INTRODUCTION

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION CONCLUSIONS MATERIALS AND METHODS LITERATURE CITED

Arbuscular mycorrhizal (AM) fungi have existed in symbiosis with plant roots for over 460 million years, since the appearance of the earliest land plants (Remy et al., 1994). This group of fungi, recently renamed Glomeromycota (Schüssler et al., 2001), is one of the most widely distributed; 95% of present-day plant species belong to families that are characteristically mycorrhizal (Smith and Read, 1997). AM fungi are able to transfer rare or poorly soluble nutrients such as phosphorous, copper, and zinc from the soil to the plant, which in turn provides carbohydrates to the fungus. This nutrient exchange may be of critical importance when soil fertility and water availability are low, conditions that severely limit agricultural production in most parts of the world. Although AM fungi are both agriculturally and ecologically important, very little is known about the cellular and molecular events that occur during establishment of the association, and in particular events that play a role in signaling and recognition of both symbiotic partners.

Before infection, AM fungi recognize and respond to their potential hosts. Compounds constitutively secreted by the roots of host plants, but not non-host plants, stimulate ramifications in hyphae from germinating spores of Gigaspora and Glomus spp. (Mosse and Hepper, 1995; Giovannetti et al., 1993b; Buée et al., 2000). These morphological changes increase the possibility of contact between hyphae and host roots, but also signal a physiological "switch" to active presymbiotic fungal growth without which hyphal attachment and appressorium formation may not occur (Giovannetti et al., 1994). Upon contact, the topographical and/or biochemical properties of host root epidermal cell walls induce the formation of AM fungal appressoria (Giovannetti et al., 1993a; Nagahashi and Douds, 1997). Although rapid stimulation of spore germination, hyphal growth, and appressorium formation by host-roots has obvious advantages for the survival of the obligately symbiotic AM fungus, no evidence to date indicates plant recognition of the fungus before contact, nor the existence of fungal signals before root penetration.

Gene expression studies indicate an active plant response to the AM fungus during the earliest stages of hyphal penetration. Studies using reverse transcriptase-PCR and northern analyses in pea (Pisum sativum) suggest that induction of PsENOD12A and Psam5 is concurrent with appressorium formation and hyphal proliferation in the cortex (Albrecht et al., 1998; Roussel et al., 2001). Use of gene-promoter -glucuronidase (GUS) fusions in rice (Oryza sativa) has revealed that expression of the lipid transferase protein (Ltp) gene in epidermal cells is associated with appressorium formation (Blilou et al., 2000). By a similar approach, Chabaud et al. (2002) have recently shown that the Medicago truncatula ENOD11 gene is transcriptionally activated in epidermal and cortical cells containing penetration hyphae during infection by Gigaspora rosea. This early nodulin gene is also expressed in M. truncatula epidermal cells in response to purified Nod factors, during infection of the root by Sinorhizobium meliloti, and in arbuscule-containing cells in mycorrhizal roots (Journet et al., 2001). Thus, in AM infection, as in root infection by Rhizobium sp. bacteria, the host plant actively and specifically responds to penetration of host root cells.

Mycorrhization and nodulation are very different processes, involving unrelated microbial symbionts, and giving rise to very different physiological structures in the host plant root. Nonetheless, the establishment of these two root symbioses appears to involve a number of related plant responses, including the expression of common plant genes (for references, see Gianinazzi-Pearson and Dénarié, 1997; Hirsch and Kapulnik, 1998; Harrison, 1999). The genetic evidence is the most striking: nodulation-defective (Nod) mutants that are also non-mycorrhizal (Myc) have been found in pea (Duc et al., 1989), alfalfa (Medicago sativa; Bradbury et al., 1991), M. truncatula (Sagan et al., 1995; Catoira et al., 2000), bean (Phaseolus vulgaris; Shirtliffe and Vessey, 1996), and Lotus japonicus (Wegel et al., 1998; Bonfante et al., 2000). The recent characterization of several M. truncatula, L. japonicus, and pea Nod/Myc mutants blocked for very early steps in Nod factor signal transduction and also required for the establishment of arbuscular mycorrhizas, indicate that certain elements of signal transduction are common to both symbiotic interactions (Catoira et al., 2000; Walker et al., 2000; Stracke et al., 2002).

These analogies have led to the suggestion that so-called "Myc factors" may be produced by AM fungi, acting as fungal signals recognized by host roots and necessary for the establishment of a successful mycorrhizal association. Although many cellular, molecular, and developmental root responses to Nod factors have been characterized in several legume systems, no such responses have been described preceding infection of plant roots by AM fungi. This is probably because of the difficulty in studying the early steps of AM fungal symbiosis, and the non-synchronous nature of AM infection. Chabaud et al. (2002) recently developed an in vitro technique for studying the early stages of AM fungal infection using Agrobacterium rhizogenes-transformed roots of M. truncatula carrying a chimeric gusA gene fusion under the control of the MtENOD11 promoter. The authors observed that during early stages of the interaction between M. truncatula Ri T-DNA-transformed roots and G. rosea, strong cortical pMtENOD11-gusA expression was often observed in noninfected roots in the vicinity of fungal-root contacts. This suggested that pMtENOD11-gusA gene expression might be induced in M. truncatula roots by G. rosea without direct contact. To test this hypothesis, we have developed a coculture system, also applicable to seedlings, where roots and fungi are separated by a membrane barrier. In this way, we have been able to demonstrate unequivocally that germinated spores of AM fungi synthesize a diffusible factor capable of triggering the expression of an early nodulin gene in host roots in the absence of fungal-plant contact.

	RESULTS

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION CONCLUSIONS MATERIALS AND METHODS LITERATURE CITED

Membrane-Separated M. truncatula/G. rosea Coculture

To test whether a fungal diffusible factor was responsible for inducing pMtENOD11-gusA expression, we inserted a cellophane membrane between the fungus and the Ri T-DNA-transformed roots from the start of coculture (Fig. 1a). In this experimental system, the two symbiotic partners are able to grow in close proximity (<1 mm), thus facilitating the exchange of pre-infection signals. Furthermore, the transparency of the cellophane membrane used made it possible to follow fungal growth and morphology, which were essential parameters in defining the bio-assay conditions. Germination of G. rosea spores occurred after 4 to 7 d and was followed 2 d later by the appearance of the hyphal ramifications characteristically produced by AM fungi in response to diffusible root factors (Figs. 1c and 2a). Histochemical GUS staining, carried out after 7 to 14 d of coculture, systematically revealed strong pMtENOD11-gusA transcriptional activation in numerous root laterals, with staining initiating just behind the root tip and extending some distance up the root (Fig. 2, a and b). This expression pattern was almost exclusively observed in the immediate proximity of the developing fungus. In control roots, grown without fungus but with the cellophane membrane, pMtENOD11-gusA expression was generally only observed at three locations in the root (root caps, the base of root laterals, and occasionally in vascular tissues: Fig. 2, c and e), corresponding to the non-symbiotic "constitutive" expression of MtENOD11 in roots of intact plants (Journet et al., 2001).

View larger version (14K): [in this window] [in a new window]   Figure 1.   Schematic representation of the membrane-separated coculture of AM fungi and M. truncatula. a, Excised (Ri T-DNA-transformed) roots; b, seedlings. c, Detail of the corridor containing dense hyphal growth and ramifications, in which MtENOD11 induction occurs. 1, Spores; 2, location of membrane; 3, primary root; 4, secondary root; 5, tertiary root; 6, negatively geotropic germ tubes; 7, corridor containing hyphal ramifications.

View larger version (102K): [in this window] [in a new window]   Figure 2.   Root pMtENOD11-gusA induction in membrane-separated coculture. a, b, d, f, and g, Ri T-DNA-transformed roots membrane-separated from G. rosea. c, e, and h, Control roots cultured with membrane separation but without fungus. a, GUS activity colocalizes with hyphal ramifications on other side of membrane (white arrowheads), bar = 0.2 cm. b, MtENOD11 induction occurs mainly in tertiary Ri T-DNA-transformed roots; c, only constitutive expression (see also e) is observed in control roots, bars = 2 cm. d, AM factor-induced GUS activity stretches from 0.1 cm behind the root cap as far as the zone of mature root hair growth, bar = 0.2 cm; e, in control roots, only constitutive expression is present in root caps and at the base of lateral roots, bar = 0.1 cm. f, Detail of pMtENOD11-gusA induction in epidermal cells including root hairs, bar = 50 µm. g and h, Seventy-five-micrometer-thick transverse sections of AM-inoculated (g) and control (h) roots, bars = 150 µm. i, Membrane-separated coculture of G. rosea with whole plants; j, control plants. i, pMtENOD11-gusA is induced in young lateral roots, bar = 0.3 cm; j, only background GUS activity is seen in vascular tissues in control roots, bar = 0.3 cm. k and l, pMtENOD11-gusA induction in membrane-separated coculture with other fungi. Induction occurs in Ri T-DNA-transformed roots cocultured with G. margarita (k), but not with Phy. medicaginis (l), bars = 0.6 cm. m, pMtENOD11-gusA induction in membrane-separated coculture of G. rosea and with Ri T-DNA-transformed roots derived from the M. truncatula Myc mutant dmi2-2, bar = 0.7 cm.

To verify that the observed reporter gene expression is also valid for whole plants, we developed an axenic petri dish experimental system for AM colonization of M. truncatula seedlings. In this whole-plant system (see "Materials and Methods"), spore germination occurred after 5 to 7 d of coculture, when seedling roots passed close to spores. Fungal hyphae generally made contact with roots 2 to 3 d later. After 1 month of coculture, daughter spores were observed, indicating that a viable symbiosis had formed between the AM fungus and the host M. truncatula plant. We then applied the membrane-separated approach to germinating spores and transgenic pMtENOD11-gusA seedlings (Fig. 1b). After 10 to 14 d of coculture, GUS expression was observed in the cortex, epidermis, endodermis, pericycle, and vascular tissues of secondary roots growing in close proximity to AM fungal germ tubes and hyphal ramifications (Fig. 2i), whereas in control roots, only constitutive expression was seen in vascular tissues (Fig. 2j), sites of root lateral initiation, and root tips (not shown) as described above.

A Diffusible Fungal Factor Induces pMtENOD11-gusA Expression

To confirm that the root response we have observed is attributable to a diffusible factor, we verified by several methods that the fungus had not traversed the membrane barrier. First, careful and extensive microscopic examination of all roots stained with Chlorazol Black E failed to detect fungal structures either at the root surface or within roots (data not shown). Second, diagnostic PCR using fungal-specific primers (see "Materials and Methods") revealed no fungal signals in DNA samples extracted either from control root cultures or roots recovered 10 d after coculture with membrane-separated fungus (Fig. 3). On the other hand, DNA extracted from both G. rosea spores and colonized roots gave a positive PCR signal. Third, replacing the cellophane membrane with a polycarbonate membrane (0.6-µm pore size) resulted in the same pMtENOD11-gusA expression pattern. Finally, the addition of a second cellophane membrane, increasing the physical distance between the fungus and the root to approximately 3 mm, did not prevent the characteristic reporter gene expression. Although the appearance of hyphal ramifications was delayed by 24 h in the double membrane tests, the number of responding roots and the intensity of GUS staining was not different from single-membrane experiments (data not shown). In addition, replacing the cellophane membrane by a dialysis membrane (3.5 kD molecular mass cut-off) also permitted pMtENOD11-gusA induction. Thus, we conclude that an AM fungal factor, possibly less than 3.5 kD in size and capable of diffusing across a variety of membrane barriers, is capable of eliciting early nodulin gene expression in host roots.

View larger version (65K): [in this window] [in a new window]   Figure 3.   PCR analysis of potential fungal contamination. Products from plant and fungal DNA amplified with universal primers ITS1/4 (odd lanes) and fungal specific primers ITS1F/4 (even lanes). Lanes 1 and 2, DNA of M. truncatula control roots; Lanes 3 and 4, DNA from roots harvested from a membrane-separated coculture of M. truncatula and G. rosea at 10 dai; Lanes 5 and 6, DNA of M. truncatula roots colonized by G. rosea; Lanes 7 and 8, DNA from G. rosea spores; and Lanes 9 and 10, water control. Note that the band (550 bp) amplified by fungal primers in spores (lane 8) and colonized roots (lane 6; arrowheads), could not be detected in roots separated from AM fungi by a membrane (lane 4). Two faint non-specific bands (600 and 680 bp) can be seen in lanes 2 and 4 corresponding to root DNA.

Localization of Diffusible AM Factor-Elicited MtENOD11 Expression

By staining for GUS activity directly in the petri dish containing the membrane-separated coculture, we were able to show that pMtENOD11-gusA induction occurred mainly in the proximity of the advancing fungal germ tubes and hyphal ramifications (Fig. 2a; for schematic localization, see Fig. 1a). Within a "corridor" containing all fungal germ tubes and hyphal ramifications (Fig. 1c), pMtENOD11-gusA induction was seen in approximately one-quarter (25% ± 4%) of the tertiary roots present, whereas in control roots only 5% ± 2% of the tertiary roots showed expression of the transgene (Table I). MtENOD11 expression was also induced to a lesser extent in secondary roots (7% ± 2% compared with 0% ± 0% in controls). Primary roots never showed transgene induction. GUS activity was limited to roots growing within the culture medium, because expression of the gusA reporter was never observed in aerial roots. Microscopic examination of stained roots revealed that pMtENOD11-gusA activity initiated 0.1 to 0.7 cm behind the root cap and continued up to 1.5 cm from the root cap (Fig. 2d). This comprises the zone of root hair emergence, root hair development, and part of the region of mature root hairs. Semithin sections of stained root segments showed that in roots exhibiting strong GUS activity, staining was observed in the vast majority of cortical cells and often in the root epidermis (including root hairs), the root endodermis, and vascular tissue (Fig. 2, f and g). In weaker-stained roots, GUS expression was primarily seen in the cortex and was often limited to isolated cells or small groups of cells in one or all of the root tissues mentioned above (data not shown).

View this table: [in this window] [in a new window]   Table I.   pMtENOD11-gusA induction Percentage of Ri T-DNA-transformed roots of M. truncatula expressing pMtENOD11-gusA after 10 d of membrane-separated coculture with G. rosea (+AM) and without (AM). Roots were counted in the "corridor" of dense fungal growth and hyphal ramifications. Values listed are means ± SE. Data presented are from a single representative experiment. Similar results were obtained in at least five independent experiments. Data in each column followed by a different letter (a or b) are significantly different according to Student's t test (P < 0.05, n = 12).

All AM Fungi Tested Produce a Diffusible Factor Capable of Inducing MtENOD11 Expression

All of the AM fungi tested (G. rosea, Gigaspora margarita, Gigaspora gigantea and Glomus intraradices; Table II) produced diffusible factors with pMtENOD11-gusA-inducing activity, although reporter gene expression varied in intensity between fungi (data not shown). Under our membrane-separated coculture conditions, pMtENOD11-gusA expression was strongest and most uniform in response to G. rosea and G. gigantea. Per equivalent fungal inoculum, G. margarita elicited weaker and more localized reporter gene expression (Fig. 2k), and G. intraradices elicited even weaker gene expression. We therefore conclude that the capacity to produce this diffusible factor may be common to all AM fungi tested, irrespective of phylogenetic diversity.

The MtENOD11-Activating Factor Is Specific to AM Fungi

Non-mycorrhizal fungi were also tested to see whether the response we have observed is specific to AM fungi or whether it might correspond to a general stress response to fungi. The three M. truncatula root pathogens tested (F. solani f.sp. phaseoli, Phoma medicaginis var pinodella, and Phythophthora medicaginis; Table I) were first tested for their effect on Ri T-DNA-transformed roots under our in vitro growth conditions. Inoculation of root explants with all three fungi resulted in negative growth effects within 4 d after inoculation (dai). For F. solani var phaseoli and Phy. medicaginis, root growth ceased at approximately 8 dai, whereas Pho. medicaginis var pinodella was less aggressive, with growth of infected roots continuing, despite signs of necrosis, until at least 20 dai. We also tested pathogenicity on seedlings, where Suc was absent from the medium. After inoculation of M. truncatula seedlings with F. solani var phaseoli and Phy. medicaginis, the first visible signs of pathogenicity (necrosis of root tissue in the primary root) were visible 2 to 3 dai, with plant death occurring 10 to 14 dai. Again, Pho. medicaginis var pinodella was less aggressive, with the plant surviving for at least 3 weeks with only moderate signs of necrosis. To cover the entire infection period, histochemical staining for GUS activity was therefore performed at 3, 6, and 10 dai. Our experiments failed to reveal any cortical or epidermal pMtENOD11-gusA expression in either Ri T-DNA-transformed root cultures or seedlings inoculated with any of the three pathogenic fungi.

Finally, we investigated whether reporter gene expression could be induced in either root organ cultures or seedlings separated from pathogenic fungi by a cellophane membrane. F. solani var phaseoli and Phy. medicaginis developed rapidly and digested the cellophane membrane within 4 dai, leading to infection of the host roots. Pho. medicaginis var pinodella had not crossed the membrane barrier by 10 dai but nonetheless elicited some necrosis in roots. Despite this, pMtENOD11-gusA expression was never observed in roots confronted with pathogens across a membrane, stained for GUS activity at 3, 6, and 10 dai (Fig. 2l). The absence of pMtENOD11-gusA induction in interactions with three different pathogenic fungi strongly suggests that the diffusible factor is specific to AM fungi.

M. truncatula Nod/Myc dmi Mutants Also Respond to the Diffusible AM Factor

Mutations in the three DMI genes of M. truncatula result in Nod/Myc phenotypes and corresponding mutants are blocked at early stages of a Nod factor signal transduction pathway (Catoira et al., 2000). To evaluate possible mechanistic parallels between the Rhizobium sp. Nod factor and the diffusible AM factor, we established pMtENOD11-gusA root cultures derived from three representative dmi mutants (dmi1-1, dmi2-2, and dmi3-1; Table II). After several weeks of coculture with G. intraradices, no new spores were formed, and in all cases no fungal growth was visible after 14 d (results not shown). In addition, microscopic observations of Chlorazol Black-stained roots from 11-d-old cocultures with G. rosea revealed that the fungus was blocked at the stage of appressorium formation, and no hyphal penetration had occurred. Despite their defective mycorrhization phenotype, these root cultures still produce the factors that stimulate AM hyphal ramification. Using the membrane barrier approach described earlier, Ri T-DNA-transformed root cultures of dmi1, dmi2, and dmi3 mutants were tested for their capacity to respond to the diffusible AM factor produced by G. rosea. Expression of pMtENOD11-gusA was induced in dmi1, dmi2, and dmi3 mutant root cultures with the same histological localization, intensity, and timing as wild-type M. truncatula roots (Fig. 2m). Membrane-separated coculture with G. rosea led to statistically significant (P < 0.001) pMtENOD11-gusA induction compared with control roots (Fig. 4). The percentage of roots showing GUS activity varied somewhat between mutants. In particular, dmi3 roots showed less GUS activity than wild-type and other dmi mutant roots in response to the diffusible AM factor, however, differences were not statistically significant (P = 0.20). Thus, mutations in the dmi1, dmi2, and dmi3 genes apparently do not alter the pMtENOD11-gusA induction response to the AM factor.

View larger version (42K): [in this window] [in a new window]   Figure 4.   pMtENOD11-gusA induction in M. truncatula Nod/Myc dmi mutants. Percentage of Ri T-DNA-transformed roots of M. truncatula wild-type and dmi mutants that show induction of pMtENOD11-gusA after 10 d of membrane-separated coculture with G. rosea (+AM) and without (AM). Roots were counted in the "corridor" of dense fungal growth and hyphal ramifications. Data from three independent experiments were pooled and treated as subsets (blocks) of a single data set for statistical analysis. Means ± SE labeled with a different letter (a and b) are significantly different according to analysis of variance followed by Tukey's test (P < 0.05, n = 7).

	DISCUSSION

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION CONCLUSIONS MATERIALS AND METHODS LITERATURE CITED

Evidence for a Diffusible Factor Produced by AM Fungi

We demonstrate that pMtENOD11-gusA expression is induced in M. truncatula roots by AM fungi in cocultures separated by a variety of membrane barriers. We have verified that in our experiments, fungal hyphae did not cross the membrane both by microscopic examination of stained roots and PCR analysis using fungus-specific primers. Furthermore, the addition of a second membrane barrier, increasing the distance between the fungus and the roots to at least 3 mm, did not alter induced pMtENOD11-gusA reporter gene expression in roots. We therefore conclude that a diffusible factor produced by AM fungi induces pMtENOD11-gusA expression in M. truncatula roots.

In these experiments, we have predominantly made use of Ri T-DNA-transformed roots expressing a pMtENOD11-gusA reporter as a convenient experimental tool. Such root cultures have been validated for studies of plant/AM fungal associations in the case of carrot (Daucus carota), pea, and M. truncatula (Bécard and Fortin, 1988; Balagi et al., 1994; Boisson-Dernier et al., 2001; Chabaud et al., 2002). Furthermore, Ri T-DNA-transformed roots derived from penetration mutants exhibit the same phenotype as the intact plant for all legume mutants tested so far (this study; Balagi et al., 1994; Chabaud et al., 2002). Despite this, we have also verified that induction of pMtENOD11-gusA expression also occurs in the same root tissues of whole plants in response to the diffusible AM factor.

All of the AM fungi tested elicited similar spatio-temporal expression of the pMtENOD11-gusA reporter. However, the intensity of GUS expression varied with fungal species. It is therefore possible that the quantity of diffusible AM factor differs between AM fungi or that factors from different fungi may have different gene-inducing activities or different diffusion rates/stabilities. By staining for GUS activity directly in the petri dish containing the coculture, we were able to conclude that pMtENOD11-gusA-expressing roots were always in the vicinity of fungal germ tubes and highly ramified hyphal structures on the other side of the membrane, whereas reporter gene activity was not observed in roots distant from the fungus. Our observations are therefore consistent with the hypothesis of a slowly diffusible or relatively unstable factor. In addition, our results with dialysis membranes suggest that the factor may have a low molecular mass (<3.5 kD). Nevertheless, because the pathogenic fungi efficiently digested the cellulose membranes, we cannot exclude the possibility that the AM fungi, in contact with the membrane for several days, might alter its cellulose structure and hence its molecular cut-off.

Significance of MtENOD11 Expression in Response to the Diffusible AM Factor

Similar results were obtained with all four AM fungal species tested, including the phylogenetically distant G. intraradices, implying that diverse AM fungi produce a diffusible factor that can be perceived by host roots without direct physical contact between the two organisms. On the other hand, pMtENOD11-gusA induction was not observed in interactions with fungal pathogens (aggressive and nonaggressive) in our study or in interactions with the biotrophic root fungus Rhizoctonia sp. (Journet et al., 2001). Because infection with aggressive fungal pathogens alters root growth and therefore may alter gene expression, we cannot directly interpret the absence of pMtENOD11-gusA induction in response to these fungi. However, our results with a range of fungal pathogens suggest that the pMtENOD11-gusA induction we observe is specific to AM fungi. Comparison with biotrophic fungi in future may be useful, because these fungi can enter host root tissues without negatively affecting root growth, and therefore may be more appropriate controls for AM fungal studies.

Diffusible AM factor-dependent induction of pMtENOD11-gusA occurred mainly in root laterals, the preferred site for AM colonization both for whole plants and in vitro root cultures (Mosse and Hepper, 1995; Chabaud et al., 2002). However, within the zone perceiving the AM factor, certain roots responded more strongly than others, and some roots not at all. We do not yet understand the reason for this variability. It is possible that phytohormonal balance or some other internal regulatory mechanism may control the susceptibility of a given root to the AM factor, as has been suggested for ectomycorrhizal fungal and Rhizobium sp. bacterial infection (Smith and Read, 1997; Mathesius et al., 2000). In the absence of a membrane, appressoria were often observed on roots showing strong cortical pMtENOD11-gusA expression (data not shown), suggesting a correlation between the gene expression pattern that we observed and the subsequent passage of the fungus through these tissues. However, a direct link between cortical pMtENOD11-gusA expression and subsequent appressorium formation and AM root infection remains to be established. Finally, it should be underlined that the expression of pMtENOD11-gusA in response to the diffusible AM factor differs significantly from that elicited at a later stage during fungal root infection (Chabaud et al., 2002). In the latter case, expression is strictly limited to individual epidermal and cortical cells penetrated by hyphae, and furthermore no expression is observed with infection-defective Nod/Myc mutants.

The physiological function of the MtENOD11 gene product is so far unknown, but based on its nucleotide sequence, MtENOD11 is predicted to encode an extracellular repetitive Pro-rich protein with low overall Tyr content. A possible role for MtENOD11 in the construction of a cell wall with reduced cross-linking and thus higher plasticity and/or matrix porosity has been suggested (Journet et al., 2001). So far, no major alterations in cell size or shape have been observed in pMtENOD11-gusA-expressing cells in semithin sections, but a more detailed analysis is clearly required to evaluate whether these cells are modified in their cellular morphology. Some other ENOD genes have been found to be induced during mycorrhizal symbiosis, including Psam5, which is expressed very early in AM fungal interactions with both Myc⁺ and Myc mutants of pea (Roussel et al., 2001). However, little is known about the products of these genes or whether they are also expressed in non-legumes during mycorrhization. Further spatio-temporal expression studies of known genes and the identification of novel genes induced by the diffusible AM factor should help toward understanding the physiological significance of both pMtENOD11-gusA induction and the activity of the diffusible factor produced by AM fungi.

Is the Diffusible AM Factor a Symbiotic Signal?

When roots and AM fungus are separated by a membrane, pMtENOD11-gusA induction correlated both spatially and temporally with the appearance of the hyphal ramifications that indicate that the fungus has "recognized" its host (Giovannetti et al., 1994). No transgene expression was observed when hyphal ramifications were not already present. Thus, a crucial step of fungal-host recognition may be required for synthesis of the diffusible AM factor, suggesting that, in the same way that legume root flavonoids activate nodulation genes in rhizobia, host root compounds may activate mycorrhization genes in AM fungi. Overall, our results suggest that a molecular dialogue takes place between germinating AM fungi and their potential host roots.

Analysis of legume Nod/Myc mutants blocked for Nod factor signal transduction has led several authors to suggest an analogous signaling mechanism in the AM fungal and rhizobial symbiosis, including a putative "Myc" factor equivalent to Nod factor (LaRue and Weeden, 1994; Albrecht et al., 1998; Catoira et al., 2000; Walker et al., 2000; Stracke et al., 2002). As the chitin backbone of the Nod factor molecule is more typical of fungi than bacteria, the diffusible AM factor could indeed be a Nod factor-like molecule or simply chitin oligomers. It has been shown that such chito-oligosaccharides can induce rapid alkalinization of tomato cell suspension cultures (Felix et al., 1993), transient GmENOD40 induction in soybean (Glycine max; Minami et al., 1996), and calcium spiking in pea (Walker et al., 2000). However, in this case, we would hypothesize that the chitin oligomers released by AM fungi during symbiotic infection are not the same as chitin fragments derived from plant chitinase-degradation of the fungal pathogens Pho. medicaginis var pinodella and F. solani.

Evidence for a DMI-Independent Symbiotic Transduction Pathway

Our study has revealed some important differences between responses induced by Nod factors and the diffusible AM fungal factor that must be considered. MtENOD11 induction by Nod factor and diffusible AM factor is not identical; pMtENOD11-gusA is expressed mainly in epidermal cells in response to Nod factor (nod-ENOD11 response), whereas the AM factor induces expression of the same gene most strongly in the cortex of a larger region of the root (myc-ENOD11 response). This indicates that, although the same gene responds to a diffusible signal from both microsymbionts, spatial and temporal gene expression is probably regulated differently in the two symbioses. In addition, the M. truncatula dmi mutants are totally blocked for certain Nod factor responses, in particular for epidermal MtENOD11 induction (Catoira et al., 2000). In our study, dmi1, dmi2, and dmi3 all responded positively to the diffusible AM factor, indicating that the DMI gene products are not required for the AM factor-induced ENOD11 expression that we have revealed. The diffusible AM factor is thus likely to activate this response by a signal transduction pathway independent of the DMI genes. This pathway may be specific to the AM factor, i.e. not activated by Nod factor signaling.

Finally, is our diffusible AM factor the "Myc factor" proposed to activate the DMI cascade in M. truncatula (Catoira et al., 2000)? Our diffusible AM fungal factor may activate only the pathway leading to myc-ENOD11 induction, whereas a different factor, the Myc factor, would activate the DMI pathway. As an alternative, we cannot rule out the hypothesis that the AM factor activates steps in the DMI pathway that are common to both Nod and Myc signaling in addition to the pathway leading to myc-ENOD11 induction. Whether or not our diffusible factor is the Myc factor, the signaling pathway that we have revealed might temporally precede DMI-dependent fungal penetration of root tissues. This would be consistent with the fact the dmi Myc phenotype is blocked at appressorium formation and requires physical contact between plant and fungus, whereas our myc-ENOD11 induction occurs in the absence of contact. Further studies are clearly now needed to distinguish between these alternatives and to firmly establish the role of this diffusible factor in the plant/AM fungal symbiotic dialogue.

	CONCLUSIONS

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION CONCLUSIONS MATERIALS AND METHODS LITERATURE CITED

This paper describes a novel system developed for the study of early signaling in AM fungal-root interactions. Our membrane-separated coculture approach provides the first evidence that AM fungi produce a diffusible symbiotic factor. This factor seems to be specific to AM fungi, and thus potentially important in a molecular dialogue associated with symbiotic infection. The transgenic root/reporter gene strategy that we have used to detect the plant response to the fungal factor, combined with our novel technique for separating the AM fungus from its host in axenic culture, should now provide the essential tools for the isolation and characterization of AM fungal diffusible factors. Recent breakthroughs in the cloning of receptor-like kinase genes (MtDMI2 and orthologous genes in alfalfa, L. japonicus, Melilotus alba, and pea) involved in Nod/Myc signaling (Endre et al., 2002; Stracke et al., 2002) now open the way to the molecular analysis of early recognition and transduction of Nod/Myc factor signals. In the foreseeable future, we can expect rapid advances in our understanding of early signaling events in AM infection and the mechanisms underlying the signaling pathways leading to mycorrhization.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION CONCLUSIONS MATERIALS AND METHODS LITERATURE CITED

Plant Material

Mycorrhizal interactions were studied in vitro using the Medicago truncatula Gaertn. excised roots (Ri T-DNA-transformed) and whole plants listed in Table II. Ri T-DNA-transformed roots derived from the Nod⁺/Myc⁺ M. truncatula transgenic line expressing the pMtENOD11-gusA fusion (L416) and the Nod/Myc mutant TR26 (dmi2-2) expressing the same fusion have been described previously (Chabaud et al., 2002). Analogous transformed root lines were obtained following Agrobacterium rhizogenes (isolate ARqua1-A4T) transformation (Boisson-Dernier et al., 2001) of C71 (dmi1-1) and TRV25 (dmi3-1) Nod/Myc mutant plants into which the pMtENOD11-gusA fusion had been introduced by genetic crossing (kindly provided by E.-P. Journet [IPM, Toulouse, France]). Before use, we confirmed that root clones possessed a normal non-symbiotic pENOD11-gusA expression pattern by staining for GUS (see below). Their Myc phenotype (absence of AM infection and fungal sporulation) was confirmed by staining roots for the presence of intraradical colonization (see below) 11 d after inoculation with G. rosea, and by examining fungal growth 6 weeks after inoculation and coculture with G. intraradices. By that time, dual culture with wild-type M. truncatula root cultures resulted in numerous mycorrhizal infection sites and de novo spore production (data not shown). All transformed roots were cultured vertically, at an angle of 70°, at 24°C in the dark (Fig. 1a; Chabaud et al., 2002) on M medium (in mg L¹: 80 KNO₃, 731 MgSO₄·7H₂O, 65 KCl, 4.8 KH₂PO₄, 288 Ca(NO₃)₂·4H₂O, 6 MnCl₂·4H₂O, 1.5 H₃BO₃, 2.65 ZnSO₄·7H₂O, 0.0024 Na₂MoO₄·2H₂O, 0.13 CuSO₄·5H₂O, 0.75 KI, 8 NaFe-EDTA, 3 Gly, 50 myo-inositol, 0.5 nicotinic acid, 0.1 pyridoxine-HCl, 0.1 thiamine-HCl, and 10,000 Suc) (Bécard and Fortin, 1988) containing 0.5% (w/v) gellan gum (Phytagel, Sigma-Aldrich, St. Louis).

When using seedlings, seeds of M. truncatula transgenic line L416 (Journet et al., 2001) were scarified with a 4-min treatment with concentrated H₂SO₄, surface-sterilized with 3% (w/v) calcium hypochlorite for 4 min, then germinated for 48 h at 15°C in the dark in petri dishes containing M medium gelled with 0.5% (w/v) gellan gum without Suc. All seedlings were grown axenically on M medium without Suc, using a culture system adapted from Wong and Fortin (1989) in which the aerial portion of the plant develops outside the petri dish, permitting optimal photosynthesis and gas exchange, while the root system remains in the humid, sterile environment of the petri dish (Fig. 1b). In brief, a 3-mm-diameter hole was burned into the top edge of a 90- × 90-mm square plastic petri dish, through which the 2-cm-long root of a germinated seed was inserted. After seedling insertion, petri dishes were sealed with parafilm, covered with aluminum foil, propped vertically at an angle of 70°, and cultured for 10 to 40 d in a controlled growth chamber at 24°C/20°C with a 16-h photoperiod and 3,200 cd m² light intensity.

Fungal Inoculum

Symbiotic and pathogenic fungal inocula used are listed in Table II. Gigaspora gigantea (Nicol. & Gerd.) Gerd. & Trappe, Gigaspora margarita Becker & Hall, and Gigaspora rosea Nicol. & Schenck spores were surface-sterilized and stored at 4°C according to Bécard and Fortin (1988). Sterile Glomus intraradices Schenck & Smith inoculum was stored in sterile water at 4°C. The phytopathogenic fungi Fusarium solani var phaseoli (Burkholder) Snyder & Hansen, Phoma medicaginis var pinodella (L.K. Jones) G. Morgan, Jones & Bunch, and Phytophthora medicaginis Drechs. were cultured on the above M medium with 1% (w/v) Suc for 2 weeks before use as inoculum in dual cultures.

Dual Cultures and Membrane Separation

Fast-growing Ri T-DNA-transformed root explants with a "fishbone" morphology were prepared according to Chabaud et al. (2002) for dual cultures. Three 5-cm-long explants were transferred to the upper one-third of 120- × 120-mm square petri dishes containing 50 mL of M medium, and fungal inoculum was added at the same time to the center of the petri dish (Fig. 1a). The AM fungal inoculum consisted of 10 surface-sterilized spores of Gigaspora spp. or approximately 500 sterile spores of G. intraradices pressed into the medium. Inocula of non-mycorrhizal fungi consisted of a 0.5-cm-diameter plug taken from 14-d-old cultures grown on M medium with Suc. Because mycelial growth of these fungi was significantly faster than that of AM fungi under our experimental conditions, the inoculated plants were harvested at 3, 6, and 10 dai. Dual cultures were incubated vertically, as for Ri T-DNA-transformed root cultures. After spore germination, AM fungal growth was observed daily using a dissecting microscope and marked on the petri dish using colored markers. Roots were harvested at 10, 20, and 30 dai to determine fungal infection and monitor pMtENOD11-gusA expression using histochemical GUS staining (see below).

Germinated seedlings were grown as described above in square petri dishes (90 × 90 mm) containing 25 mL of M medium without Suc, one seedling per petri dish. Dual cultures were incubated vertically, as described above for plant growth. At the time of seedling insertion, fungal inoculum was added as described above for root coculture inoculations (Fig. 1b).

To physically separate fungal inoculum from roots, membranes were inserted as a physical barrier. A cellophane membrane (Couvre-Confitures, Hutchinson, Chalette sur Loing, France) was generally used, or alternately, a polycarbonate membrane of 0.6-µm pore size (catalog no. 7062 4706, Whatman International, Maidstone, UK) or a dialysis membrane with a molecular mass cut-off of 3.5 kD (Spectra/Por, Spectrum Laboratories, Inc., Rancho Dominguez, CA). All membranes were rinsed with distilled water for 30 min and autoclaved in distilled water before use. For experiments with Ri T-DNA-transformed roots, fungal inocula were inserted into the medium in the center of a 120- × 120-mm square petri dish containing 50 mL of M medium (with Suc). The membrane (120 × 120 cm) was then laid on top, 25 mL of the same medium with Suc was then added, and finally, three rapidly growing root explants were transferred to the petri dish, as described above (Fig. 1a). This system was modified to increase the distance between the fungus and the roots, by adding first a cellophane membrane, followed by 25 mL of M medium with Suc, then a second membrane, followed by 25 mL of the same medium, and finally three explants. For experiments with seedlings, fungal inocula were inserted into the medium in the center of a 90- × 90-mm petri dish containing 25 mL of M medium without Suc, the membrane was laid on top, 12 mL of the same M medium without Suc was added, and finally, the single seedling (L416) was inserted into the petri dish as described above (Fig. 1b). For both plant and Ri T-DNA-transformed culture systems, fungus and roots were cocultured for 10 to 14 d and then harvested for study. All plant and transformed root experiments were performed at least three times, with three to five replicates per experiment. In all experiments, control roots were grown under identical conditions, including membrane inserted in the medium, but in the absence of fungal inocula.

Histochemical Localization of GUS Activity

Roots were evaluated for GUS activity after 6 h incubation at 37°C with the substrate 5-bromo-4-chloro-3-indolyl glucuronide, cyclohexylammonium salt (X-gluc, Biosynth AG, Staad, Switzerland), as described previously (Journet et al., 1994). To visualize the spatial relationship between fungal structures and pMtENOD11-gusA expression, whole-root systems were stained in situ by adding the X-gluc substrate-buffer combination directly to the petri dish (0.3 mg X-gluc mL¹ medium). The number of roots showing pMtENOD11-gusA induction and the total number of roots were counted in membrane-separated coculture experiments. Differences between AM treatments were assessed by pair wise Student's t Tests (Microsoft Word 97 SR-1, Microsoft, Redmond, WA), and differences between dmi mutants were compared by analysis of variance followed by Tukey's test (STATISTICA v6, StatSoft, Tulsa, OK) at the P = 0.05 significance level.

For histological observations, stained root segments were embedded in 3% (w/v) low-gelling temperature agar (type III, Sigma-Aldrich). Semithin sections (50-75 µm) were prepared using a Microcut H1250 vibrotome (Energy Beam Sciences, Agawam, MA), and observed immediately with bright-field microscopy (LEICA DM IRB/E, Leica Microsystems, Wetzlar, Germany).

Observation of Fungal Root Infection

To check for fungal colonization and to verify that the fungus had not crossed the membrane and contacted the roots in membrane-separated coculture, whole-root systems were cleared with 10% (w/v) KOH for 1 h at 90°C, thoroughly rinsed with distilled water, and stained with 0.05% (w/v) Chlorazol Black for 1 h at 90°C. Roots were destained overnight in 50% (v/v) glycerol and carefully examined by bright-field microscopy.

PCR Analysis of Potential Fungal Contamination

To further verify that the fungus had not crossed the cellophane barrier, the presence of fungal DNA in the root compartment was tested by PCR using fungus-specific and universal primers. Total DNA was extracted from liquid-nitrogen frozen plant and fungal tissue using the Wizard Genomic DNA Purification kit (Promega, Madison, WI) and stored at 4°C. DNA samples were collected from control roots, from roots grown in the presence of G. rosea but physically separated by a cellophane membrane, from roots colonized by G. rosea, and from spores of G. rosea. Two independent PCRs were performed concurrently on each DNA sample, using primers specific to the internal transcribed spacer (ITS) region in the nuclear ribosomal repeat unit. The first primer set, a so-called "universal" primer combination (ITS1 and ITS4) designed to amplify DNA from a broad range of organisms including fungi, plants, protists, and animals (White et al., 1990), was used to amplify plant and fungal DNA. The second primer set, ITS1-F coupled with a universal primer ITS4, was used to amplify fungal DNA more specifically (Gardes and Bruns, 1993). The following PCR conditions were used: denaturation at 93°C for 3 min, followed by 35 cycles of denaturation at 93°C for 30 s, annealing at 62°C for 1 min, and extension at 72°C for 1 min, with a final extension at 72°C for 10 min. Negative controls (no DNA template) were included in every experiment to test for contamination of reagents and reaction mixtures. DNA from three independent experiments was tested in this way, and similar results were obtained.