) were tested at 10−7 and 10−9 mol l−1 on M. truncatula roots: (i) the commercial chitotetraose with four GlcNAc residues, (ii) a derivative of this chitin tetramer N-deacetylated at the non-reducing end, and (iii) a derivative N-deacetylated at the non-reducing end and O-sulphated on the C6 of the reducing GlcNAc residue. The three compounds provoked a slight, but not significant, decrease in the number of LRs at both concentrations. No significant difference could be found between the activities of these compounds (data not shown). These results strongly indicate that the presence of an N-acyl chain on the chitin oligomer backbone is required to elicit LRF in our assay.

To determine whether the elicitation of the LRF response by NFs is specific, we studied the effect of NFs (at 10−9 mol l−1) prepared from two other rhizobial species, Sinorhizobium fredii and Rhizobium leguminosarum bv. viciae (Figure 1), symbionts of soybean and pea, respectively, that cannot nodulate Medicago sp. Only the cognate S. meliloti NFs induced numbers of LRs significantly (P < 0.001) different from control plants (Figure 3C). These data indicate that NF structural requirements are similar for inducing both LRF and symbiotic responses. Moreover, since LRF appears to be stimulated specifically by cognate NFs, this response could also be considered a symbiosis-associated response.

Genetic analysis of the Nod factor signalling pathway leading to stimulation of lateral root formation

Genetic studies of M. truncatula have identified genes that control a NF-activated signal transduction pathway leading to symbiotic responses and nodulation (Ben Amor et al., 2003; Catoira et al., 2000). Phenotypic comparison of the corresponding mutants suggests the following intervention order for these genes: NFP acts upstream of DMI1 and DMI2, and all three genes are required for calcium spiking in root hairs. DMI3 acts downstream of calcium spiking, followed by NSP1 (Ben Amor et al., 2003; Wais et al., 2000). While all five genes are required for nodulation, only DMI1, DMI2 and DMI3 are required for mycorrhization.

To determine whether the signalling pathway leading to a NF-induced modification of root development shares steps with the NF signalling pathway leading to symbiotic responses and nodule formation, we compared the effect of S. meliloti NFs on LRF in seedlings of wild-type M. truncatula and representative mutants in the NFP, DMI1, DMI2, DMI3 and NSP1 nodulation genes (Table 1). Seedlings were grown on plates in the presence of S. meliloti NFs at 10−7 and 10−9 mol l−1.

At both concentrations NFs significantly (P < 0.01) stimulated LRF in the wild-type plants, but no significant LRF stimulation could be detected in any of the tested mutants (Figure 4A–C for 10−9 mol l−1 and Figure S2 for 10−7 mol l−1). These results indicate that both types of NF-induced response, previously described symbiotic responses and the modification of root development, share the same NFPDMINSP signalling pathway, and reinforce the idea that the NF-induction of LRF might also be a symbiosis-associated response.

Figure 4.

Genetic analysis of the Nod factor-activated signalling pathway leading to stimulation of lateral root formation.
(A) Histogram showing the effect of NFs (10−9 mol l−1; grey columns) on LRF of M. truncatula wild-type and mutant plants. Statistical comparisons were made by analysis of variance with repeated measures of data for days 5, 6, 7, 8 and 10. For each plant genotype, data from at least two independent experiments with 30 repetitions (plants) per treatment in each were used. Means are presented as percentage of the control value, (**) indicates a highly significant (P < 0.01) difference. The time scale of NF effects is shown by data from one representative experiment for wild-type (B) and nfp mutant plants (B), with 30 plants/treatment. Bars show SEM.

A diffusible AM fungal factor stimulates lateral root formation

To study whether AM fungi other than G. margarita can stimulate LRF in M. truncatula, we tested a close and a distant relative of G. margarita, Gigaspora rosea and Glomus intraradices, respectively. Spores of G. rosea were transferred individually and those of Gl. intraradices were transferred as a water suspension. Germinating spores of both species significantly (P < 0.05) stimulated LRF (Figure 5A,B), indicating that this ability is a general feature in AM fungi. Moreover, LRF stimulation was observed at a very early stage (4–5 days after inoculation), at which time no direct contact of germinating spores and plant roots could be detected, indicating that the stimulation might be mediated by one or more diffusible compounds.

Figure 5.

Effect of two AM fungi (G. rosea and Gl. intraradices), and of a diffusible factor from G. margarita on lateral root formation.
Curves show the time scale of the effect of G. rosea (A), Gl. intraradices (B) and the effect of G. margarita on LRF without (C) and in the presence of a physical barrier in the form of a membrane (D). The stimulation of LRF is statistically significant (P < 0.05) from day 6 onwards (A and C), from day 5 onwards (B), and from day 12 onwards (D). Data from one representative experiment with 30 plants/treatment. Bars show SEM.

When seedlings of M. truncatula are separated from AM fungi by a membrane, we had previously shown that AM fungi produce a diffusible factor that can be perceived by roots and induce expression of the symbiotic plant gene MtENOD11 (Kosuta et al., 2003). To assess whether G. margarita stimulates LRF by a diffusible factor, germinating spores were co-cultivated with wild-type M. truncatula seedlings, either separated by a cellophane membrane or in conditions allowing direct physical contact between the two partners, as previously described (Kosuta et al., 2003). In control plates, seedlings were grown in the absence of the fungal partner, either with or without a membrane.

When separated by a cellophane membrane germinating spores were found to significantly (P < 0.05) stimulate LRF, while the membrane control had no effect on the number of LRs (Figure 5C,D). This stimulation was slightly delayed by comparison with membrane-free co-cultures, similar to results of previous experiments describing the ability of a diffusible factor to induce expression of MtENOD11 (Kosuta et al., 2003). We can conclude that germinating spores of G. margarita produce a diffusible factor that stimulates LRF in M. truncatula.

To determine whether this ability to stimulate LRF was general among root-associated fungi, M. truncatula seedlings were grown in the presence of two fungi (Phoma medicaginis and Fusarium oxysporum), not belonging to Glomeromyceta, and an Oomycete (Aphanomyces euteiches). The three species investigated are all pathogens of pea and alfalfa, and A. euteiches has been shown to be pathogenic on M. truncatula roots (Nyamsuren et al., 2003). As the saprophytic growth ability of these three pathogenic organisms is potentially very high, inoculant sizes were chosen so as to prevent excessive mycelial development during the early period of each experiment. As a result, the overall growth of seedlings was not affected, although various degrees of browning were observed on roots after several days.

No significant stimulation of LRF could be observed when plants were grown in the presence of P. medicaginis or F. oxysporum and the same was true for A. euteiches, except at day 10 when a significant (P < 0.05) stimulation of LRF was seen (Figure S3). Since this is much later than the stimulation caused by AM fungi, we conclude that the ability to rapidly stimulate LRF is not a general property of root-associated fungi.

Comparison of the effect of AM fungi, Nod factors and auxin on root development

Many fungal species produce auxin or auxin-like compounds (Griffin, 1994). To determine whether the diffusible fungal factor responsible for inducing LRF is likely to be auxin or an auxin-like compound(s) we compared the effect of germinating spores of G. margarita on root development of M. truncatula to the effect of indole acetic acid (IAA) or NFs at several concentrations (10−7, 10−8 and 10−9 mol l−1). All treatments were performed in plates, and, in addition to LRF, the length and thickness of the primary roots were measured, the latter by scanner image analysis.

Whereas IAA at 10−9 and 10−8 mol l−1 did not elicit detectable effects on LRF, at 10−7 mol l−1 it elicited a transient increase in both the number of LRs and the thickness of the primary root which could only be observed until the fifth day (data not shown). The presence of IAA at 10−7 and 10−8 mol l−1 significantly (P < 0.01 for 10−7 mol l−1) decreased the length of the primary root (Figure 6A,C, data not shown). In contrast, both NFs and germinating spores of AM fungi (G. margarita or Gl. intraradices) resulted in stimulation of LRF without any detectable inhibition of primary root elongation (Figure 6A,D–F). In some experiments even, the length of the primary roots was slightly increased by NFs. No change in the geotropism of roots could be observed in the presence of the fungus or NFs, while IAA induced a change in positive root geotropism (Figure 6C).

Figure 6.

Effect of Nod factors, auxin, G. margarita and Gl. intraradices on root development of M. truncatula.
The histogram (A) shows the effect of S. meliloti NFs and auxin (IAA) on the length of primary roots of M. truncatula (a, control; b, NF 10−9 mol l−1; c, NF 10−7 mol l−1; d, IAA 10−9 mol l−1; e, IAA 10−7 mol l−1). Bars show SEM, (**) indicates a highly significant (P < 0.01) difference. Photos show the root systems of 8-day-old M. truncatula control plants (B), and plants in the presence of 10−7 mol l−1 IAA (C), 10−7 mol l−1 Nod factors (D), germinating spores of G. margarita (E) and Gl. intraradices (F). Arrows indicate the positions of spores in the Petri dishes.

We can conclude from these results that the diffusible AM fungal factor(s) is unlikely to be an auxin-like compound. Interestingly, AM fungi and NFs have similar effects on root development of M. truncatula: a stimulation of LRF associated neither with the inhibition of primary root elongation nor with a change in root geotropism.

Genetic analysis of the signalling pathway used by the diffusible AM fungal factor

Since NFs and diffusible AM fungal factor(s) elicit a similar LRF response, this suggests that the corresponding pathways might share one or more common steps. We therefore studied the effect of germinating spores of G. margarita on LRF in M. truncatula mutants altered in the NFP, DMI1, DMI2, DMI3 and NSP1 genes (Table 1). Assays were carried out in test-tubes and data are presented in Figure 7.

Figure 7.

Genetic analysis of the arbuscular mycorrhiza fungal-activated signalling pathway leading to stimulation of lateral root formation.
Histogram shows the effect of G. margarita (grey columns) on LRF of M. truncatula wild-type and mutant plants (A). Means are presented as percentage of the control value, (*) indicates a significant (P < 0.05) difference. Statistical comparisons by analysis of variance with repeated measures of data for days 5, 6, 7, 8 and 10. For each plant genotype, data from at least two independent experiments with 30 repetitions (plants) per treatment in each were used. The time scale of the effect of G. margarita for wild-type (B) and nfp mutant plants (C), from one representative experiment with 30 plants/treatment. Bars show SEM.

Fungal spores elicited a statistically significant (P < 0.05), increase in LRF in nfp mutant seedlings, unlike what was observed in the presence of NFs. In contrast, a fungal presence did not elicit any detectable stimulation of LRF in dmi1 and dmi2 mutants, indicating that DMI1 and DMI2 are common to the bacterial and fungal signalling pathways leading to this root developmental response. Germinating spores did elicit statistically significant (P < 0.05) LRF in dmi3 and nsp1 mutants.

In NF-induced LRF experiments, we previously observed a correlation between the defects exhibited by nfp, dmi1, dmi2, dmi3 and nsp1 mutants for nodulation and LRF phenotypes: all the nodulation-defective mutants were defective for the LRF response. In contrast, in fungal-induced LRF experiments the defects in the LRF response were observed in two of the three mycorrhization-defective (Myc) mutants, dmi1 and dmi2, but not in dmi3 mutants. However, whereas the mycorrhizal phenotype (Myc+) of the M. truncatula nfp mutant C31 was established with G. margarita (Ben Amor et al., 2003), the Myc phenotype of Y6 (dmi1), TR25 (dmi2) and TRV25 (dmi3) mutants had been previously established with Gl. intraradices or Glomus mosseae, but had not been tested with G. margarita (Sagan and Morandi, 1998; Sagan et al., 1995). We thus microscopically examined the mycorrhizal behaviour of these mutants with G. margarita 6 and 7 weeks after inoculation. In the presence of the three dmi mutants, G. margarita germinating hyphae were well developed and appressoria were formed normally on root surfaces, but no hyphae or arbuscules could be detected inside roots, indicating that the Myc phenotype of these mutants is also valid with G. margarita.

Taken together these results indicate that the fungal and bacterial signalling pathways, leading to the LRF response, share two common steps controlled by the DMI1 and DMI2 genes. Interestingly, the DMI3 gene that is required for mycorrhization seems not to be required for the AM fungal-induced LRF response.


In this study we have used the model legume M. truncatula to analyse cross-talk between the signalling pathways leading to formation of mycorrhiza and root nodules that are activated by AM fungi and rhizobia, respectively. Growing seedlings of M. truncatula in the presence of NFs resulted in the stimulation of root colonization by the AM fungus G. margarita, and in an increase in root branching. Since LRs are targets for AM fungal infection, the stimulation of LRF by NFs might be one of the mechanisms by which the nodulation signal stimulates formation of mycorrhiza. Since the infection site density was also stimulated, another mechanism, directly involved in the response of root tissues to AM fungal infection, might also be involved. A diffusible factor secreted by the AM fungus G. margarita also stimulated LRF. Genetic analysis showed that this LRF developmental response induced by bacterial and fungal symbiotic signals requires the DMI1 and DMI2‘common’ symbiotic genes.

Nod factors, diffusible AM fungal factors and root development

It was previously reported by Souleimanov et al. (2002) that the NFs of Bradyrhizobium japonicum can stimulate development of soybean root systems by increasing their total lengths, surface areas and dry weights, suggesting that these molecular signals may have a general plant growth-regulating effect in addition to their symbiotic roles in nodulation. We have confirmed these observations in M. truncatula, since NFs of S. meliloti, the M. truncatula symbiont, clearly stimulated LRF and the root system as a whole, resulting in an increase in root dry weight. Furthermore, we have shown that this response appears to depend on the specific lipochito-oligosaccharidic structure of cognate NFs, since no effect could be detected with chitin tetramers or with NFs from two rhizobial species unable to nodulate M. truncatula. The structural requirements for stimulation of root branching and for nodulation responses therefore appear to be the same.

No significant NF stimulation of LRF could be observed with plants carrying mutations in any of five genes (NFP, DMI1, DMI2, DMI3, NSP1) that control a NF-activated signal transduction pathway leading to symbiotic responses, such as early nodulin gene expression and cortical cell division (Ben Amor et al., 2003; Catoira et al., 2000). Nod factor-regulated root development and NF-induced symbiotic responses therefore share the same NFPDMINSP signalling pathway, suggesting that this stimulation of LRF might also be a symbiosis-associated response.

It was previously reported that plant colonization by AM fungi is associated with stimulation of development of the root system, mainly by increasing the formation of new LRs (Berta et al., 2002). These observations were made primarily on well-mycorrhized roots at a late stage of the symbiosis. Here, we found that three AM fungal species (G. margarita, G. rosea and Gl. intraradices) could increase the number of lateral roots as soon as the fourth or fifth day after inoculation, a very early stage of the interaction before fungal hyphae had established contact with roots. In addition, using a membrane-separated dual-culture system, we showed that a diffusible factor produced by G. margarita is responsible for stimulation of root development. These data therefore suggest that secretion of diffusible compounds capable of stimulating LRF is a common property of AM fungi.

The type of root developmental response elicited by the mycorrhizal fungal factor(s) is similar to the effect induced by NFs, with a stimulation of LRF without inhibition of primary root elongation and with no modification of root geotropism. This type of response is quite different from that observed in the presence of the plant hormone auxin, where the appearance of new lateral roots was associated with an inhibition of primary root elongation and with an inhibitory effect on the positive geotropism of primary roots. These results show that auxin has the same effect on roots of M. truncatula as on roots of diverse plant species (Casimiro et al., 2003; Leyser, 2002), and that NFs and the mycorrhizal fungal factor(s) appear to have a novel regulatory effect on root development.

Non-legume nodules clearly have the structures of modified roots, and although legume nodules appear to be anatomically and developmentally distinct from LRs (Hirsch and LaRue, 1997), their organogenesis presents several similarities with the development of LRs: (i) forming post-embryonically near the xylem poles of pre-existing roots; (ii) requiring activation and division of the pericycle adjacent to the inner cortex (Timmers et al., 1999; Yang et al., 1993); (iii) involving local auxin transport in the development of the primordium and in the initiation and differentiation of the vasculature (de Billy et al., 2001; Mathesius et al., 1998). These developmental similarities suggested the hypothesis that during evolution nodule organogenesis might have incorporated some already existing mechanisms involved in LR development (Hirsch and LaRue, 1997; Kistner and Parniske, 2002). The finding that NFs increase the formation of LRs provides new evidence for this hypothesis.

What could be the benefit for the symbiotic partners of the existence of microbial signals that stimulate LRF? The establishment of symbiosis with nitrogen-fixing bacteria or AM fungi is a continuous process, with old roots becoming inactive and new roots being constantly infected. Symbiotic signals such as Nod and AM fungal (Myc) factors might activate plant symbiotic programmes that are not only involved in infection itself but possibly also in the persistent establishment of symbioses, by creating new potential infection sites, e.g. by stimulating the formation of new (lateral) roots that could subsequently be colonized. Symbiotic signals and programmes that allow this continuous generation of infection might have been selected in nature.

Cross-talk and specificities in the Nod and Myc signalling pathways

To explain the NF stimulation of AM fungal colonization we can hypothesize that NFs provoke symbiotic responses related to nodule formation and that these responses favour mycorrhizal colonization. For example, NFs provoke stimulation of LRF and this response provides new targets for mycorrhizal infection. Whether or not NFs increased infection density (number of infection units/root length unit) by increasing LRF or by increasing the sensitivity of roots to mycorrhizal fungal infection, our data suggest an influence of NFs on the expression of the plant mycorrhizal programme. Moreover it is likely that NFs contribute to the regulation of the plant mycorrhization programme via the DMI genes that control common steps in the signalling pathways leading to nodulation and formation of mycorrhiza (Figure 8).

Figure 8.

Working model for the Nod- and Myc-activated signal transduction pathways leading to the stimulation of lateral root formation.
Dark grey: Nod factor-activated pathway; light grey, mycorrhizal factor-activated pathway.

The stimulation of root branching by NFs requires five genes controlling the NF signalling pathway leading to nodulation, that is the NFP and NSP1 genes that are specifically required for nodulation, as well as the three ‘common’ genes, DMI1–3, that are required for both nodulation and mycorrhization. NFP is believed to be the first gene of the NF signalling pathway (Ben Amor et al., 2003) and encodes a LysM-receptor-like kinase (J.F. Arrighi, B. Ben Amor and C. Gough, personal communication) and is likely to be a NF receptor. The mycorrhizal factor does not need this putative NF receptor to activate the pathway leading to root branching, suggesting that the fungal and Nod factors have structures sufficiently different to be discriminated by the NFP protein. The cross-talk between the two pathways would not thus be simply due to a structural similarity of the Nod and Myc signals. This was already suggested by the observation that an nfp mutant (and lotus and pea mutants in orthologous genes) is defective for nodulation and not for mycorrhization (Ben Amor et al., 2003; Radutoiu et al., 2003; Walker et al., 2000).

DMI1 and DMI2 seem to be key components of the cross-talk. They are required for root branching induced by both bacterial and fungal signals. DMI1 has similarity with an ion channel and DMI2 codes for an LRR-receptor-like kinase (Anéet al., 2004; Endre et al., 2002). Their function in the perception of symbiotic signals is not clear but these two genes are absolutely required for the NF elicitation of the calcium-spiking response in root hairs (Wais et al., 2000). Interestingly, the DMI3 gene, which acts downstream of the calcium-spiking response, is required for the root-branching response when induced by NFs but not when induced by the AM fungal factor. DMI3 encodes a calcium/calmodulin-dependent protein kinase (Lévy et al., 2004; Mitra et al., 2004) that might recognize the calcium-spiking response elicited by NFs. That this calcium-sensing protein is required for root branching when it is induced by NFs and not by mycorrhizal factors indicates that the mycorrhizal and nodulation pathways leading to root branching diverge before DMI3 (Figure 8). The Myc-activated pathway might thus shortcut this calcium-sensitive step. The fact that NSP1, which acts downstream of the DMI genes in a nodulation-specific pathway, is required for the NF-induced but not for Myc-induced root branching is consistent with the hypothesis that the Nod and Myc pathways leading to LRF diverge after the DMI1/2 genes. It is worth noting that in all mutant seedlings for which the treatment by either NFs or AM fungi does not elicit a statistically significant stimulation of LRF, there are nonetheless small, consistent, but non-significant increases. This suggests that, in addition to the DMI pathway, one or more additional pathways exist by which these symbiotic signals might influence root development.

Diffusible mycorrhizal fungal factors

From the genetic analysis of NF signalling in M. truncatula, a hypothesis has been proposed that upstream of the DMI genes, symmetrically to the putative NFP receptor activated by NFs, a mycorrhiza-specific receptor might be activated by a putative ‘Myc factor’ (Catoira et al., 2000; Ben Amor et al., 2003) (Figure 8). Would this Myc factor be a diffusible compound like its bacterial counterpart? The existence of a diffusible mycorrhizal fungal factor that activates the transcription of the symbiotic gene MtENOD11, in a DMI-independent manner, was previously demonstrated (Kosuta et al., 2003). The localization of ENOD11 gene expression induced by NFs and the fungal factor are not identical (epidermal cells in the former case and both epidermal and cortical cells in the latter case), suggesting that the two signals act through different pathways, with the fungal-activated one not involving the DMI proteins. Here we demonstrate the existence of a diffusible fungal factor, which provokes a plant response (LRF) in a DMI1/2-dependent way. The fact that this fungal factor induces a signal transduction pathway involving genes required for mycorrhization is compatible with what is expected for a putative symbiosis-specific Myc factor. MtENOD11 expression associated with a later step of mycorrhizal interaction, the formation of appressoria, has also been studied in M. truncatula. This expression, which requires the DMI2 gene, appears to be linked to the penetration of AM fungal hyphae across the root epidermis (Chabaud et al., 2002).

Does the diffusible fungal factor identified by the LRF bioassay and that identified by Kosuta et al. (2003), correspond to one and the same compound? The observation that the ENOD11 response is independent of the DMI genes, whereas the LRF response is dependent on DMI1/2, might suggest that the two factors are distinct. However, we cannot exclude that the same compound could activate two different pathways, one leading to ENOD11 expression being DMI independent and another one leading to LRF stimulation being DMI1/2 dependent. Biochemical purification and characterization of these factors are now required. Concerning the chemical nature of the AM fungal factor identified with the root-branching assay, our data suggest that it is not a simple chitin tetramer or a non-sulphated NF, fucosylated or not, or an auxin-like compound. By comparing the root-branching responses induced in wild-type and in a dmi1 (or dmi2) mutant of M. truncatula by AM fungal exudates we are currently working on the purification of this (or these) presumably symbiotic specific ‘Myc factor’ compound(s).

Experimental procedures

Plant material

Medicago truncatula plants are listed in Table 1. Seeds were sterilized and germinated as described by Catoira et al. (2000).

AM fungal inoculum

Fungal inocula are listed in Table 1. Spores of G. margarita were obtained either from pot culture with sorghum under greenhouse conditions by BIORIZE (21110 Pluvault, France), or from in vitro mycorrhizal root culture systems (Bécard and Fortin, 1988). Spores were cold-treated at 4°C for 3 weeks, a treatment necessary for germination, before use. Spores produced in pot culture were collected, sterilized and stored as described by Jargeat et al. (2004). To pre-germinate spores, they were placed on M medium (see below) in a 2% CO2 incubator at 28°C for 4–5 days. Gigaspora rosea spores were produced in leek pot culture, surface-sterilized and stored at 4°C according to Bécard and Fortin (1988). Sterile Gl. intraradices spores (PREMIER TECH 2000 Ltée, Rivière-du-Loup, Québec, Canada) were stored in sterile water at 4°C.

Nod factor, chitin oligomer and auxin treatments

Nod factors (NFs) from S. meliloti, S. fredii and R. leguminosarum bv. viciae were purified from bacterial cultures as described by Roche et al. (1991) by F. Maillet and S. Uhlenbroich (LIPM, CNRS-INRA, Toulouse). NF structures were analysed by mass spectrometry by V. Poinsot (LIMRCP CNRS-UPS, Toulouse). To test the effect of NFs, 10 M. truncatula seedlings were grown axenically in 120 × 120 mm Petri dishes on 85 ml M medium (Bécard and Fortin, 1988), with some modifications (MM medium). MM medium was sucrose-free, with 0.8 mm KNO3 instead of 3.2 mm, CaCl2 instead of Ca(NO3)2, and solidified with 4 g l−1 Phytagel (Sigma, St Louis, MO, USA). NFs were incorporated directly into the sterile medium from a stock solution (10−3 mol l−1) of NFs dissolved in 50/50 ethanol/water. Control experiments were always performed with the appropriate amount of ethanol. To keep the root systems in the dark, Petri dishes were partially covered with paper and aluminium foil, propped vertically at an angle of 70°. Seedlings were grown for up to 18 days in a room at 24/20°C with a 16 h photoperiod, and 3200 cd m−2 light intensity. Chitin tetramers (E. Samain, CERMAV-CNRS, Grenoble, France) and auxin (IAA, Sigma) were tested for activity as described above for NF treatment. Lateral root formation (LRF) was monitored every day for up to 2 weeks. All experiments were performed at least three times, with three to five replicates (Petri dishes) per experiment.

Dual cultures and membrane separation of plants and fungi

For dual cultures, plants were grown either in test tubes on slopes of 20 ml of MM medium (one seedling/tube) as described in Ben Amor et al. (2003), or on Petri dishes as described above. In Petri dishes, AM fungal inocula consisted of 50 (G. margarita), 100 (G. rosea) or approximately 500 (Gl. intraradices) surface-sterilized spores. Spores were inserted into the medium, at the same time as the seedlings were put on plates, in a line 5 cm from the bottom of the Petri dish. In test tubes, five spores (G. margarita) were put in the bottom of each slope near the seedling roots. Spores of Gigaspora species were used without (experiments presented in Figure 2) or with (all other experiments) pre-germination. Glomus intraradices spores were always used without pre-germination. Arbuscular mycorrhiza fungal growth was observed using a binocular microscope.

For membrane-separated co-cultures, fungal inocula were inserted into a 120 × 120 mm Petri dish containing 70 ml of MM medium. To physically separate fungal inoculum from roots, a 120 × 120 mm cellophane membrane (Couvre-Confitures, Hutchinson, 45120 Chalette-sur-Loing, France), soaked in distilled water for 30 min, and autoclaved in distilled water before use, was laid on the medium. Twenty millilitres of MM medium was then layered on the membrane, and finally 10 seedlings were put on the medium.

For all systems, plants and fungi were co-cultured for 10–18 days, and LRF was scored every day. All dual cultures were performed at least three times, with three to four (Petri dishes) or 20–40 (test-tubes) replicates/experiment.

Observation of fungal root infection

To observe fungal colonization or to verify that the fungus had not crossed the membrane and contacted roots in membrane-separated co-cultures, whole root systems were cleared with 10% KOH for 1 h at 90°C, thoroughly rinsed with distilled water, and stained with 0.05% Chlorazol Black for 2 h at 90°C. Roots were then de-stained overnight in 50% glycerol and examined by bright-field microscopy. To calculate the average number of infection units/plant and the density of infection (number of infection units/root section), root systems were cut into fragments of 1 cm and infection units were counted in each fragment. The mycorrhizal phenotype of plants was studied by staining roots for the presence of intraradical colonization 6 and 7 weeks after inoculation and co-culture with five G. margarita spores/plant.

Inoculations with root pathogens

Inocula of Phoma medicaginis, Fusarium oxysporum and Aphanomyces euteiches were obtained from fresh in vitro cultures and inoculations were performed on M. truncatula wild-type seedlings growing on MM medium in 120 × 120 mm Petri dishes (six seedlings/dish). For P. medicaginis, agar plugs (4 mm diameter) were put in between and 2.5 cm below seeds of 1-day-old seedlings. For F. oxysporum, 50 μl of a conidial suspension (106 conidia ml−1 in water) was applied to root tips of 4-day-old seedlings. For A. euteiches, three spots of 3 μl each of an oospore suspension (2.5 × 105 oospores ml−1 in 0.01% Tween) were applied at regular distances on roots of 4-day-old seedlings. Root systems were kept in the dark in a growth room, as described above. Lateral root formation was scored every day for 10–12 days. All experiments were performed twice, with seven or eight replicates (Petri dishes) per experiment.

Statistical and image analysis of data

All statistical analyses were made with SPSS statistical software (SPSS Inc., Chicago, IL, USA). Homogeneity of variance between groups was tested by the Levene test and normality of the residues was verified by the Kolmogorov–Smirnov test. In Figure 3(A), the effect of NF concentration on the number of LRs was tested by a one-way analysis of variance with repeated measures (days 5, 6, 7, 8 and 10). Data collected from all these time points were used to make use of information provided by the time scale from a single experiment. In Figures 4(A) and 7(A), effects of NFs or G. margarita on the number of LRs were tested by analysis of variance of data from several (at least two) independent experiments, for each plant type (wild-type and mutants) separately. In order to take account of variations between experiments, a two-way factorial design (one factor for the presence of NFs or G. margarita, and one factor for experiments) with repeated measures (days 5, 6, 7, 8 and 10) was used. After analysis of variance, marginal means and their corresponding standard errors were calculated. Data were then standardized by dividing by the mean of the control value and expressed in the figures as a percentage of the control. This allowed the differences in NF responses between plant types to be highlighted. The effects of pathogens on LRF were tested by two-way analysis of variance, with repeated measures (days 5–10), of data from two independent experiments. Data in Figures 2 and 5 were analysed by the t-test, for each day. When we have chosen a ‘representative experiment’ to illustrate a time-course LRF response, this means that more than one experiment of this type showed a statistically significant difference. Image analysis of root thickness was made using IMAGE-PRO PLUS (Media Cybernetics, Silver Spring, MD, USA).


We are extremely grateful to J.-M. Prospéri (INRA, Montpellier), for providing wild-type seeds of M. trunctula. We are also grateful to E. Samain (CERMAV-CNRS, Grenoble) for providing chitin oligomers, F. Maillet and S. Uhlenbroich (LIPM CNRS-INRA, Toulouse) and V. Poinsot (LIMRCP CNRS-UPS Toulouse) for providing purified NFs, D. Girard (SCSV CNRS-University Toulouse 3) for spores of G. rosea, Premier Tech (Rivière-du-Loup, Canada) for spores of Gl. intraradices and A. Jauneau (IFR 40, Toulouse) for image analyses. Inocula of F. oxysporum and A. euteiches were very kindly provided by M. Rickauer (Ecole Nationale Supérieure d'Agronomie, Toulouse) and C. Jacquet (UMR CNRS-UPS 5546, Toulouse), respectively. Boglárka Oláh was supported by post-doctoral fellowships of the IFR40 (grant from the French Research Ministry) and of INRA (fellowship for international cooperation).