Presymbiotic factors released by the arbuscular mycorrhizal fungus Gigaspora margarita induce starch accumulation in Lotus japonicus roots


Author for correspondence:
Paola Bonfante


  • • Nutrient exchange is the key symbiotic feature of arbuscular mycorrhiza (AM). As evidence is accumulating that plants sense presymbiotic factors from AM fungi and prepare for colonization, we investigated whether modifications in plant sugar metabolism might be part of the precolonization program.
  • • Inoculation of Lotus japonicus roots in a double Millipore sandwich with the AM fungus Gigaspora margarita prevented contact between the symbionts but allowed exchange of signal molecules. Starch content was used as a marker for root carbohydrate status.
  • • Mycorrhizal colonization of L. japonicus roots led to a decrease in starch concentration. In roots inoculated in the double sandwich, the polysaccharide accumulated after 1 wk and persisted for at least 4 wk. The response was absent in the castor myc mutant, sym4-2, while transcript levels of both CASTOR and POLLUX were slightly enhanced in the wild-type L. japonicus roots, suggesting a requirement of the corresponding proteins for the starch-accumulation response. Exudates obtained from fungal spores germinated in the absence of the plant also induced starch accumulation in wild-type L. japonicus roots.
  • • We conclude that factors released from germinating AM fungal spores induce changes in the root carbon status, possibly by enhancing sugar import, which leads to starch accumulation when colonization is prevented.


Members of most land-plant families and fungi of the group of Glomeromycota unite to form the most widespread symbiosis on land called arbuscular mycorrhiza (AM) (Smith & Read, 2008). The most important mutual benefit is nutrient exchange. The fungi efficiently take up inorganic ions (especially phosphate) from the soil and provide these to the plant (Karandashov & Bucher, 2005), while the plant supplies the biotrophic fungus with carbohydrates produced as a result of photosynthesis (Harrison, 1999). Dependence on the plant's carbon supply most probably makes the fungus an obligate symbiont (Smith & Read, 2008). The fact that light-limitation results in decreased photo-assimilate production by the plant and in limited mycorrhizal colonization (Koide & Schreiner, 1992) supports this view. Recent transgenic approaches also showed that carbon under-supply of the root leads to a lower degree of colonization (Schaarschmidt et al., 2007a,b). More direct evidence was provided by a carrot root split-plate system in which AM fungi (AMF) took up glucose exclusively from the intraradical environment and not at all from the growth medium via extraradical hyphae (Pfeffer et al., 1999).

Ho & Trappe (1973) were the first to discover that 14C-labelled photosynthate from the plant was transported to extraradical hyphae of an AM fungus and this has been confirmed by others (reviewed in Bago et al., 2000). In fact, the AM fungus enhances root sink strength for carbon and up to 20% more photosynthate is transported to the roots of mycorrhizal faba bean and clover compared with control noninfected plants (Wright et al., 1998 and references therein).

A presymbiotic molecular dialogue between the plant and the AM fungus precedes root colonization. A plant signal that induces germination and hyphal branching of AMF has been characterized as a strigolactone (Akiyama et al., 2005; Besserer et al., 2006). The existence, but not yet the identity, of one or more fungal signals has been shown by transcriptional and developmental reactions of plant roots to AMF in spite of separation of the symbionts by a semipermeable membrane. A ß-glucuronidase (GUS) reporter of the early nodulin gene (ENOD11) was induced in Medicago truncatula hairy roots in this manner (Kosuta et al., 2003) and lateral root formation was elevated in M. truncatula plants (Olah et al., 2005). A transient increase in calcium in response to fungal exudates was shown in soybean cell culture cells (Navazio et al., 2007), and calcium-spiking was observed in epidermal cells of M. truncatula roots in the vicinity of branching AMF hyphae (Kosuta et al., 2008). These responses seem to be mediated via partly different signalling components as ENOD11 induction is independent of the common symbiosis (SYM) proteins (Does Not Make Infections) DMI1, DMI2 and DMI3 (corresponding to POLLUX, SYMRK and CCAMK in the Lotus japonicus nomenclature; Stracke et al., 2002; Imaizumi-Anraku et al., 2005; Tirichine et al., 2006), whereas lateral root formation and calcium-spiking are dependent on functional DMI1 and DMI2 (Kosuta et al., 2003, 2008; Olah et al., 2005).

As carbohydrate supply by the plant is vital for completion of the AM fungal life cycle, we wondered whether carbon transport to the root is induced before physical contact between the symbiotic partners. We co-inoculated L. japonicus with the AMF Gigaspora margarita, separating roots and fungus by a semipermeable membrane in the ‘double-sandwich’ set up (Fig. 1). We demonstrated that in this set up starch accumulated in L. japonicus roots. The response was dependent on CASTOR, a functional twin of POLLUX whose putative orthologue in M. truncatula and soybean has only recently been identified (Chen et al., 2009). Starch accumulation was also observed in response to exudates from germinating G. margarita spores produced in the absence of the plant.

Figure 1.

(a) Set up of the double sandwich. (b) Fungal growth on the nitrocellulose membrane in the double sandwich 1 and 4 wk postinoculation (wpi). Bars, 1 mm.

Materials and Methods

Plant cultures and mycorrhizal inoculation

Lotus japonicus (Regel) Larsen seeds (Gifu, wild type and mutant Ljsym4-2 defective in CASTOR) were surface sterilized and scarified for 3 min in sulphuric acid, washed three times for 15 min each with sterile, distilled water and germinated on water agar (0.6%) in Petri dishes, as described in Bonfante et al. (2000).

Mycorrhizal colonization was induced as described in Novero et al. (2002). For the double sandwich, two seedlings were placed between two nitrocellulose membranes, as described before, while 15 (in total 30) spores were placed at both external sides of the sandwich and were then covered with further membranes (Fig. 1a). The double sandwich was placed in sterile quartz sand.

Periodic acid-Schiff staining and starch extraction

Periodic acid-Schiff (PAS) staining was performed, according to Ruzin (1999), on semithin (1 µm) microtome sections prepared as described by Novero et al. (2002). Root sections were cut from fragments obtained from secondary roots cut 1 cm above the root tips. Micrographs from sections were taken under light microscopy (Nikon Eclipse E400; Nikon, Tokyo, Japan) using a digital camera (Nikon Coolpix 4500; Nikon, Tokyo, Japan). The number of amyloplasts in each root section was determined by counting. Amyloplast numbers in controls were set at 100% and those in treated roots were expressed in relation to this. Two independent experiments were performed.

For starch extraction, excised root systems were rapidly blot-dried on filter paper and weighed. Samples were then frozen in liquid nitrogen, transferred to 2-ml Eppendorf tubes (Eppendorf, Hamburg, Germany) and thoroughly homogenized using a pestle in liquid nitrogen. The samples were further homogenized in 0.5 ml of absolute ethanol. After addition of 0.5 ml of 80% ethanol, the tubes were incubated at 70°C for 90 min and then centrifuged for 10 min at 11 337 g and the pellet was resuspended in 1 ml of 80% ethanol. Two more washings were performed with 1 ml of 80% ethanol (and 10 min of centrifugation). The pellets were finally resuspended in 400 µl of 0.2 m KOH and incubated at 95°C for 60 min. After neutralization with 70 µl of acetic acid, the samples were centrifuged for 10 min and the supernatant was used for starch quantification (Starch Test-Combination enzymatic analysis kit, cat. no. 207748; Boehringer, Mannheim, Germany;), according to the manufacturer's instructions. At least three independent experiments, including at least three plants each, were performed to obtain all results of enzymatic starch quantification.

Production of fungal exudates and treatment of plants

G. margarita spore exudates were produced as described by Navazio et al. (2007). The germination water was removed and concentrated 5- or 10-fold in a SpeedVac (KNF Laboport, Neuberger). Seedling roots were placed in 1 ml of concentrated germination water in a 2 ml microcentrifuge tube and grown in a climatic chamber at 22°C in 60% humidity with 14 h of light per day. After 1 wk, plants were harvested and total starch from roots was quantified as described in the previous section.

Real-time reverse transcription–polymerase chain reaction

For use in the real-time reverse transcription–polymerase chain reaction (RT-PCR), RNA was extracted using the pine tree protocol (Chang et al., 1993) from two root systems per sample of each treatment. Two independent biological replicates were performed. Reverse transcription and real-time RT-PCR were performed as described by Guether et al. (2009). CT values of all genes were normalized to the CT values of ubiquitin (TC3806) in each sample. Primers used for the amplification of ubiquitin were as follows: forward, 5′-TTCACCTTGTGCTCCGTCTTC-3′; and reverse, 5′-AACAACAGCACACACAGACAATCC-3′. The elongation factor of G. margarita (GI:37653264) was amplified using the following primers: forward, 5′-TGAACCTCCAACCAGACCAACTG-3′; and reverse, 5′-GGTAAGACCAACTGGGGCGAATG-3′. For CASTOR and POLLUX amplification, primer sequences were taken from Imaizumi-Anraku et al. (2005).


Probability values were calculated using the statistical software systat 10. The Kruskall–Wallis test for nonparametric data was used to analyze the results of most experiments. Data for starch accumulation in response to spore exudates produced in the absence of the plant were calculated using the Bonferroni post-hoc test of an ANOVA.


Root starch responses to AM colonization and presymbiotic AM fungal signals

Four weeks after inoculation with G. margarita in the Millipore sandwich system, L. japonicus roots were colonized along approx. 70% of their total length (Novero et al., 2002). At this time point, the starch content in L. japonicus wild-type roots was about half the amount of the mock-inoculated control (Fig. 2a). In the myc mutant, sym4-2, which is defective in the CASTOR gene and does not allow cortex colonization and the formation of arbuscules (Imaizumi-Anraku et al., 2005), no difference in starch content was found (Fig. 2a).

Figure 2.

Root starch diminishes upon arbuscular mycorrhizal (AM) colonization. (a) Comparison of the mean (± SE; n = 9; Kruskall–Wallis test; *, P ≤ 0.05) quantities of starch between Lotus japonicus wild-type and castor mutant control roots (open bars) and roots colonized by Gigaspora margarita (grey bars) at 4 wk post inoculation (wpi). Colonization by G. margarita for 4 wk leads to significant (n = 8; Kruskall Wallis test; *, P ≤ 0.05) decreases in amyloplast numbers. (b) Cross-sections of wild-type L. japonicus roots after resin-embedding and staining by the periodic acid-Schiff (PAS) reaction for the visualization of amyloplasts (indicated by arrows). For quantification, two independent experiments were performed on four plants each. Each of the eight plants was represented by five cross-sections. Amyloplast numbers are expressed as a percentage of the control and are displayed in the lower left corner of images. A, arbuscule. Bars, 60 µm.

Quantification of amyloplast numbers in root sections confirmed this observation. Amyloplasts mostly accumulated in cortical cells (Fig. 2b) and were absent from arbuscule-containing cells of mycorrhizal roots (Fig. 2b). Quantitative image analysis showed that significantly fewer amyloplasts (less than half) (P ≤ 0.05) were found in colonized roots compared with roots from noncolonized controls (Fig. 2b). No difference between the G. margarita-inoculated and -noninoculated castor mutant was observed (data not shown). Taken together, the results suggest that starch is diminished in mycorrhizal roots as a result of carbohydrate uptake by the AM fungus.

To understand whether the modification in root sugar metabolism begins before fungal colonization, we measured starch, as a marker for carbon supply, in roots that were separated from the AM fungus by a semipermeable membrane (double sandwich, Fig. 1a). In this set up the spores of G. margarita germinated within 1 wk and after 4 wk the germinating mycelium completely covered the nitrocellulose filter membrane (Fig. 1b). The quantity of starch doubled in roots growing in the presence of G. margarita, as compared with the control, within 1 wk of treatment in the double sandwich (Fig. 3a). The increase in starch persisted until at least 4 wk of treatment (Fig. 3a). In G. margarita and mock-inoculated roots grown in the double-sandwich set up, amyloplasts had a cortical localization, as in the single sandwich. However, the number of amyloplasts had almost doubled (P < 0.05) after 4 wk in the presence of the fungus as compared with the mock-inoculated control (Fig. 3c).

Figure 3.

Root starch increases in Lotus japonicus in response to diffusible signals released by Gigaspora margarita. (a, b) Comparison of the mean (± SE; n = 9; Kruskall–Wallis test; *, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001) starch quantities (a) at 0, 1, 2, 3, or 4 wk postinoculation (wpi) in wild-type roots and (b) at 1 and 4 wpi in castor (sym4-2) inoculated in the double sandwich in the presence and absence of G. margarita. (c) Double sandwich treatment for 4 wk leads to a significant (n = 8; Kruskall–Wallis test; *, P ≤ 0.05) increase in amyloplast numbers in the presence of G. margarita compared with the control. Cross-sections of wild-type L. japonicus roots after resin-embedding and staining by the periodic acid-Schiff (PAS) reaction for the visualization of amyloplasts (indicated by arrows). For quantification, two independent experiments were performed on four plants each. Each plant was represented by five cross-sections. Amyloplast numbers are expressed as a percentage of the control and are displayed in the lower left corner of the images. Bars, 50 µm. (d) Comparison of the mean (± SE) quantities of starch in wild-type L. japonicus control roots (n = 24) and in roots after treatment with 5× (n = 15) and 10× (n = 24) concentrated exudates from germinating G. margarita spores. Different letters indicate significant differences (Bonferroni post hoc test of an ANOVA; F-ratio = 12.664; P = 0.000).

To test whether presymbiotic signalling, leading to starch accumulation in roots, was dependent on the common SYM signalling pathway, we measured starch accumulation in the L. japonicus castor mutant, sym4-2. In this mutant, starch did not accumulate in the double sandwich at 1 and 4 wk postinoculation (wpi; Fig. 3b). The same was found for amyloplasts at 4 wpi (data not shown).

We conclude therefore that factors produced by the AMF and passing through the nitrocellulose filter membrane (pore size 0.45 µm) induce starch accumulation in wild-type L. japonicus roots. This response depends on CASTOR.

Starch accumulation upon treatment with AM fungal germination water

To examine whether the production of fungal signalling factors that trigger starch accumulation was induced by the presence of the plant (Parniske, 2005), L. japonicus seedlings were treated with the water in which fungal spores had germinated for 5 d in the absence of the plant. After 1 wk of treatment, root starch concentrations had almost doubled (although this was not significant) when plants were treated with G. margarita germination water concentrated 5×, and had increased by eightfold (P = 0.000) when treated with germination water concentrated 10× (Fig. 3d). Therefore, G. margarita produced molecules, independently of plant signals, that act as signals to induce starch accumulation in plant roots.

Transcriptional response of CASTOR and POLLUX

CASTOR and POLLUX are two highly homologous proteins, belonging to the common SYM signalling pathway, that are indispensable for AM formation (Imaizumi-Anraku et al., 2005; Parniske, 2008). Although it has been reported that expression of CASTOR and POLLUX does not change upon mycorrhizal colonization (Gutjahr et al., 2008; Chen et al., 2009), we tested whether the CASTOR and POLLUX genes would transcriptionally respond to AMF signals and AMF presence in L. japonicus roots.

Twenty-four hours of treatment with AM fungal exudates produced in the absence of the plant did not lead to any change in expression of CASTOR and POLLUX (Fig. 4a). However, in the double sandwich at 4 dpi, when the spores had just begun to germinate, CASTOR and POLLUX were slightly induced in roots in the separated double sandwich but not upon direct contact with the germinating spores. The presence of germinating spores even led to a slight repression of POLLUX at 4 dpi (Fig. 4b, c). At full mycorrhizal colonization (28 dpi), CASTOR and POLLUX expression was equal to that of the control (Fig. 4b, c).

Figure 4.

Transcriptional response of CASTOR and POLLUX genes to fungal signalling molecules. (a) CASTOR and POLLUX expression in Lotus japonicus roots after 24 h of treatment with signalling factors (grey bars) produced by Gigaspora margarita spores in the absence of the plant compared with the water control (open bars). (b, c, d) Gene expression at 4 days postinoculation (dpi) in the double sandwich (dark grey bars) and at 4 and 28 dpi in the single-sandwich set up, allowing colonization by G. margarita (light grey bars) compared with nonmycorrhizal control roots (open bars), as assessed by real-time reverse transcription–polymerase chain reaction (RT-PCR). (b) LjCASTOR transcript level, (c) LjPOLLUX transcript level, (d) transcript level of G. margarita elongation factor. Three technical replicates were performed for each sample (n = 2).

To control for the separation of plant and fungus by the membrane in the double sandwich and for the absence of colonization at 4 dpi, but the presence of colonization at 28 dpi, in the single-sandwich set up, we included a constitutively expressed gene from G. margarita, the elongation factor GI:37653264 (GmEF), in the real-time RT-PCR analysis. No fungal transcript was detected in any set up at 4 dpi or in the mock-inoculated control at 28 dpi. The GmEF transcript level was high only in the single-sandwich set up at 28 dpi (Fig. 4d).


Carbohydrate nutrition appears to be the main reason that AMF develop symbiotic relationships with host plants because AMF are incapable of taking up glucose from the surrounding medium (Pfeffer et al., 1999). Evidence is accumulating that plants are able to sense as yet unknown AMF signalling molecules (Kosuta et al., 2003, 2008; Olah et al., 2005; Navazio et al., 2007) to prepare for symbiont accommodation (Genre et al., 2005). Here we demonstrate that a change in carbohydrate metabolism is one of these preparative reactions. We tested root starch content as a marker of root carbohydrate status and found that the concentration of starch is differently modulated depending upon whether the fungus develops inside the root or is perceived by the plant before colonization.

Root starch decreases upon AM colonization but accumulates upon presymbiotic signalling

Quantification of total extracted starch and amyloplasts demonstrated that the concentration of starch decreased in the mycorrhizal roots of wild-type L. japonicus, whereas it did not change in the nonmycorrhizal castor mutant sym4-2 (Fig. 2). The results indicate that starch stored in the root system is broken down when carbohydrates are taken up by the fungus upon colonization, as also observed in leaves of cucumber infected with the biotrophic leaf pathogen powdery mildew (Abood & Lösel, 2003). Starch breakdown in mycorrhizal roots may be supported by the up-regulation of amylase genes, as observed in mycorrhizal M. truncatula (Gomez et al., 2009). This might coincide with the structural changes of root plastids observed during arbuscule formation, that is starch disappearing from arbusculated cells (Bonfante, 1984) and (nonamyloplast) plastids accumulating around arbuscules (Fester et al., 2001) as well as in ready-to-be-colonized cortical cells (Genre et al., 2008).

An increase in sugar transport into the root, or a decrease in sugar consumption without import changes, should lead to starch accumulation. This was observed when roots were subjected to the double-sandwich set up that separates plant and fungus but allows the exchange of signal molecules. Because, under these conditions, the amount of starch doubled (Fig. 3a), we conclude that yet-unknown fungal signalling factors alter the root carbon storage metabolism. Root colonization by AMF enhances the import of photosynthetic carbon into roots (Kucey & Paul, 1982; Koch & Johnson, 1984; Wang et al., 1998). Therefore, we consider it more likely that starch accumulated because of enhanced sugar import into the root, which resulted in starch accumulation in the absence of AM colonization, instead of a change in root metabolism leading to a decrease in sugar consumption. This would imply that enhanced carbon import into mycorrhizal roots is not simply caused by fungal carbohydrate uptake and the resulting increase in steepness of the sucrose gradient between shoot and root, but is an actively regulated process resulting from perception of fungal signals. 14/13/11CO2 labeling studies to measure sucrose flux into roots and root respiration measurements in a set up that allows the perception of AMF signals by the root but inhibits colonization, are necessary to distinguish between enhanced sugar import and decreased sugar use and to test the hypothesis described above. Interestingly, starch granule accumulation in the root cortex and in developing root nodules has been repeatedly observed in response to Nod factors (Truchet et al., 1991; Ardourel et al., 1994), indicating that presymbiotic shifts in the carbon-sink strength of the root might also play a role in nodule symbioses that share a common signaling pathway with AM symbioses (Parniske, 2008). Infection with biotrophic leaf pathogens leads to enhanced starch accumulation in chloroplasts at sites of colonization and this has been attributed to cytokinin signalling (Sziraki et al., 1984; Scholes & Farrar, 1987; Tang et al., 1996; Walters & McRoberts, 2006; Doehlemann et al., 2008). Cytokinin signalling is also an essential component of the Nod factor signalling pathway downstream of common SYM signalling, leading to nodule formation in the root cortex (Murray et al., 2007; Tirichine et al., 2007). Furthermore, the cytokinin concentration is elevated in mycorrhizal roots (van Rhijn et al., 1997). It will be interesting to determine in the future whether cytokinin plays a role in myc factor signalling and whether it is important for the observed starch accumulation.

Mycorrhizal colonization leads to the induction of invertase and sucrose synthase genes (Hohnjec et al., 2003; Ravnskov et al., 2003; Schaarschmidt et al., 2006), an indication of the enhanced sink status of the mycorrhizal root. In a microarray experiment performed at presymbiotic stages of the interaction between L. japonicus and G. margarita, such induction was not found (M. Guether & P. Bonfante, unpublished). However, it is possible that starch accumulation before colonization is regulated post-transcriptionally (Tiessen et al., 2002).

We further demonstrated that AMF signalling molecules that lead to starch accumulation in roots are produced by germinating spores independently of plant signalling molecules such as strigolactones (Bouwmeester et al., 2007). This plant response reminds of a transient elevation in calcium exhibited by soybean cells after perception of the same fungal exudates (Navazio et al., 2007).

Starch accumulation induced by AMF signalling molecules requires the LjCASTOR gene

The common SYM pathway, consisting of seven proteins known to date, is required for AM formation (Parniske, 2008) and involves two highly homologous potassium channels called CASTOR and POLLUX (Charpentier et al., 2008). The protein LjCASTOR was shown to control two check-points of root colonization by the AMF: the first at the epidermis and the second at the cortex (Novero et al., 2002). The castor mutant, Ljsym4-2, which lacks intraradical colonization (Novero et al., 2002) does not accumulate starch when in direct contact with the fungus (Fig. 2a) or in the double-sandwich set up (Fig. 3b). Thus, we conclude that starch accumulation is under the control of CASTOR. Both CASTOR and POLLUX are required for calcium-spiking, which is central to the common SYM signalling pathway (Imaizumi-Anraku et al., 2005; Charpentier et al., 2008; Kosuta et al., 2008). Therefore, it is likely that calcium-spiking or the calcium elevation observed by Navazio et al. (2007) is a prerequisite for the starch-accumulation response.

As for all other known common SYM genes, CASTOR and POLLUX expression was described to be equivalent between mock-inoculated and fully colonized rice roots (Gutjahr et al., 2008). However, slight regulation has also been reported: for example, Navazio et al. (2007) observed an induction of DMI1, the orthologue of POLLUX, in soybean cell culture upon 24 h of treatment with G. margarita exudates, while Imaizumi-Anraku et al., (2005) found that CASTOR and POLLUX expression was weakly suppressed in L. japonicus roots at early stages (12–24 h) of rhizobium colonization and Nod factor treatment. Similarly, a slight repression of POLLUX was observed at 4 dpi when plants and germinating hyphae were establishing their early contacts (Fig. 4a), confirming the weak down-regulation of CASTOR observed in laser-dissected cortical cells derived from roots treated in the same way (J. Gomez-Ariza & P. Bonfante, unpublished). The transcriptional response of the two twin genes to AMF signals depended on the plant presence: treatment of the roots with spore exudates for 24 h did not lead to changes in expression of CASTOR and POLLUX; whereas at 4 dpi in the double sandwich, both genes were very slightly induced, confirming unpublished microarray data (M. Guether & P. Bonfante, unpublished). We speculate that in the double-sandwich set up, germinating spores might be stimulated by plant exudates to produce higher amounts of signal molecules. This might lead to an over-stimulation of the root causing weak induction of CASTOR and POLLUX. The slight repression observed during the early interaction prompts the question of whether the direct contact between AMF and plant epidermal cells may induce the release of other, currently unknown factors that are perceived exclusively by the contacted cells, lead to the development of responses that are related to the formation of the pre-penetration apparatus (PPA) (Genre et al., 2005; Siciliano et al., 2007) and possibly regulate the expression of CASTOR and POLLUX genes.

In conclusion, we have presented evidence that diffusible factors released from germinating G. margarita spores induce starch accumulation in L. japonicus roots. This response is mediated by CASTOR. Thus, fungal signals released during the asymbiotic and presymbiotic phases have an unexpected impact on carbohydrate metabolism in the host plant, which is mediated by the common symbiotic signalling pathway. It will be interesting to determine in the future whether Nod factor-induced starch accumulation is also under the control of common SYM signalling.


We are grateful to Dr Andrea Genre (University of Torino, Italy) for taking part in preliminary experiments and for comments, and Prof. Martin Parniske (University of Munich, Germany) for providing sym4-2 mutant seeds. C.G., M.G., and O.M. were funded by a Marie Curie PhD Fellowship of the European Union (Sixth Framework) in the frame of the project INTEGRAL (Intensifying Training in Europe for Genomic Research Activity on Legumes).