Targeted inoculation of Medicago truncatula in vitro root cultures reveals MtENOD11 expression during early stages of infection by arbuscular mycorrhizal fungi

Authors

  • M. Chabaud,

    Corresponding author
    1. Laboratoire de Biologie Moléculaire des Relations Plantes-Microorganismes, UMR215 INRA-CNRS, BP 27, 31326 Castanet Tolosan Cedex, France;
      Author for correspondence: M. Chabaud Tel: +33 (0)5 61 28 53 24 Fax: +33 (0)5 61 28 50 61 Email: mchabaud@toulouse.inra.fr
    Search for more papers by this author
  • C. Venard,

    1. Laboratoire de Biologie Moléculaire des Relations Plantes-Microorganismes, UMR215 INRA-CNRS, BP 27, 31326 Castanet Tolosan Cedex, France;
    Search for more papers by this author
  • A. Defaux-Petras,

    1. Laboratoire de Biologie Moléculaire des Relations Plantes-Microorganismes, UMR215 INRA-CNRS, BP 27, 31326 Castanet Tolosan Cedex, France;
    Search for more papers by this author
  • G. Bécard,

    1. Signaux et Messages Cellulaires Chez les Végétaux, UMR 5546 CNRS-Université P. Sabatier, Pôle de Biotechnologie Végétale, BP 17, 31326 Castanet Tolosan Cedex, France
    Search for more papers by this author
  • D. G. Barker

    1. Laboratoire de Biologie Moléculaire des Relations Plantes-Microorganismes, UMR215 INRA-CNRS, BP 27, 31326 Castanet Tolosan Cedex, France;
    Search for more papers by this author

Author for correspondence: M. Chabaud Tel: +33 (0)5 61 28 53 24 Fax: +33 (0)5 61 28 50 61 Email: mchabaud@toulouse.inra.fr

Summary

  • • An in vitro targeted inoculation technique has been developed for studying the earliest stages of arbuscular endomycorrhizal (AM) infection of Medicago truncatula roots, and in particular the spatio-temporal expression of the early nodulin gene MtENOD11.
  • •  Agrobacterium rhizogenes transformed root explants were derived either from Myc +M. truncatula or from the infection-defective Myc mutant TR26 ( dmi2–2 ), both expressing the pMtENOD11-gusA fusion. The normal positive geotropism of these roots, coupled with the negative geotropism of Gigaspora germ tubes allowed oriented growth of the two symbiotic partners, facilitating the identification of initial fungal/root contacts.
  • • Early infection events at the stage of appressoria and/or internal hyphae could be observed for over 50% of the inoculated explants, revealing that MtENOD11 is expressed transiently in both epidermal and cortical cells at sites of hyphal penetration in Myc + roots, but not in epidermal cells in contact with appressoria in Myc roots.
  • • We propose that a direct link exists between MtENOD11 gene expression and cellular events required for fungal penetration, thereby extending analogies between rhizobial and AM host root infection processes.

Introduction

The arbuscular mycorrhizal (AM) symbiosis is a wide-spread and agronomically important plant–fungal association, involving the majority (80%) of land plants and a restricted number of obligate biotrophic soil fungi belonging to the Glomeromycota, such as Glomus or Gigaspora (Smith & Read, 1997; Schussler et al., 2001). After spore germination, fungal hyphae penetrate host root tissues, triggering the formation of differentiated endosymbiotic structures (called arbuscules) in inner cortical cells. Subsequently, internal fungal colonization of the host root cortex takes place with the formation of additional arbuscules, and the fungus develops an extensive extraradical hyphal network. Arbuscules are believed to be the major site of metabolite exchange between the two organisms, providing the plant with important nutrients such as phosphate. In return the plant furnishes reduced carbon in the form of photosynthates to the fungal partner, and in addition provides the appropriate environment for the fungus to complete its life cycle. Despite the importance of the AM symbiosis relatively little is known about the molecular and cellular mechanisms involved in the establishment and development of this beneficial plant–fungal association (Harrison, 1999).

Studies of the earliest stages of the AM interaction are not straightforward, due partly to technical problems associated with handling an edaphic obligate biotrophic fungus, and partly to the nonsynchronous and infrequent infection process. Nevertheless, it is now clear that an important early step preceding infection is the ability of the fungus to distinguish between host and nonhost. This can be visualized by the stimulation of hyphal growth and branching in the physical presence of the host plant root (Giovannetti et al., 1993b). Recently, Buée et al. (2000) have partially purified this so-called ‘branching factor’ from Agrobacterium rhizogenes-transformed carrot roots, and shown that this activity is not present in exudates of nonhost plants such as the Brassicaceae. Following hyphal branching, appressoria are then formed at the root epidermis, and are the sites of hyphal penetration inside the root. These infection hyphae, after traversing the outer layers of the cortex, penetrate inner cortical cells, and then differentiate into arbuscules.

Over the last decade, genetic studies on a variety of legume species have provided a number of useful mutants for AM studies. In particular, a subgroup of nodulation-defective (Nod) mutants are also refractive to colonization by AM fungi (Myc). In pea (Gollotte et al., 1993), alfalfa (Bradbury et al., 1991), and Medicago truncatula (Sagan et al., 1995; Calantzis et al., 2001) the majority of these Nod/Myc mutants have been classified as Pen since appressoria are formed on the root surface but hyphae do not penetrate the epidermal cell layer. This often leads to deformed or highly branched appressoria. For other mutants hyphae penetrate the epidermis but are then unable to pass into the root cortex (for Lotus: Wegel et al., 1998; Bonfante et al., 2000 and for alfalfa: Bradbury et al., 1991). In the case of M. truncatula, the three Nod/Myc mutants so far characterized are all blocked in a signaling pathway related to Rhizobium Nod factor perception/transduction (Catoira et al., 2000), suggesting that there are common early signaling steps between nodulation and mycorrhization.

In addition to the genetic data, a number of plant genes, specifically expressed during different stages of nodulation, such as MsENOD2 and MsENOD40 (Van Rhijn et al., 1997), PsENOD12 and PsENOD5 (Albrecht et al., 1998), VfENOD5, VfENOD12 and VfLb29 (Frühling et al., 1997), MtENOD11 and MtENOD12 (Journet et al., 2001) are also transcribed during the AM interaction. However, during the earliest stages of the AM interaction (prior to arbuscule formation) plant gene expression is poorly documented at the cellular level, mainly due to the difficulties of synchronizing infection. Thus, while it has been deduced from RT-PCR and Northern analyses that expression of PsENOD12A (Albrecht et al., 1998) and Psam5 (Roussel et al., 2001) correlates with the formation of appressoria and the first internal hyphae, it is not yet known in which plant cells these genes are transcribed. The use of reporter genes such as gusA represents a particularly attractive method for determining spatio-temporal gene expression during the earliest stages of the AM interaction. For example, Blilou et al. (2000) have exploited a lipid transferase promoter-gusA fusion to reveal epidermal expression associated with the presence of fungal appressoria in rice.

In our laboratory we are currently studying the endomycorrhizal association using the model legume M. truncatula (Cook et al., 1997). Previously, we were able to show that the MtENOD11/12 genes are expressed in arbuscule-containing cortical cells of plants colonised by Glomus (Journet et al., 2001). However, the lack of synchronised infection in these experiments prevented us from studying gene expression during the early infection stages. In this article we describe an in vitro system for studying these early infection stages based on targeting M. truncatula Ri T-DNA-transformed roots with Gigaspora hyphae, analogous to that developed previously by Bécard & Fortin (1988) using carrot Ri T-DNA-transformed roots. The choice of Gigaspora over Glomus in this context is due to the negative geotropism of the Gigaspora germ-tube. M. truncatula root cultures have been obtained from both Myc+ and Myc plants containing the pMtENOD11-gusA reporter and the targeted inoculation system has allowed us to analyze at the cellular level expression of the MtENOD11 gene during the earliest stages of the AM interaction. We present data showing that this early nodulin gene is transcribed transiently in epidermal and cortical cells associated with hyphal penetration of the host root.

Materials and Methods

Plant and fungal material, and bacterial strains

The M. truncatula transgenic line L416 (Journet et al., 2001), single-copy and homozygous for the pMtENOD11-gusA transgene is derived from Jemalong A17. The same reporter fusion was introduced into a dmi2–2 Nod/Myc mutant background (TR26 Sagan et al., 1995) by crossing with L416 (Vernoud et al., 1999). F3 homozygous progeny derived from this cross are referred to as F3 TR26×L416. Agrobacterium rhizogenes-based transformation was used to generate Myc+ and Myc root clones derived from L416 and F3 TR26×L416. The A. rhizogenes strain used was Arqua1, a Smr-derivative of A4T (Quandt et al., 1993).

Spores of Gigaspora rosea Becker and Hall (DAOM 194757), formerly called G. margarita (Lanfranco et al., 2001), were produced in vitro according to Diop et al. (1992). Spores of G. gigantea were produced using pot-grown leek plants according to Buée et al. (2000).

Ri T-DNA-transformed roots of M. truncatula; establishing a root culture for targeted inoculation

Inoculation of M. truncatula by A. rhizogenes, excision of Ri T-DNA-transformed roots, bacterial decontamination and root propagation were performed according to Boisson-Dernier et al. (2001). For both lines (L416 and F3 TR26×L416), 10 transformed root clones were initially propagated in vitro. A single representative clone with both a typical in vitro growth pattern and characteristic non symbiotic pMtENOD11-gusA expression (root cap and root lateral initiation sites; Journet et al., 2001), was selected for each line (named, respectively, L416-Ar4 and TR26×L416-Ar8). These clones were maintained by monthly in vitro subculture in Petri dishes (14 cm diameter) containing M medium (Bécard & Fortin, 1988) with 0.3% Phytagel (Sigma, St Louis, MO, USA).

To obtain optimal growth of the explants for inoculation, roots were cultured vertically on M medium containing 0.5% Phytagel in two steps. An initial preculture was set up with 3-cm-long explants (primary root apex with several laterals) and grown on square Petri dishes (12 × 12 cm) for 2–3 wk. From this preculture, fast-growing explants with a ‘fish-bone’ morphology (Fig. 1b), were transferred to square Petri dishes (8 × 8 cm) containing exactly 20 ml of medium and grown for 2–3 d before inoculation. Root growth was followed on a daily basis. All cultures (initial root cultures and inoculated root explants) were incubated at 25°C in the dark.

Figure 1.

Targeted inoculation of M. truncatula Ri T-DNA-transformed roots with pregerminated Gigaspora spores. (a) Preculture of roots growing vertically on a Phytagel support (13-d-old-culture). (b) Root explant with a ‘fish-bone’ morphology appropriate for targeted inoculation. Bar, 1 cm. (c) Schematic representation of typical root and fungal growth as recorded on the bottom of the Petri dish with contacts occurring during the first days of coculture. (d) Detail of first contacts of vertically growing hyphae with lateral root (arrows). In this case, two germ tubes developed from a single spore of G. gigantea . Bar, 1 mm. (e) Dense highly ramified external Gigaspora hyphae formed within 3–5 d of coculture. Bar, 500 µm.

In vitro inoculation with pregerminated spores of G. rosea or G. gigantea

Axenic spores of G. rosea or G. gigantea were gently inserted into the Phytagel medium in fresh square Petri dishes (8 × 8 cm) containing exactly 20 ml of M medium with 0.5% Phytagel, and cultured at 32°C at a slope of approximately 70° with a 2% CO2 atmosphere for optimal germination according to Bécard et al. (1992). Under these conditions spores germinated within 3–6 d. Because germ tubes of Gigaspora had a negative geotropism (Watrud et al., 1978) once they reached the surface of the gel they grew along the Phytagel surface towards the top of the dish. These pregerminated spores were then transferred in a gel plug containing the spore and germinating hyphae to the Petri dish with the root culture, and positioned directly underneath a growing secondary root (see Results). Note that it is essential that the same concentration and thickness of medium is used for spore germination and root culture to facilitate transfer and subsequent hyphal growth. Explant and hyphal growth (including branching), as well as all physical contacts between explant and hyphae, were recorded each day on the underside of the dish.

In situ histochemical localization of GUS activity and staining of fungal structures

A double-staining procedure was established to colocalize both GUS activity and fungal structures. Histochemical staining for β-glucuronidase (GUS) activity was performed according to Journet et al. (1994) using the substrate X-glcA (5-bromo-4-chloro-3-indolyl-β-D-glucuronide, cyclohexylammonium salt; Biosynth AG, Staad, Switzerland). GUS staining was performed in situ on whole root explants to allow simultaneous macroscopic observation of GUS expression and external fungal hyphae. Ten ml of X-glcA-containing solution was directly added to each square Petri dish containing the inoculated root explants and the dishes were incubated overnight in the dark at 37°C. For microscopic observations of fungal structures inside the root, sampled root pieces were subsequently stained using the convenient ink and vinegar method described by Vierheilig et al. (1998). In brief, roots were cleared with 10% KOH (w/v) for 6–7 min at 95°C, rinsed gently 3× with water, and stained for 3 min in 5% (v/v) black ink (Sheaffer, Ft Madison, IA, USA)-vinegar solution at 95°C. Roots were then rinsed and destained for 20 min with water (acidified with a few drops of vinegar). Microscopic observations were performed with a Zeiss Axiophot microscope (Carl Zeiss, Oberkochen, Germany) with bright field optics.

Results

M.truncatula Ri T-DNA-transformed roots for targeted inoculation with Gigaspora hyphae

In 1988, Bécard & Fortin first described a targeted inoculation system using excised and propagated Ri T-DNA-transformed carrot roots in conjunction with germinating spores of Gigaspora. In this in vitro system, carrot roots growing horizontally on a solid medium were targeted to the post-elongation zone with vertically growing germ tubes from spores positioned below the roots. We have adapted and modified this protocol for use with Ri T-DNA-transformed root cultures of the model legume M. truncatula. Unlike carrot, M. truncatula Ri T-DNA-transformed roots conserve a normal positive geotropism. Hence, for optimal growth, the explants are grown vertically rather than horizontally. In addition, the growth pattern of the M. truncatula root culture is very different from that of carrot, since the elongation zone is relatively short with frequent and highly regular lateral initiation (Fig. 1a). For this reason, we target the growing laterals rather than the elongation zone of the primary root.

A two-step culture was used to generate fast-growing explants with a characteristic ‘fish-bone’ morphology (see Materials and Methods). In our experimental system typical target explants consist of a primary root with six to eight laterals (Fig. 1b), which are grown for several days prior to targeting in order to establish regular growth. Because of the negative geotropism of the germ-tube, germinated spores of either G. rosea or G. gigantea were used to facilitate targeting, and the germinating hyphal tip was positioned just below a convenient secondary root so that contact would be made after 1 or 2 d of coculture (Fig. 1c,d). Growth rates of both the secondary roots and the germinating hyphae were approximately 0.3 cm per day. Following initial root contact, hyphae generally continued to grow upwards, ramifying at the same time. In this way additional contacts were made with secondary and tertiary roots growing above (Fig. 1c). In the majority of cases, hyphal branching rapidly led to dense, highly ramified external fungal structures in the medium, which could be observed within 4–5 d of coculture (Fig. 1e). In contrast, and as expected, germinating hyphae grown in the absence of root explants did not branch beyond secondary ramification.

In addition to extensive branching in the medium, the hyphae frequently grew and ramified along the root as shown in Fig. 2(a). Appressoria were generally observed at hyphal apices within 2 d following the first contact of the root with the germ tube (corresponding to 4 d of coculture, Fig. 2b), usually at subsequent contacts made by secondary hyphae. Internal hyphae then developed intra- and intercellularly (Fig. 2c,d) and invaded the cortex. Internal hyphae and Paris-type arbuscules (Fig. 2e) in inner cortical cells were visible 1 d later, suggesting that cortical invasion is very rapid. These arbuscules are presumed to be efficient symbiotic structures since lengthy cocultures of M. truncatula roots with Gigaspora rosea resulted in de novo viable spore production. In general, successful plant/fungal interactions (revealed by the presence of appressoria, internal hyphae and arbuscules) were only observed when dense, highly ramified fungal structures were present in the medium (Fig. 1e). Within less than 1 wk of coculture, infection events at the stage of appressoria and/or internal hyphae could be observed on the first two contacted secondary roots and their laterals for over 50% of the explants tested, with one or two infection sites/explant. This targeted inoculation technique therefore considerably facilitates observation of the early stages (prior to arbuscule formation) of the AM fungal/plant interaction.

Figure 2.

Gigaspora fungal structures (ink-stained) which develop in association with M . truncatula root explants following targeted inoculation. (a) External ramified hyphae on the root surface. (b) Detail of appressoria on the root epidermis (stars). (c) Intercellular hyphae invading the cortex (arrows). (d) Detail of intracellular hyphae (arrow) in outer cortical cell (star). (e) Symbiotic arbuscules (arrow) formed in inner cortical cell. Bars, 30 µm.

Expression of pMtENOD11-gusA in the epidermis and outer cortex correlates with appressorium formation and root infection

Having established the experimental conditions necessary for studying in detail the initial stages of root infection by Gigaspora AM fungal species, our objective was to evaluate the spatio-temporal expression pattern of the MtENOD11 gene. We have previously shown that this early nodulin gene is expressed during arbuscule development (Journet et al., 2001) and we wished to examine whether transcription of MtENOD11 also occurs during earlier stages of root infection. Ri T-DNA-transformed root clones were generated from the M. truncatula homozygous line expressing the pMtENOD11-gusA gene fusion (line L416; Journet et al., 2001). A single representative root clone (named L416-Ar4) was then propagated in vitro and used for targeted inoculation. In order to correlate reporter gene expression with early stages of fungal root infection we developed a double-staining protocol enabling fungal structures to be colocalized on histochemically GUS-stained roots (see Materials and Methods). Analysis of L416-Ar4 roots targeted with G. rosea germ tubes allowed us to identify a total of c. 40 early infection sites and to conclude that expression of the pMtENOD11-gusA transgene is indeed associated with all stages of root infection by the endosymbiotic fungus. Fig. 3(a,b) show that low level reporter gene activity is first detectable in epidermal cells directly associated with visible appressoria on the root surface. Note that GUS staining appears as a blue particulate precipitate due to post-treatment of the root with KOH, required for fungal staining. Epidermal cells have a characteristic elongated shape, and can be easily distinguished from the underlying cortical cells (e.g. Figure 3b). It is important to note that pMtENOD11-gusA expression is clearly limited to those cells in the immediate vicinity of the fungal appressorium. In a minority of cases (c. 30%), GUS staining was not observed in epidermal cells located below appressoria (discussed later).

Figure 3.

Expression of p MtENOD11-gusA in L416-Ar4 (a–e,h–j)) and TR26xL416-Ar8 (f,g,k) root explants inoculated with Gigaspora rosea revealed after double-staining for GUS activity (blue staining) and fungal structures (black ink staining). (a) Weak GUS expression (black arrows) in elongated epidermal cells associated with two appressorial sites (white arrows). Out of focus external ramified hyphae are above the epidermis (black arrowheads). (b) Epidermal GUS expression (black arrows) located directly under the appressorium (white arrow). The epidermal cell layer is labelled with a white star and the cortex with a black star. (c) GUS staining (black arrow) in outer cortical cell infected by internal hyphae (white arrow). (d) GUS staining in inner cortical cells containing arbuscules (e.g. white arrow). (e) GUS staining in inner cortical cells associated with propagating internal hyphae (black arrow) at the extremity of a colonised region with arbuscules (white arrow). (f) Ramified external hyphae on Myc TR26xL416-Ar8 roots. (g) Appressoria (white arrows) formed on TR26xL416-Ar8 roots. Note the absence of GUS activity associated with the fungal structures. (h–k) Schematic representation showing p MtENOD11-gusA expression throughout AM fungal infection of Myc + (h–j) and Myc (k) Ri T-DNA-transformed roots. Bars, 50 µm.

Transgene expression was also observed in outer cortical cells infected by penetrating hyphae, which had initiated from surface appressoria (Fig. 3c). Again, GUS activity was only found in cortical cells in the immediate vicinity of the hyphae. At later stages of infection, the pMtENOD11-gusA reporter is expressed at high levels in inner cortical cells of G. rosea-colonised roots either containing arbuscules (Fig. 3d) or associated with propagating internal hyphae (Fig. 3e). This is in line with our previously reported observations for MtENOD11 expression in arbuscule-containing cells of Medicago cultured roots colonized with G. intraradices (Boisson-Dernier et al., 2001). Since GUS activity in epidermal and outer cortical cells is no longer detectable once arbuscules have formed in the inner cortex we deduce that MtENOD11 expression is probably transient in the outer root cells. In conclusion, our results argue that the MtENOD11 gene is transcribed in epidermal and cortical root cells throughout all stages of fungal infection. This is presented schematically in Fig. 3(h–j).

Finally, in addition to the expression of the pMtENOD11-gusA reporter gene directly correlated with the infection process, GUS staining was also observed in cortical cells of tertiary roots in the vicinity of dense, highly ramified external fungal structures, but not colonised by the fungus (data not shown). The origin of this cortical expression pattern, independent of appressorium formation and infection, is currently being studied in more detail.

pMtENOD11-gusA is not expressed in epidermal cells associated with appressoria formed on roots derived from the Mycdmi2 mutant TR26

The fact that pMtENOD11-gusA expression was not observed in all epidermal cells in contact with appressoria (30% of cases) suggested that the triggering of gene expression may only occur in contacts leading to successful infection. To investigate this in more detail we made use of a MycM. truncatula mutant allele of the dmi2 gene, known as TR26 or dmi2–2 (Sagan et al., 1995; Catoira et al., 2000). This mutant has been characterized as Pen, that is to say capable of forming appressoria but defective in hyphal penetration (Calantzis et al., 2001). We have already described the introduction of the pMtENOD11-gusA reporter into the TR26 mutant background by genetic crossing with the L416 line (Vernoud et al., 1999). A. rhizogenes-transformed root cultures were generated from the homozygous line F3 TR26×L416, and a representative root clone (called TR26×L416-Ar8) was selected and propagated in vitro. No differences in morphology or root growth rate were observed between L416-Ar4 and TR26×L416-Ar8.

When TR26×L416-Ar8 root explants were used for targeted inoculation with Gigaspora rosea pregerminated spores, we observed normal fungal branching and the formation of dense, highly ramified external fungal structures in the medium within 4–5 days of coculture, as described earlier for the Myc+ clone L416-Ar4. This shows that the roots derived from the Myc mutant are still capable of stimulating hyphal branching. Moreover, typical highly branched external hyphae in contact with the root were frequently observed with TR26×L416-Ar8 (Fig. 3f), similar to those described by Calantzis et al. (2001) for TR26 plants inoculated with Glomus mossae. The first appressoria were observed within 3–4 d following the first contact with the root (Fig. 3g), and these appressoria generally developed on young tertiary roots. Most importantly, penetrating fungal hyphae were never observed in cortical cells, and despite lengthy coculture, arbuscules were never detected in the inner root cortex. The absence of root colonization for the TR26×L416-Ar8 clone was confirmed for other endomycorrhizal fungi such as Glomus intraradices (results not shown). We therefore conclude that the root clone TR26×L416-Ar8 derived from the dmi2 mutant TR26 faithfully reproduces the Myc phenotype of the original plant.

In order to examine the relationship between appressorium formation and MtENOD11 expression in the epidermis, root explants of TR26×L416-AR8 were targeted with G. rosea germ tubes and then double-stained for colocalization of GUS activity and the presence of the fungus. In all cases examined (a total of 15 appressorial sites in three independent experiments) reporter gene expression was totally absent in epidermal cells in contact with appressoria on the Myc root (Fig. 3g and corresponding schematic Fig. 3k). This strongly argues that the MtENOD11 early nodulin gene is only expressed in successful associations that lead to the formation of penetrating hyphae at the appressorium/epidermal cell interface. Thus, taken together, our findings suggest that the transcriptional activation of MtENOD11 during early stages of root infection is linked to the penetration of endomycorrhizal hyphae both across the root epidermis and through the outer cortex.

Discussion

Targeted fungal inoculation of M. truncatula root cultures as a means of studying early stages of the arbuscular mycorrhizal interaction

We have established an experimental system, based on controlled in vitro targeted inoculation of transgenic M. truncatula root cultures, specifically aimed at studying early stages of the endomycorrhizal association (prior to arbuscule formation). This approach allows continuous and simultaneous monitoring of fungal and root development. Within 4–5 d of coculture, appressoria and/or first internal hyphae were observed in greater than 50% of the inoculated explants and successful interactions were always correlated with the formation of dense, highly ramified external fungal structures in the vicinity of the root target. This is in line with the observations of Giovannetti et al. (1993b), who identified this form of fungal differentiation as a prerequisite for subsequent appressorium formation, root penetration and colonization. Following the first contact of the G. rosea germ tube with the Medicago root, appressoria are generally formed within 2 d, hyphal penetration occurs within 3 d, and arbuscules in the inner cortex are observed within 3–4 d. This rapid infection process is comparable with that described by Bécard & Fortin (1988) for carrot Ri T-DNA-transformed roots inoculated with the same fungal isolate (formerly known as G. margarita), and also by Brundrett et al. (1985) for in vivo infection of leek plants by Glomus versiforme. In conclusion, comparable infection processes and kinetics, as well as the maintenance of host-specificity (Bécard & Piché, 1990) and mutant phenotypes (see below) argue strongly that in vitro root culture is a valid system for studying both the fungal and plant partner during early stages of mycorrhizal colonization.

We have shown that the use of excised Ri T-DNA-transformed Medicago roots can be extended to mutant phenotype analysis. Ri T-DNA transformed roots derived from the TR26 (dmi2–2) Nod/Myc mutant are refractive to colonization by either Gigaspora or Glomus in our in vitro coculture system, with blockage occurring after formation of the appressorium and prior to epidermal cell penetration. This agrees with the previous findings of Balagi et al. (1994) who demonstrated that excised Ri T-DNA-transformed roots of the pea Nod/Myc mutants R25 and R72 inoculated with G. rosea, conserved the Pen mutant phenotype. The mutations in these two alleles have been assigned to the sym8 and sym9 pea loci, respectively, which are believed to correspond to the dmi1 and dmi3 loci of M. truncatula (J.-M. Anéet al., 2002; Walker et al., 2000). Thus, Ri T-DNA-transformed roots faithfully reproduce the Pen phenotype for all legume mutants so far tested, showing that the shoot is not required for the expression of this particular phenotype. Moreover, it is important to note that in our in vitro system dense, highly ramified external fungal structures were also observed following coculture of Myc roots with the fungus. This is consistent with the results of Buée (2000) who showed that root exudates from a Myc pea mutant (DK10, an allele of sym8: G. Duc, INRA, Dijon, France, pers. comm.) are able to elicit fungal branching. This implies that the initial stages of fungal differentiation, elicited by plant exudates, are not altered in interactions with these mutants deficient in root penetration.

The MtENOD11 early nodulin gene is expressed throughout the infection process in epidermal and outer cortical cells

We have exploited the targeted inoculation approach described in this article for studying the spatio-temporal expression of the early nodulin MtENOD11 gene during early infection stages of the endomycorrhizal interaction. During root nodulation MtENOD11 (and MtENOD12) expression correlates with preinfection responses to Sinorhizobium meliloti in root hairs, and then with formation of the infection thread in root hairs and its progression through cortical tissues (Pichon et al., 1992; Journet et al., 1994; Journet et al., 2001). In addition, and as mentioned earlier, expression of this gene has been observed in inner cortical cells containing recently formed arbuscules after colonization of M. truncatula with the endomycorrhizal fungus G. intraradices (Journet et al., 2001). These conclusions have been drawn from studies using transgenic Medicago plants containing the pMtENOD11/12-gusA reporters, supported by confirmatory data provided either by RT-PCR (Pingret et al., 1998; Journet et al., 2001) or in situ hybridization (Pichon et al., 1992; Journet et al., 2001).

In Myc+ Ri T-DNA-transformed roots inoculated with either G. rosea or G. gigantea, pMtENOD11-gusA is expressed in the majority of epidermal cells in contact with appressoria, as well as the outer and inner cortical cells in the immediate vicinity of internal hyphae and arbuscules. This suggests that the MtENOD11 gene is expressed throughout the infection process in root cells in contact with the fungus. Similar results have also been obtained with either Ri T-DNA-transformed roots or with plantlets (transgenic plants containing the pMtENOD11-gusA fusion) inoculated in vitro with Glomus intraradices (data not shown), although the number of early infection events observed in both cases was severely limited by the lack of synchronization. The rarity of these early infection events and the fact that gene expression is transient explains why this expression pattern was not previously observed (Journet et al., 2001) and underlines the advantage of this targeted inoculation technique for early gene expression studies. On the other hand, GUS activity was not detected in epidermal cells in contact with appressoria on Ri T-DNA-transformed roots of the Pen mutant TR26 (dmi2–2), implying that either hyphal penetration or essential processes preceding penetration are required for induction of MtENOD11 in the root epidermis. In this context, it is interesting to note that Nagahashi (2000) has attempted to distinguish between two kinds of appressoria. ‘Functional’ appressoria, as defined by Garriock et al. (1989), are those which adhere to the host epidermal cell surface and initiate penetration hyphae, whereas ‘nonfunctional’ appressoria are formed either on Myc mutants (Calantzis et al., 2001), on nonhost roots (Giovannetti et al., 1993a) or on isolated host epidermal cell walls (Nagahashi & Douds, 1997). According to this definition, ‘nonfunctional’ appressoria develop in the epidermal cell junction and never initiate penetration hyphae. If MtENOD11 is only expressed in the presence of ‘functional’ appressoria this could explain the absence of GUS expression on Pen roots. The percentage of appressoria without associated epidermal GUS expression that we observed on Myc+ roots might therefore correspond either to nonfunctional appressoria formed on parts of the root that are not responsive to infection, or alternatively to early stages of appressorium development. Interestingly, the spatio-temporal expression pattern of MtENOD11 that we have observed throughout hyphal infection is consistent with the results of Albrecht et al. (1998), who used an RT-PCR approach to show that the pea ENOD12A gene is transcribed during early stages of root infection. Furthermore, in the same article it was shown that AM infection-induced PsENOD12A expression is blocked in a pea Nod/MycSym8 mutant. Nevertheless, these apparent similarities in expression need to be validated since PsENOD12A does not appear to be expressed during arbuscule development in pea (Albrecht et al., 1998).

Comparing MtENOD11 expression during rhizobial and endomycorrhizal root infection, it is striking that transcription is triggered in both cases in epidermal and cortical cells physically associated with microsymbiont penetration: the infection thread in the case of S. meliloti, internal hyphae in the case of the endomycorrhizal fungus. MtENOD11 is predicted to encode a repetitive proline rich protein (RPRP), targeted to the extracellular matrix (Journet et al., 2001). As we have previously suggested, this atypical RPRP, low in cross-linking tyrosine residues, could play a role in the modification of cell wall plasticity necessary for cell penetration. As mentioned earlier, the MtENOD11 gene is also transcribed in root epidermal tissues prior to rhizobial infection, and this expression can be mimicked by the addition of purified rhizobial Nod factors (Journet et al., 2001). Indeed it has been suggested by several authors that endomycorrhizal fungi secrete soluble signalling factors analogous to Nod factors (e.g. Gianinazzi-Pearson & Dénarié, 1997). Since M. truncatula dmi2 mutations are blocked in Nod factor-elicited MtENOD11 expression (Vernoud et al., 1999; Catoira et al., 2000) it is conceivable that the absence of GUS reporter expression at appressorial sites of dmi2 roots results from an analogous block in a hypothetical ‘Myc’ factor signaling pathway. However, comparative studies of signal transduction leading to MtENOD11 expression will have to await the identification of such putative symbiotic fungal elicitors.

In conclusion, our results indicate that M. truncatula Ri T-DNA-transformed roots can be successfully used for studying the earliest stages of the arbuscular mycorrhizal association. These root cultures grow vigorously, displaying positive geotropism and organized lateral root development, and most importantly can be efficiently colonized by both Glomus and Gigaspora endomycorrhizal fungi. The possibility of targeting roots with infectious fungal hyphae, as described in this article for ENOD gene expression studies, now opens the way to focusing on the detailed cellular responses of the host plant during early plant/fungal contact in outer root tissues. Such studies will be greatly facilitated by the convenience of introducing reporter genes into M. truncatula roots via A. rhizogenes, and by the availability of extensive genetic resources for this model legume including numerous symbiotic mutants.

Acknowledgements

We are grateful to G. Duc and M. Sagan (INRA, Dijon, France) for initially providing us with seeds of the TR26 Nod/Myc mutant, to E.-P. Journet for providing seeds of F3 TR26 × L416 and to M. Buée and B. Tamaslov for kindly providing us with Gigaspora spores. Our thanks also to V. Vernoud for her valuable advice on in vitro mycorrhization, to E.-P. Journet and F. Carvalho-Niebel for critical reading of the manuscript and other colleagues for helpful comments. Financial support was provided by the French Ministry of Research and Technology as a grant to the Toulouse Institute of Federative Research in Plant Biotechnology (IFR 40).

Ancillary