Dual requirement of the LjSym4 gene for mycorrhizal development in epidermal and cortical cells of Lotus japonicus roots

Authors


Author for correspondence: Paola Bonfante Tel: +39 011 650 2927Fax: +39 011 670 7459 E-mail: p.bonfante@csmt.to.cnr.it

Summary

• The LjSym4 mutation leads to Lotus japonicus plants that are defective in arbuscular mycorrhiza (AM) development.

• Two alleles of LjSym4 with different phenotypic strength are compared here. The development of AM was assessed by considering five parameters related to fungal structures present in root segments from wild-type and mutant plants. The distribution of intercellular hyphae was determined using semithin sections from resin-embedded roots. Cellular interactions were investigated ultrastructurally, whereas cell wall components from the host plant were identified using immunogold labeling.

• In roots of Ljsym4-1 mutant, fungal hyphae were mostly restricted to the intercellular spaces of the cortex, indicating a block to infection by mutant cortical cells, which resulted in a very low number of arbuscules.

• This observation suggests the presence of an additional, genetically defined ‘checkpoint’ for mycorrhizal development, located at the wall of cortical cells. The LjSym4 gene is therefore required for infection of both epidermal and cortical cells by AM fungi.

Introduction

Plant organisms do not act alone, but usually establish complex interactions with microbes that can range from beneficial to detrimental. Intensely studied interactions exist between plant and root symbionts that lead to the establishment of nitrogen-fixing nodules or mycorrhizas. Both are very important biologically and agriculturally. Symbiotic interactions cannot be addressed using Arabidopsis (The Arabidopsis Genome Initiative, 2000): developmental processes associated with legume nodule symbionts (rhizobia) and arbuscular mycorrhizas are therefore approached using different strategies (Downie & Bonfante, 2000; Stougaard, 2001). One of the most attractive is provided by the generation of mutant plants that are defective in one or both symbiotic properties, but which may grow well even if these developmental pathways are blocked. Many mutants have therefore been identified by screening the roots by microscopy and are being used to understand the biology of the symbiotic interaction. Particular attention was focused on mutants blocked at early stages of symbiotic interactions; some of these mutants are blocked for both rhizobial and mycorrhizal symbioses and others are specifically blocked for either one or the other. The discovery of such mutants has already suggested an overlap in the genetic programmes for the two diverse symbiotic interactions (Duc et al., 1989; Hirsch & Kapulnik, 1998; Albrecht et al., 1999) and their analysis offers a powerful tool to identify genetically defined steps in the development of the symbiotic interaction (Marsh & Schultze, 2001; Stougaard, 2001).

The final goal to clone and functionally characterize the genes responsible is hampered in some legumes, for example in pea, by their large genome size, while other plants as Lotus japonicus and Medicago truncatula seem to be more amenable models (Downie & Bonfante, 2000 and references therein).

A large collection of Lotus lines has been described, which are mutated in several loci (Marsh & Schultze, 2001; Stougaard, 2001) and where the infection process aborts following appressoria formation or prior to cortex invasion. In this case, hyphae penetrate epidermal cells but hyphal development is arrested following epidermis or exodermis colonization (Wegel et al., 1998; Parniske, 2000; Senoo et al., 2000). However, only a detailed cytological analysis of the mycorrhizal phenotype in the mutant line Ljsym4-2 demonstrated that the successful colonization process requires the active involvement of specific cell types: fungal penetration in epidermal cells is a prerequisite for normal colonization and arbuscule development in cortical cells (Bonfante et al., 2000). The presence of the fungus in the epidermal cells of the mutant Ljsym4-2 ultimately leads to the disappearance of cytoskeletal structures and to death of epidermal cells (Genre & Bonfante, 2002). These results, together with previous observations, showed that the weak allele of Ljsym4, identified as Ljsym4-1, allows occasional colonization of the inner cortex (Wegel et al., 1998; Bonfante et al., 2000). We therefore performed a cytological comparative analysis of this allele in order to verify whether the mutations of the locus LjSym4 are both stage specific and cell-type specific or alternatively whether they act at different times during the infection process, causing pleiotropic effects. Light microscopy observations performed so far on the whole roots of the mutant Ljsym4-1 and mutants with similar phenotypes (Senoo et al., 2000) did not allow a clear identification of the target cells.

Ultrastructural observations, together with a detailed quantification of fungal structures show that the weak allele Ljsym4-1 produces a novel mycorrhizal phenotype never yet described: not only is the fungus often blocked in the epidermal cells, as in Ljsym4-2, but the number of intercellular hyphae in the cortex significantly increases compared with the wild type; however, only a limited number of hyphae cross the cortical cell wall, leading to arbuscule formation. This suggests the presence of an additional genetically defined ‘checkpoint’ for mycorrhizal development, which is located at the wall of cortical cells.

Materials and Methods

Plant culture

Lotus japonicus seeds (Gifu, wild type and mutants Ljsym4-1 and Ljsym4-2) were surface-sterilized and scarified for 5 min in sulfuric acid, washed three times with sterile water and germinated on water agar (0.6%) in Petri dishes.

Details of the seed origin and of mutant line production are provided in Wegel et al. (1998) and Bonfante et al. (2000).

Mycorrhizal synthesis

Ten-day-old seedlings were inoculated with Gigaspora margarita Becker and Hall (strain deposited in the Bank of European Glomales as BEG 34) using the Millipore (Bedford, MA, USA) sandwich method (Giovannetti et al., 1993). Two seedlings were placed between two membranes (pore diameter 0.45 µm, Millipore), with 10–15 fungal spores (or without any spores for controls). Membranes containing the seedlings were planted in sterile acid-washed quartz sand in Magenta GA-7 vessels (Sigma, St Louis, MO, USA) and grown in climatic chamber at 22°C, 60% humidity, with 14 h of light per day. After 3 wks, samples from roots were cut after observation under a stereo microscope and processed for quantification, immunofluorescence and cytological analyses (see below). In order to obtain tissue samples with uniform age, segments were excised from the distal area of the root hair region. Root segments from 12 independent experiments and a total of 60 sandwich cultures were analysed.

Fungal staining and quantification

A total of 50 root segments (1 cm long) for each genotype were incubated overnight at room temperature in 0.1% Cotton blue (w : v) in lactic acid. Segments were then washed in lactic acid, and observed under the microscope for quantification assays. According to the method by Trouvelot et al. (1986), segments were classified into four categories depending on the percentage of segment length occupied by mycelium and by arbuscules. Five parameters were considered: (1) F%, reporting the percentage of segments showing internal colonization (frequency of mycorrhization); (2) M%, indicating the average per cent colonization of root segments (intensity of mycorrhization); (3) a%, quantifying the average presence of arbuscules within the infected areas (percentage of arbuscules); (4) A%, quantifying the presence of arbuscules in the whole root system (percentage of arbuscules in the root system); (5) the number of appressoria per centimeter of root.

The presence of intercellular hyphae and arbuscules was calculated on semithin sections from resin-embedded roots (see below).

Viability test

Fungal viability was evaluated by testing succinate dehydrogenase activity (Smith & Dickson, 1991). Roots were sectioned into 5 mm segments on ice and incubated overnight at 4°C in the following solution: 2.5 ml Tris-HCl 0.2 m pH 7.4; 1.0 ml MgCl2 5.0 mm; 2.5 ml sodium succinate 1.0 m; 4.0 ml distilled water; 10 mg nitroblue tetrazolium (NBT). The samples were then washed in distilled water and counterstained with acidic fuchsin (Brundrett et al., 1994). They were then observed under a light microscope.

Electron microscopy

Root segments from six independent experiments were fixed in 2.5% (v : v) glutaraldehyde in 10 mm sodium phosphate buffer (PBS, pH 7.2) for 2 h at 4°C. After rinsing with the same buffer, samples were postfixed in 1% (w : v) osmium tetroxide in double-distilled water for 1 h, washed three times with double-distilled water and dehydrated in an ethanol series (30, 50, 70, 90 and 100%; 10 min each step) at room temperature. The root segments were infiltrated in 2 : 1 (v : v) ethanol–LR White resin (Polysciences Inc., Warrington, PA, USA) for 1 h 1 : 2 (v : v) ethanol–LR White for 2 h and 100% LR White overnight at 4°C according to Moore & Staehelin (1998). Semithin sections (1 µm) were cut from 15 to 20 root embedded samples for each genotype (wild type, WT and mutant) in the presence and in the absence of the fungus. Sections were stained with 1% toluidine blue for morphological observations.

Thin sections were counterstained with uranyl acetate and lead citrate. Alternatively they were treated for immunogold reactions by using rat monoclonal antibodies (McAb) JIM5 and JIM7 for the location of nonesterified and methylesterified pectins, and a mouse McAb for detection of β-1,3 glucans. Experimental details are provided in Bonfante et al. (2000).

Results

Mycorrhizal phenotype of Lotus japonicus mutants

The mycorrhizal phenotype of the Gifu parental line and of the mutant Ljsym4-2 has been described in detail elsewhere (Wegel et al., 1998; Bonfante et al., 2000). Figure 1 briefly summarizes the most important features of the colonization process, as seen on semithin sections. The fungus proceeds through epidermal cells of the wild type to the outer cortex, via intracellular hyphae. They become intercellular at the contact with the inner cortical layer, where they re-enter and eventually produce abundant arbuscules (Fig. 1a). A large number of appressoria occurs at the surface of Ljsym4-2 roots (Fig. 1b), where the block is localized at the epidermis. Figure 1c shows a characteristic section from Ljsym4-1 roots, where the fungus is mostly found in the cortex as intercellular hyphae (C). However, a full sequence of the colonization events in Ljsym4-1 consists of: the fungus being blocked at the surface (Fig. 2a); penetrating the epidermal or outer cortical layers (Fig. 2b); reaching the inner cortex with many intercellular hyphae (Fig. 2c); and colonizing the cortical cell layer forming arbuscules (Fig. 2d). Abundant starch is present in the surrounding cortical cells.

Figure 1.

Semithin sections illustrate the colonization process in the wild type roots (a), in the LjSym4-2 mutant (b) and in the LjSym4-1 mutant (c). Details are provided in the text. Arrows point to penetration attempts in (b) and (c). A, arbuscules; C, cortical cells; E, epidermal cells; EH, extraradical hyphae. Bars, 30 µm.

Figure 2.

Semithin sections illustrate the different phenotypes shown by the LjSym4-1 mutant in the presence of Gigaspora margarita. The fungus is blocked at the surface (a), colonizes the epidermal and the first cortical layer (b), develops abundant intercellular hyphae (c) and very rarely produces arbuscules (d). Amyloplasts are abundant in the surrounding cortical cells. A, arbuscules; C, cortical cells; E, epidermal cells; EH, extraradical hyphae; CH, coiled intracellular hyphae; IH, intercellular hypha; N, host nucleus; a, amyloplast. Bars, 10 µm, 16 µm, 9 µm and 8.5 µm in (a), (b), (c) and (d), respectively.

The detail of the infection process for each genotype can be seen in Figs S1–S3.

Quantification of fungal structures

Data from the measurements of 50 root segments for each genotype were used to calculate four parameters indicating the intensity of root colonization by G. margarita (Trouvelot et al., 1986): (1) the frequency of mycorrhization, F%, reporting the percentage of segments showing internal colonization; (2) the intensity of mycorrhization, M%, indicating the average per cent colonization of root segments; (3) the percentage of arbuscules, a%, quantifying the average presence of arbuscules within the infected areas; (4) the percentage of arbuscules in the root system, A%, quantifying the presence of arbuscules in the whole root system. Table 1 shows the strong difference among the genotypes in terms of fungal colonization: the number of root segments with intraradical fungal structures (F%) was comparable in WT and Ljsym4-1, but the intensity of mycorrhization, M% (i.e. the average extension of colonized areas in the segment), was about fourfold higher in the WT than in the mutant. Even stronger differences are highlighted by the two last parameters, a% and A%, indicating that the frequency of arbuscules within the infected areas and within the whole root system are about fivefold and 23-fold higher, respectively, in the WT. Table 1 shows that the fungus colonizes a comparable number of segments in WT and mutant plants, with a rather homogeneous distribution pattern along the root system, but the fungal internal structures (arbuscules) are quantitatively different.

Table 1.  Parameters for the intensity of colonization in wild type and mutant Lotus japonicus roots
 F%M%a%A%
  1. F%, frequency of mycorrhization; M%, intensity of mycorrhization; a%, percentage of arbuscules; A%, percentage of arbuscules in the root system.

Wild type72.331.472.922.9
Ljsym4-163.3 7.9 8.9 0.7
Ljsym4-2 0 0 0 0

In order to verify whether the lower intensity of mycorrhization shown by Ljsym4-1 when compared with the WT was caused by a partial block of the colonization process, a quantification of the penetration points was first performed on the root segments of the two genotypes. The mutant Ljsym4-2 was included for a comparison. For this analysis, the number of appressoria per centimeter of root was quantified after cotton-blue staining. The results are summarized in Table 2.

Table 2.  Average number of appressoria per centimeter of root length
 Wild typeLjsym4-1Ljsym4-2
  1. Different letters indicate significantly different values according to the Kruskal–Wallis test of variance (P < 0.01).

Appressoria per cm0.13 (A)1.31 (B)2.17 (C)

The average values show a 10- and 20-fold higher number of appressoria per centimeter in the mutant lines compared with the wild type.

The second step focused on the intraradical fungal structures: a thorough survey of the infected areas was performed by microscope observation of semithin sections (Table 3). A total of 50 root sections for the Ljsym4-1 genotype and 30 root sections for the WT, all containing fungal structures, were observed. Three parameters were considered: Ap : I, reporting the ratio between appressoria and intercellular hyphae; I : Ar, reporting the ratio between intercellular hyphae and arbuscules; and R%, reporting the percentage of root sections hosting arbuscules. Because intraradical colonization was never observed in mutant 1749 (Bonfante et al., 2000), this line Ljsym4-2 was not included in this experiment.

Table 3.  Parameters reporting the pattern of intraradical fungal distribution in the two Lotus japonicus genotypes
GenotypeAp : II : ArR%
  1. Ap : I, ratio between the number of appressoria and intercellular hyphae; I : a, ratio between the number of intercellular hyphae and arbuscules; R%, percentage of root sections containing arbuscules. Different letters indicate significant differences according to the Kruskal–Wallis test of variance (P < 0.01).

Wild type0.01 (A)1.26 (B)62.75 (D)
Ljsym4-10.11 (A)3.11 (C)21.42 (E)

The WT roots show an almost threefold more extensive presence of arbuscules (R%), while the I : Ar ratio indicates that almost two more intercellular hyphae per each arbuscule are present in the mutant line. By contrast, a statistically nonsignificant difference (0.10) was found for the ratio Ap : I in WT and mutant plants.

Fungal viability

A viability test was performed in order to verify whether the higher number of penetration attempts present in the mutant lines (Table 2) was effective or abortive.

In order to obtain a comparable set of data, only extraradical hyphae and appressoria were considered, as the different extension of intraradical fungal structures could unbalance the test. Ten segments (5 mm long) were analysed for each genotype. The average data are reported in Table 4. The analysis of root segments stained with NBT and acidic fuchsin showed that in the WT, the majority of hyphae are vital, while in Ljsym4-1 and Ljsym4-2 vital hyphae are more restricted.

Table 4.  Average data from the fungal viability test performed on the three genotypes
 Average number of vital appressoria per cmAverage number of nonvital appressoria per cmAverage percentage of vital extraradical mycelium
  1. Different letters indicate significant differences according to the Kruskal–Wallis test of variance (P < 0.01).

WT1 (C)0 (F)73.75% (I)
Ljsym4-11.45 (A)1.82 (D)30.90% (G)
Ljsym4-20.37 (B)3.37 (E) 3.75% (H)

Plant–fungal cell interactions in the mutant Ljsym4-1

After a profuse growth at the root surface of the mutant Ljsym4-1, G. margarita starts an infection process very similar to that described for the WT and the allelic mutant, Ljsym4-2 (Bonfante et al., 2000). An intercellular hypha – produced by the appressorium – separates two adjacent epidermal cells (Fig. 3a,b) forming a penetration peg into the epidermal cells (Fig. 4a,b).

Figure 3.

Gigaspora margarita separates two epidermal cells and penetrates into both (a, semithin section under the light microscope) or a single one (b, thin section under the electron microscope). In all the cells, the host nuclei are located close to the intracellular hypha, which is surrounded by the invaginated host membrane (arrows). E, epidermal cells; C, cortical cells; PH, penetrating hypha; CH, coiled intracellular hyphae; N, host nucleus. Bars, 8.5 µm in (a) and 5 µm in (b).

Figure 4.

Magnifications of Fig. 3b. showing the ultrastructural details of the early colonization events. (a) The penetrating hypha which enters the epidermal cell is vital with small nuclei and electron transparent vacuoles. Bar, 3.5 µm. (b) The area surrounded by the white square is further magnified to illustrate the thin fungal peg which starts the penetration event at the base of the epidermal cells. It causes a host wall loosening and host membrane invagination. Host wall material is poorly labeled by the monoclonal antibody (Mab) JIM7 (arrow heads). Bar, 0.5 µm. (c) The fungus occupies a part of the epidermal cell with a coiled hypha which contacts the underlying cortical cell. Bar, 2.5 µm. (d) The area surrounded by the white square is further magnified to show the membranous body in the fungal cytoplasm and the limited host wall thickening, which is poorly reactive to Mab JIM7 (arrow heads). Bar, 0.5 µm. E, epidermal cells; C, cortical cells; PH, penetrating hypha; CH, coiled intracellular hyphae; PL, host invaginated membrane; FM, fungal membranes.

This last event is associated with a proliferation of a host-derived membrane that surrounds the fungus (Fig. 3b) and with the movement of the host nucleus towards the fungus (Fig. 3). The epidermal cell is vital, with a large vacuole and peripheral organelles. Immunogold labeling with antibodies targeted to host cell wall components revealed a loose presence of methylated pectins, as detected by the monoclonal antibody (Mab) JIM7, and more abundant nonesterified pectins, revealed by the Mab JIM5 (not shown). At the fungus–plant wall contact point, limited thickenings were observed, both when the intercellular hypha entered the epidermal cell (Fig. 4a,b) and when the intracellular hypha contacted the underlying cortical cell to move towards the inner layers (Fig. 4c,d). Antibodies to locate β-1,3 glucans did not label these wall thickenings (not shown).

Even if rare (Table 1) arbuscules presented a regular morphology (Fig. 5a), with each arbuscule trunk and branch being surrounded by the invaginated host membrane (Fig. 5b). Pectins were present in the interfacial material laid down between the fungal wall and the host membrane.

Figure 5.

Ultrastructural details of an arbuscule colonizing a cortical cell. (a) A large arbuscule trunk and thin arbuscular branches which fill up the cell volume; monoclonal antibody (Mab) JIM 5 regularly labels the middle lamella (arrow heads). Bar, 1.5 µm. (b) A detail of the plant–fungal interface; mab JIM 5 labeling indicates the presence of pectic material (arrowheads) in the space between the fungal wall and the plasma membrane. IH, intercellular hypha; AT, arbuscule trunk; A, arbuscular branches; PL, host invaginated membrane; FW, fungal wall. Bar, 0.2 µm.

More often, cortical tissue only contained intercellular hyphae: they were distributed along the corner of the outer or inner cortical cells (Figs 2c, 6a, 6b and 7). The intercellular hyphae proliferating among the outer cortical cells were vital, with a high number of nuclei involved in mitotic events (Fig. 6b,c). By contrast, those present in the inner cortical spaces were more vacuolated, and penetration attempts into the cortical cells were often present (Fig. 7a), eventually leading to collapsing hyphal segments (not shown). Surrounding cortical cells regularly presented abundant starch in their amyloplasts (Fig. 2d). Electron-dense material was often stored in these cortical cells (Fig. 7a,b). A quantitative analysis on the number of cells containing electron-dense phenol-like material, demonstrated that a higher number of these cells was present in each root section of the mutant vs the WT (2.36 cells vs 1.62 cells per section). However, the difference was not significant according to the Kruskal–Wallis test of variance (P > 0.01).

Figure 6.

Intercellular hyphae: ultrastructural details. (a) Intercellular hyphae spread in the cortex and surround vacuolated cortical cells. They are seen at the corner of the cells with the typical triangle profile or are longitudinally sectioned. In all the situations they are rich in nuclei (b), which seem to be involved in simultaneous mitosis. (c) Magnification of (b) with the two still-joined nuclei. Monoclonal antibody (Mab) JIM 5 regularly labels the middle lamella (arrow heads). IH, intercellular hypha; C, cortical cells; Nf, nucleus of the fungus. Bars, 3, 1 and 0.35 µm, respectively.

Figure 7.

Intercellular hyphae: ultrastructural details. (a) Attempts to penetrate by an intercellular hypha; there is a loosening of the host wall (arrow heads) with a localized thickening. Bar, 0.5 µm. (b) The surrounding cortical cells often show an electron-dense content which is present in the vacuole or packed in dense bodies. IH, intercellular hypha; C, cortical cells; N, host nucleus; HT, host microtubules; V, vacuole; DM, electron-dense deposits. Bar, 2.5 µm.

When the fungus was blocked in the epidermal cells, a degeneration of the host cell was observed with the following features: detachment of the plasma membrane from the cell wall, suggesting an incipient plasmolysis, organelle breakdown, nuclear degradation and, finally, cell death, as already found in the other mutant, Ljsym4-2 (not shown).

Discussion

The analysis of the weak mutant allele, Ljsym4-1, of the LjSym4 gene made it possible to reveal a succession of ‘checkpoints’ in mycorrhizal development (Fig. 8).

Figure 8.

Scheme representing the steps of root colonization in wild type and mutant lines of Lotus japonicus. The diagram highlights the three ‘checkpoints’ outlined by the present research: the extra-epidermal block, the intraepidermal block and the cortical block, which are either effective or overcome in the different genotypes (different arrow thickness). The effect of the blocks on the upstream steps are represented by the gray arrows on the right side.

First, we identified an epidermal block, which is located on the surface or inside the epidermal cells; when these ‘checkpoints’ are overcome, another block is present in the cortical tissue, hampering the cell penetration and the subsequent arbuscule development. The epidermal block corresponds to the one found in Ljsym4-2, even if in this latter mutant the block is more effective, as demonstrated by the highest number of appressoria, the lowest fungal viability and the death of all the epidermal cells experiencing fungal penetration (Bonfante et al., 2000 and present data). This major surface block has also been identified in the Pisum sativum mutant, E107 line (Resendes et al., 2001). However, in Ljsym4-1 the epidermal block may be overcome, the fungus may move towards the outer cortex and become intercellular. On the basis of the quantitative observations (Table 3) in Ljsym4-1, the cortical block seems to be stronger than the epidermal block. The cytological analysis adds some indirect evidence showing unexpected features at this step: the proliferation of intercellular hyphae is probably performed through multiple nuclear divisions, which have so far been only rarely described (Bianciotto et al., 1995). This aspect provides a convincing explanation for the high number of root segments showing intraradical fungal structures reported in Table 1. The fungus still attempts to penetrate the cortical cells, but success is very rare, as demonstrated by the low number of arbuscules. As a consequence, many senescing intercellular hyphae are found in the inner cortical layers. Irrespective of their low number, arbuscules show a regular morphology, leading to a phenotype that strongly differs from that described in the pea mutant RisNod24, where stumpy arbuscules are formed (Gianinazzi-Pearson et al., 1991).

The presence of amyloplasts in all the cells surrounding the intercellular hyphae suggests that the sugar stock is not available for the fungus. It is well known that intraradical fungus uses glucose from the plant (Bago et al., 2000) and amyloplasts do not tend to accumulate in functional mycorrhizas. Fungal infection might induce a stronger sink for sugars in the mutant roots which are weakly released to the intercellular hyphae, leading to starch accumulation.

The second feature is an accumulation of phenolic material in cortical cells contacted by intercellular hyphae. Interestingly, these cells do not die and differ from epidermal cells, which show a sort of hypersensitive response (Heath, 1997). In addition, cortical cells show no relevant modifications in their cell wall structure or cell wall components after immunogold labeling.

To our knowledge, this is the first example of a comparative analysis between two alleles in a defective mycorrhizal plant. Catoira et al. (2000) compared the nodulating phenotype of a few alleles in Medicago truncatula. They found the same responses to Nod factor, indicating that they were the result of a mutation in the same gene in each case. Only one case of allelic variation was detected with the mutant lines B85 and C54, which differed slightly in their root-hair branching phenotype (Catoira et al., 2000). Interestingly, the root hair response of the two Ljsym4 mutant lines to Mesorhizobium loti inoculation was significantly different. This was analysed using a strain harboring a lacZ gene under the control of a constitutive promoter (Leong et al., 1985). Whereas root hairs of mutant Ljsym4-2 responded exclusively with tip swelling to rhizobial inoculation (Bonfante et al., 2000), mutant Ljsym4-1 showed pronounced and exaggerated root hair branching. This result underpins the notion that Ljsym4-1 might be a weaker mutant allele than Ljsym4-2. Mesorhizobium loti was never observed to induce infection threads on either Ljsym4-1 or Ljsym4-2 mutants.

Distinct morphological stages have been defined for the development of the AM symbiosis (Smith & Read, 1997). The following abbreviations have been used for the description of phenotypes of mutant legumes defective in their mycorrhizal capacities (Marsh & Schultze, 2001). Pen: the fungus is blocked before the root penetration. Coi: penetration of epidermal cells, but absence of cortex invasion. Only a few cases are known showing absence of inner cortex invasion (Ici), absence of arbuscules (Arb) or abnormal arbuscule development (Ard).

The detailed quantitative and morphological analysis of the mutant line of L. japonicus Ljsym4-1 demonstrates that this line does not fit a single category but a combination of phenotypic categories: Ljsym4-1 shows Pen, Coi, Ici and Arb features. It may therefore be compared with the pea mutant E107, which was first described as a low nodulating mutant and recently identified as a low mycorrhizal mutant, and described as Apr+, low Pen, low Coi (Resendes et al., 2001).

Our investigation, together with that on E107 pea, clearly shows how quantification of the infection process is crucial to evaluate the effect of the mutation on the fungal development and more precisely on the specific structures it forms (appressoria, intercellular hyphae, intracellular hyphae and arbuscules). In addition, cytological details of the interactions between plant and fungus are very important in order to identify the cellular target of the mutation. This approach made it possible to reveal that LjSym4 gene is required for infection of both epidermal and cortical cells by AM fungi, therefore extending our previous observations on the allele Ljsym4-2, where only the role played by epidermal cells was highlighted (Bonfante et al., 2000).

Different models have so far been proposed to locate the mutations along the pathways of plant–microbe symbioses (Catoira et al., 2000; Walker et al., 2000; Stougaard, 2001). The positioning of LjSym4 expression is not defined precisely: it likely acts after the Nod factor perception, and therefore downstream genes involved in such process (Ljsym1, LjSym5, PsSym10), but perhaps upstream the activation of Nsp genes, since the nodulating capacity is missing in both the alleles. Mycorrhization seems therefore to be under the control of multiple mechanisms, which may be partly unrelated with the nodulating pathways.

The availability of nonlegume plants defective in mycorrhizal properties such as, the tomatoes first described by Barker et al. (1998) and the new mutants developed by David-Schwartz et al. (2001) all suggest a more complex mechanism. The tomato rmc mutant shows a range of phenotypes (from aborted penetration to relatively normal colonization) depending on the fungal species (Gao et al., 2001). Irrespective of the role played by each fungal isolate, which have been demonstrated to be evolutionary distant (Morton & Redecker, 2001) and – as a consequence – theoretically capable of producing different signal molecules, the data of Gao et al. (2001), together with our present results, confirm that the mechanism leading to mycorrhizal establishment is a multi-step process, influenced by factors of fungal and host origin. This type of process is probably based on a threshold value for signals that change depending on the host cell type and/or on the fungal genotype, allowing a specific fungal strain to partly compensate for a mutation in a symbiosis gene of the host plant. This might represent one of the keys of AM fungal success.

Acknowledgements

This research was funded by grants to PB from the Università di Torino (Fondi di Ateneo) and the Consiglio Nazionale delle Ricerche. Thanks to Mr Silvano Panero (LMA–Dipartimento di Biologia Vegetale dell’Università di Torino) for his collaboration in micrograph preparation.

Supplementary material

The following material is available from http://www.blackwell-science.com/products/journals/suppmat/NPH/NPH424/NPH424sm.htm

Fig. S1 Colonisation of Lotus japonicus Wild Type Gifu by Gigaspora margarita. The fungus proceeds through epidermal cells to the cortex, where arbuscules are formed.

Fig. S2 Colonisation of Lotus japonicus Ljsym 4-1 mutant by Gigaspora margarita. The fungus is blocked at the root surface, but can penetrate the epidermal and outer cortical layers.

Seldom, the fungus can reach the inner cortex with many intercellular hyphae and occasionally colonise the inner cortical cells forming.

Fig. S3 Colonisation of Lotus japonicus Ljsym 4-2 mutant by Gigaspora margarita. A large number of branched appressoria occurs at the surface of the root, where the block is localised at the epidermis.

Ancillary