Transmission of rice blast from seeds to adult plants in a non-systemic way

Authors


E-mail: tharreau@cirad.fr

Abstract

Rice blast, caused by the fungal pathogen Magnaporthe oryzae, is a serious threat to rice production worldwide. In temperate regions, where rice is not cultivated for several months each year, little is known about the initial onset of the disease in the field. The main overwintering and primary inoculum sources reported are infested residues and seeds, but the subsequent steps of the disease cycle are largely unknown, even though a systemic infection has been proposed but not demonstrated. The present work follows rice blast progression in infected seeds from germination to seedling stage, with direct and detailed microscopic observations under both aerobic conditions and water seeding. With the use of GFP-marked M. oryzae strains, it was shown that spores are produced from contaminated seeds, infect emerging seedling tissues (coleoptile and primary root) and produce mycelium that colonizes the newly formed primary leaf and secondary roots. Using different rice cultivars exhibiting distinct levels of resistance/susceptibility to M. oryzae at the 2/4-leaf stage, it was observed that resistance or susceptibility of a considered genotype is already established at the seedling stage. The results also showed that when plants are inoculated either at ripening stage (mature panicles), heading stage (flowering/immature panicles) or even before heading (flag leaf fully developed), they produce infested seeds. These seeds produce contaminated seedlings that mostly die and serve as an inoculum source for healthy neighbouring plants, which gradually develop disease symptoms on leaves. The possible rice blast disease cycle was reconstructed on irrigated rice in temperate regions.

Introduction

Rice is grown in a wide range of geographic and climatic conditions. It constitutes the most important staple food for half of the world's population, especially in Asia (http://www.fao.org). Its production has increased over the last decades due to improvements in cultivation practices and the introduction of high-yielding cultivars (Sakamoto & Matsuoka, 2004). However, rice is exposed to diseases that contribute to limiting its production. Among them, rice blast, caused by the fungus Magnaporthe oryzae, is a major constraint for the productivity of this crop worldwide (Wilson & Talbot, 2009). In the absence of rice blast control strategies, M. oryzae can cause high annual yield losses (Oerke & Dehne, 2004; Skamnioti & Gurr, 2009). Rice blast disease has been found in all rice-growing countries (Kato, 2001; International Plant Protection Convention, 2011), covering a wide range of agro-ecosystems with very diverse environmental conditions from rain-fed uplands in the tropics to irrigated plains in temperate areas.

In field conditions, the fungus is able to infect all aerial parts of rice, resulting in leaf, node, neck and panicle blast. The fungal infectious process on its host is well studied. Leaf infection by M. oryzae is initiated by attachment of a spore that germinates and forms a melanized appressorium on the rice cuticle (Wilson & Talbot, 2009). This appressorium generates turgor pressure that ruptures the leaf cuticle, allowing M. oryzae to infect leaf epidermal cells with a bulbous invasive hypha (IH). Subsequently, filamentous IH are suggested to invade neighbouring cells through plasmodesmata (Kankanala et al., 2007). Fungal progression within rice cells results in the death of the infected tissues and the appearance of necrotrophic lesions on leaves by 3–4 days after inoculation (Wilson & Talbot, 2009).

In controlled conditions, and upon artificial inoculation with mycelium, M. oryzae is also able to colonize roots by forming hyphopodia (Sesma & Osbourn, 2004). However, little is known about the initial onset of the disease and how rice blast disease spreads in the field. In the tropics, overwintering is presumed not to be important because airborne conidia are present throughout the year (Ou, 1985). This assumption may not hold true in tropical regions of elevated altitude where rice is not grown in winter. In temperate regions where rice is absent for several months, overwintering and primary inoculum sources have not yet been determined. Overwintering of mycelium and conidia on straw has been proposed, but these inocula do not survive under moist conditions (Ou, 1985). Harmon & Latin (2005) reported that, in north-central Indiana (USA), M. oryzae can overwinter in infested residues of perennial ryegrass, but the proportion of the surviving population is insufficient to serve as effective primary inoculum for summertime epidemics. In addition, such survival was not demonstrated for M. oryzae populations pathogenic to rice which have previously been suggested to overwinter on an alternative host (Ou, 1985) and which are different from populations pathogenic to perennial ryegrass.

Most of the studies carried out so far support the occurrence of infested seeds as a source of primary inoculum. Long et al. (2001) showed that dead infested grains could serve as primary inoculum when placed in the field at the time of seedling emergence. Transmission of M. oryzae from rice seeds to seedlings has already been documented in different countries (Kuribayashi, 1928 and Suzuki, 1930 cited by Manandhar et al., 1998; Lamey, 1970; Mayee, 1974 cited by Long et al., 2001; Chung & Lee, 1983; Reddy & Bastawsi, 1989; Long et al., 2001), but no cytological observations of the infection process from the contaminated germinating seed to the seedling have been performed so far. Using naturally infested seeds as primary inoculum in field conditions in Nepal, Manandhar et al. (1998) showed that panicle symptoms and seed contamination are correlated. They observed that sporulation of M. oryzae on infested seeds was preferentially found at the embryonic end of germinating seeds, while occurring homogeneously all over the surface for non-germinating seeds. Moreover, a seed lot with 21% contamination led to <4% seedlings with blast lesions. Tests employing different ways of covering seeds with soil (light or complete) and under water seeding (no covering) pointed out that complete covering or seeding under water induce a lower infection frequency (Manandhar et al., 1998). Guerber & TeBeest (2006) reported similar experiments in the USA. They showed that 23% of the seed lots examined, including commercial lots, were contaminated and that the corresponding rate of diseased seedlings varied from 0·3 to 11% in greenhouse conditions. However, no disease was observed when infested seeds were germinated under water. When infested seeds were sown in the field, the fungus was recovered from different seedling parts, including roots, from plants with and without symptoms. These results clearly show that the fungus can survive on the grains used for seeding and that these contaminated grains can serve as primary inoculum. However, Guerber & TeBeest (2006) did not detect the fungus in seeds of highly resistant rice varieties. Bernaux & Berti (1981) reported that overwintering of mycelium and conidia could occur between the glumella and caryopsis because of the susceptibility of palea and lemma before anthesis.

The observation that inoculation of roots with mycelium could result in sporadic disease symptoms on the leaves has led some authors to hypothesize that M. oryzae could move systemically within a seedling (Sesma & Osbourn, 2004; Marcel et al., 2010). However, the same authors did not check for the presence of M. oryzae within the symptomless tissues. Similarly, systemic movement within a seedling from an infested seed has not been demonstrated experimentally (Lamey, 1970; Mayee, 1974; Chung & Lee, 1983; Kingsolver et al., 1984; Ou, 1985; Manandhar et al., 1998). To summarize previous reports, (i) seeds on panicles can be infested up to c. 25% and give rise to up to roughly 10% of diseased seedlings; (ii) infested seeds can serve as primary inoculum in the field; (iii) complete covering of infested seeds or infested seeds sown under water drastically reduces the frequency of seedling infection in the field; and (iv) no fungus was detected on seeds of resistant rice accessions.

Until now, no precise seedling infection time-course from infested seeds has been documented, and in particular it remains unclear whether the fungal progression is systemic. Here, the fungal progression has been followed from contaminated seeds (harvested and then stored for a few months) to seedlings by detailed cytological observations using two seeding conditions (aerobic/anaerobic). Seeds were previously contaminated using green fluorescent protein (GFP)-tagged M. oryzae strains at different plant development stages: (i) panicle (flag) leaf, (ii) immature panicles, or (iii) mature panicles. Fungal progression was monitored from seed to seed using different rice cultivars exhibiting diverse susceptibility levels.

Material and methods

Magnaporthe isolates, plant material and infection assays

GFP-tagged M. oryzae strains were used to investigate seedling infection. Magnaporthe oryzae isolate FR13-GFP was generated using Agrobacterium-mediated transformation, as previously described (Rho et al., 2001; Sesma & Osbourn, 2004). Magnaporthe oryzae strain CL3.6.7-GFP was engineered through protoplast isolation and transformation using the plasmid pfC2ORFGFP (CNRS-Bayer) and the protocol of Fudal et al. (2007). These strains were selected because of their routine use in this laboratory. Rice plants used in this study are temperate japonica cultivars: Sariceltik from Turkey, and Maratelli and Gladio from Italy. Plants and M. oryzae strains were grown as described by Faivre-Rampant et al. (2008). Inoculation was carried out by spraying 50 000 conidia mL−1 of FR13-GFP and CL3.6.7-GFP either on the panicle leaf (before heading) or on immature panicles (heading stage) or mature panicles (ripening stage). Rice plants were then placed in the dark with 100% relative humidity (RH) for 16 h and transferred back to the greenhouse until seed harvest. Seeds were then placed at 37°C for 2 days and stored at 20°C for at least 6 months prior to cytological observations.

To measure seed contamination frequency, 50 seeds per experiment were placed either in two Petri dishes on a Whatman no. 5 filter paper moistened with sterile distilled water (SDW) placed in a plastic box (‘aerobic conditions’), or in 15 mL tubes containing 2 mL of SDW (‘seeding under water’). Plates or tubes were then incubated at 25°C in 12 h light/dark cycles. Seed contamination percentage was calculated for each seed lot as the number of seeds infested by the fungus divided by the total number of seeds examined (50) and multiplied by 100. The final contamination percentage for each seed lot was determined as an average of at least three replicates ± standard error.

Sariceltik and Maratelli infected (with symptoms) and non-infected (without symptoms) seedlings grown in water were transferred in the glasshouse when 2 weeks old. Seedlings with and without symptoms were cultivated in the same tray. The soil was not flooded permanently, but kept moist by frequent watering. After 1 month, plants were placed 1 night per week in a dew chamber with 100% RH for 16 h and brought back in the glasshouse. This was repeated until seed harvest. Exposure to high RH was used to mimic field conditions where moist conditions are necessary for sporulation. In the glasshouse, RH was maintained below 70% to avoid sporulation and uncontrolled infections.

Cytological analysis

Seed and plantlet infection by M. oryzae was monitored over time using an Olympus BX60 microscope equipped with an UV lamp. The progression of the fungus was followed on 25 seeds per plate and a total of two plates, at 3, 6, 10, 13 and 15 days after sowing (d.a.s.), and the experiment repeated independently three times. For detailed inspection, rice tissues were observed on a Zeiss LSM 700 confocal microscope using × 10 N-Achroplan (numerical aperture 0·25) or × 20 Plan-Apochromat (numerical aperture 0·8) objective lenses. Fluorescence of GFP was detected using an excitation wavelength of 488 nm. Signals emitted were recorded from 555 to 639 nm. To check for the presence of M. oryzae inside the root, sections of 50 μm from embedded roots in 5% agarose were obtained using a vibratome (Microm, Thermo Scientific).

Results

Seed contamination rates by M. oryzae

To determine how rice seeds are contaminated by M. oryzae, adult rice plants were first inoculated at the developmental stage of mature panicles (ripening stage), with conidia of the green fluorescent protein (GFP)-expressing M. oryzae strain FR13. Seeds were then harvested, stored for 6–9 months, and seed contamination documented over a period of 13 days by fluorescence and confocal microscopy. The seeds of three temperate rice cultivars, showing different responses to the M. oryzae strain FR13 (Sariceltik, very susceptible; Maratelli, susceptible; Gladio, resistant; Fig. S1), were placed on moist filter paper (aerobic conditions) to examine the frequency of infested seeds. Among the three cultivars, seed contamination percentages varied from 4% for Gladio up to 21% for Maratelli (Fig. 1a), but all the seed samples germinated at high rates (75–98%) like the original non-inoculated (control) seed lots (Fig. 1a). After germination, the percentage of seedlings with blast lesions ranged from 0 for Gladio to 14% (±9) for Maratelli (Fig. 1a).

Figure 1.

(a) Germination rates and infection percentages of Sariceltik (SK), Maratelli and Gladio seeds by Magnaporthe oryzae (strains FR13 and CL3.6.7) collected from inoculated mature panicles (MP), immature panicles (IP) or panicle leaf (PL) of rice adult plants. The percentages of blast diseased seedlings are also shown. Three independent sowings were carried out for each seed lot and the mean with standard deviation was calculated. On the right side of the main figure are the germination percentages of the original non-inoculated/control seed lot. (b) Infection frequencies of rice seed teguments and rice hulled seeds. *Significant frequency difference between fluorescence from teguments and from hulled seeds (χ2 test, < 0·05).

Adult plants of the susceptible (Maratelli) and the resistant (Gladio) cultivars were then inoculated at an earlier developmental stage (immature panicles) with FR13-GFP. Harvesting of mature seeds and storage were carried out as for mature panicle-inoculated plants. Seed contamination percentages were similar for the two cultivars (Fig. 1a) and the seed lots germinated at 59% (±36) for Gladio and 91% (±8) for Maratelli, like the original non-inoculated seed samples (Fig. 1a). After germination, the percentage of diseased seedlings ranged from 0 for Gladio to 9% (±6) for Maratelli (Fig. 1a).

To determine seed contamination rate following inoculation of the panicle leaf, only the highly susceptible cultivar Sariceltik was tested. Seed infection rate reached 22·5% (±32) and 15% (±10) of seedlings exhibited blast lesions (Fig. 1a). A last inoculation on adult Sariceltik plants at the mature panicle stage was subsequently performed with a less aggressive strain, CL3.6.7-GFP. Thirty-three per cent (±17) of seeds were contaminated and 27% (±11) of seedlings showed blast lesions (Fig. 1a).

Seed contamination by rice blast was observed after inoculation at panicle leaf, immature and mature panicle stages. Moreover, in all tests M. oryzae was predominantly located on seed teguments, as shown in Figure 1b.

Time-course of seedling contamination by M. oryzae in aerobic conditions

When infested seeds were placed on moist filter paper (aerobic conditions), spores were visible at 3–4 d.a.s. (Fig. 2a). Contamination occurred mostly at the embryonic end of the rice seeds for the three cultivars and also at a lower rate (44%) on the awn in Sariceltik (Fig. 2b). For the two susceptible cultivars (Sariceltik and Maratelli), spores of FR13-GFP spread on the sterile lemmas (Fig. 2c) within 7 d.a.s. Magnaporthe oryzae progression resulted in an infected coleoptile producing mycelium and spores at 10 d.a.s. (Fig. 2d,e). From the coleoptile, mycelium and spores were able to infect the emerging leaf (Fig. 2f) within 13 d.a.s. Mycelium and spores were also observed on the primary and secondary roots at this time (Fig. 2g,h). Observations with confocal microscope confirmed the presence of fungal hyphae in the root cells (Fig. 2i). The difference in the infection process between Sariceltik and Maratelli was quantitative, not qualitative, because the fungus progression was faster and seedling susceptibility was higher for Sariceltik. After 2 weeks, plants of Maratelli and Sariceltik that were infected/with symptoms were transplanted to soil and grown in the glasshouse. All Sariceltik plants did not recover from infection and died within 2 weeks while 40% (±10) of Maratelli seedlings survived. It was confirmed that plant death was due to FR13-GFP infection in leaves as well as in roots (Fig. 2j,k).

Figure 2.

Sariceltik seed infection by GFP-transformed Magnaporthe oryzae rice FR13 strain in aerobic conditions. (a,b) Spores on the embryonic end of the seed (a), and on the awn (b), within 3 days after sowing (d.a.s.). (c) Spores infecting the lemma resulting in necrotic symptoms on the coleoptile (white arrow) by 7 d.a.s. (d) Infected upper part of the coleoptile covered by fluorescent mycelium. (e) End of the coleoptile, shown in (d), also infected by spores. (f) Infected primary leaf within 13 d.a.s. showing hyphal progression. (g,h) Infected primary (g) and partially infected secondary (h) roots with mycelium; a non-infected root is shown inset in (g); white arrows indicate fungus progression within the root and red arrow indicates external mycelium hyphae. (i) Longitudinal section of the infected root in (g) illustrating the presence of mycelium inside root cells; inset shows a transversal section of the same root. (j,k) Dead infected plantlet by 25 d.a.s. with fluorescent spores on the leaf (j) and fluorescent root (k, white arrow; the red arrow shows an uninfected root). Bar scale = 5 μm (magnification in i); 20 μm (i); 50 μm (b, magnification in e); 100 μm (a, c, d, e, f, g, h, j, k). All pictures were taken by epifluorescence microscopy except (i) which was by confocal microscopy. Col., coleoptile.

Regarding the resistant cultivar Gladio, spores and mycelium were present on the seed and infected the emerging coleoptile at 3–4 d.a.s. (Fig. 3a,b). Fluorescent appressoria were visible on the coleoptile (data not shown) within 5 d.a.s. resulting in an hypersensitive response and cell wall autofluorescence of the rice epidermis (Fig. 3c). Spores were also observed on the primary root and mycelium spread to the secondary root within 13 d.a.s. (Fig. 3d,e). However, no further infection occurred. After 2 weeks no fluorescence was observed, indicating no living fungal material was present; seedlings grew normally and did not show any macroscopic lesions (data not shown).

Figure 3.

Gladio seed infection by GFP-transformed Magnaporthe oryzae rice FR13 strain in aerobic conditions. (a) Spores on the seed within 3 days after sowing (d.a.s.). (b) Mycelium present on the seed infecting the emerging coleoptile at 4 d.a.s. (c) Hypersensitive response from epidermal cells (red arrow) of the coleoptile and epidermal cell wall autofluorescence (white arrow) due to the presence of appressoria. (d,e) Spores and mycelium on the primary (d) and secondary (e) roots at 13 d.a.s. Bar scale = 50 μm (a,b,c,e), 100 μm (d). All pictures were taken by epifluorescence microscopy.

Seed contamination by the blast fungus occurred irrespective of the level of susceptibility/resistance of the rice cultivar and led to seedling colonization for the susceptible cultivars but not for the resistant one.

Time-course of seedling contamination by M. oryzae under water seeding

The ability of the fungus to grow on infested seeds when placed under water, mimicking seed germination conditions in the field, was examined. For this purpose, seeds of Sariceltik (SK) were inoculated with CL3.6.7 at mature panicle stage (MP). The germination rate of the tested seed lot, 78% (±20), was similar to that obtained when seeds were placed in aerobic conditions (85% ± 12; Fig. 1a, SK MP CL3.6.7). Blast occurred on 30% (±33) of the infested seeds, which is similar to the contamination frequency of the same seeds when placed on moist filter paper (33% ± 17; Fig. 1a, SK MP CL3.6.7). Spores and mycelium were visible on the infested seeds within 2 d.a.s., preferentially at the embryonic end. At 4 d.a.s., a mycelium layer covered the emerging coleoptile (Fig. 4a), resulting in a total infection within 9 d.a.s. (Fig. 4b) and spore production (Fig. 4c). Fluorescent extended lesions then appeared on the growing primary leaf due to new infections by spores and/or colonization by fungal hyphae (Fig. 4d,e). At root emergence, the mycelium cover colonized the developing primary root to form a coat (Fig. 4f,g). Within 13 d.a.s., primary and secondary roots were infected by the fungus (Fig. 4h) and produced spores outside the root epidermis (Fig. 4i).

Figure 4.

Sariceltik seed infection and Maratelli adult plant infection by GFP-transformed Magnaporthe oryzae rice FR13 strain under water seeding. (a) Mycelium present on the seed forming a coat around the emerging coleoptile within 4 days after sowing (d.a.s.). (b) Infected coleoptile at 9 d.a.s.; inset is a non-infected coleoptile. (c) Magnification of (b) showing the presence of spores (white arrow) and mycelium at the coleoptile surface. (d) Infected primary leaf at 10 d.a.s. (e) Cellular events in an infected leaf at 13 d.a.s. (f,g) Mycelium present on the seed forming a coat around the emerging primary root at 6 d.a.s.; spores can also be observed on the root as shown in inset. (h,i) Infected root (h) producing spores (i) within 13 d.a.s.; inset in (i) shows longitudinal section of the infected root in (h), illustrating the presence of hyphae inside root cells. (j,k) Leaf necrotic lesion on 2-month-old plants before (j) and after (k) sporulation on moist filter paper overnight; inset in (j) is a macroscopic necrotic lesion due to rice blast. (l–o) Adjacent 5 mm long cuttings of an infected leaf showing a necrotic lesion (l, circled with broken line) illustrating the mycelium progression within the leaf from the necrotic lesion as indicated by the arrows in each picture; the end of the progression is shown in (o) as indicated by the arrowhead. (p) Necrotic lesion of Maratelli 3-month-old leaf under epifluorescence microscope. Bar scale = 5 μm (longitudinal root section in i), 50 μm (magnification in b and g, c, i) or 100 μm (a,b,d–h,j–p). All pictures were taken by epifluorescence microscopy except (e) and inset in (i), which were by confocal microscopy.

Contamination of seeds sown under water followed by colonization of seedlings by rice blast fungus occurred in these experiments, in contrast to what has been previously reported.

Infection of symptomless plants

To follow the blast infection process on rice adult plants, the 13-day-old Sariceltik plantlets with symptoms arising from germination of infested seeds under water were then transferred into the glasshouse with symptomless (non-fluorescent) Sariceltik and Maratelli seedlings. Plants with and without symptoms were grown together in the same tray and the soil was not flooded permanently. The plantlets with symptoms did not recover and died very quickly after transfer, while the symptomless Sariceltik and Maratelli seedlings grew normally (data not shown). After 2 months, typical blast necrotic lesions (Fig. 4j) appeared on leaves of half of the Sariceltik adult plants derived from symptomless seedlings. When these lesions were placed on moist filter paper overnight, fluorescent spores were clearly visible and developed at the top of the necrotic symptoms (Fig. 4k). After two additional weeks, 100% of the Sariceltik plants, symptomless when transferred in the glasshouse, exhibited necrotic lesions due to infection by the GFP strain used. To check for a putative systemic colonization of the plant by the fungus via the stem vascular system, the fungus was looked for in internodes and nodes. Five Sariceltik plants showing typical blast lesions were cut into 5 mm transverse sections and placed on wet filter paper overnight to favour potential sporulation from these cuttings. No fluorescence was visible on the cuttings (data not shown). When infected leaves were cut into small pieces on both sides of the macroscopically visible necrotic lesions, limited hyphal progression from the lesion could be observed within the leaf epidermis (Fig. 4l–o). Hyphae were observed in the vascular tissues of the leaf, as previously shown by Berruyer et al. (2006). After 3 months, 30% of the Maratelli plants which were symptomless when transferred in the glasshouse exhibited typical blast disease symptoms on the leaves fluorescing under the epifluorescence microscope (Fig. 4p). Two weeks later, 50% of the Maratelli plants showed fluorescent blast lesions. Panicles of the Sariceltik and Maratelli plants which were gradually infected in the glasshouse were harvested when mature. Full grains and empty seeds/secondary branches were placed on moist paper to check if this second generation of seeds could be infected. Results are given in Fig. S2a. For Sariceltik full grains, 1·8% (±1·7) exhibited fluorescent spores either on the tegument or on the awn (Fig. S2b). Sariceltik empty seeds and panicle secondary branches were also infected at a higher rate, i.e. 2·6% (±2·5; Fig. S2a,b). Maratelli full seeds were infected at a lower rate (0·7% ± 1·1) while Maratelli empty seeds or secondary branches were not infected (data not shown).

Discussion

In agreement with Lamey (1970), who previously showed that only 2% of hulled seeds were contaminated by M. oryzae, this study demonstrated with independent experiments that the fungus is preferentially located on the seed coat. This observation reinforces the hypothesis that spores or mycelium of M. oryzae do not easily enter seed tissues of rice. Rice blast occurred at the embryonic end of rice seeds, as previously reported by Chung & Lee (1983), Manandhar et al. (1998) and Long et al. (2001). In this way, the rice blast fungus gets easy access to the coleoptile emerging from the seed as well as to the primary root, and consequently colonizes the first leaf and the secondary roots. However, the present study noted that M. oryzae was also frequently present on the awn of Sariceltik, which is particularly long in this variety. From the awn, spores can attach to the growing coleoptile and first leaf, or mycelium can grow from the awn to reach and infect the seed.

Leaf blast severity on young rice plants (3/4-leaf stage) is greater than on older plants (7/8-leaf stage; Hwang et al., 1987; Roumen et al., 1992). Few studies have addressed the importance of rice resistance in the early stages of epidemics in younger plants. In previous reports, the fungus was not detected on seeds of highly resistant rice accessions (Guerber & TeBeest, 2006). Using fluorescent M. oryzae strains enabled the rice blast progression on infested seeds and seedlings to be followed easily and in detail. In the present work, different rice cultivars with diverse levels of resistance to the FR13 strain at an early vegetative stage were used, and it was observed that this level of resistance or susceptibility was already established at the developmental stage of the seedling. Indeed, despite being present on the seeds and the growing coleoptile, M. oryzae FR13 was not able to infect Gladio, the most resistant accession of the study. Gladio harbours at least one resistance gene, Pia, and FR13 the corresponding avirulence gene (AvrPia; D. Tharreau, unpublished data). Appressoria were formed on the coleoptile, but an hypersensitive response as well as autofluorescence at the plant cell wall were observed. In contrast, M. oryzae FR13 was able to infect and penetrate into the coleoptile, leaf and root of Sariceltik and Maratelli. The fungus carried on its progression within the vascular tissues of the leaf to colonize neighbouring tissues. The infection process led to the death of all Sariceltik seedlings, while some of the Maratelli seedlings survived. This difference of infection rate was consistent with the known susceptibility of these two cultivars at older stages (Vergne et al., 2010).

The results of these experiments also showed that rice blast is found in seedlings arising from seeds harvested from plants which were previously inoculated either at ripening stage (mature panicles), heading stage (flowering/immature panicles) or even before heading (flag leaf fully developed). Previous work on rice and M. oryzae from different parts of the world (USA, Egypt, Nepal) reported that seeds can be infested in the field on mature and flowering/immature panicles (Lamey, 1970; Bernaux & Berti, 1981; Reddy & Bastawsi, 1989; Manandhar et al., 1998). For some authors (cited in Guerber & TeBeest, 2006), M. oryzae could infect rice grains via three hypothetical routes of entry: (i) spikelet colonization from an infected boot leaf while the panicle is still enclosed within the boot, (ii) seed infection after panicle emergence through infected glumes, and (iii) fungus entry via the hilar region from infected vascular tissues of the mother plant. In this study, rice plant infection before anthesis and early before the emergence of the panicle tip from the flag leaf sheath led to seed contamination by M. oryzae at a significant rate. These results show that the infected flag leaf contaminated the panicle when emerging. This is consistent with neck blast symptoms frequently observed in the field. Sporulating lesions are often observed on the flag leaf in the field and the panicle stem often shows symptoms at the collar, probably after infection by spores drained from the flag leaf.

In the present work, some of the germinating infested seeds died because of blast colonization after a few days of growth. Thus, two non-exclusive scenarios of rice blast epidemics in the field can be proposed. The first scenario is the systemic colonization of the plant from the seed via the stem vascular system, without symptom expression (as proposed by Guerber & TeBeest (2006) and more recently by Marcel et al. (2010)). In this work, in favourable conditions, blast sporulated on previously symptomless seedlings, potentially supporting a systemic colonization of the stem. But, when diseased adult rice plants coming from symptomless seedlings were cut, no fungus was detected within the main culm. Moreover, blast is a hemibiotrophic fungus: an initial biotrophic infection phase, during which the pathogen spreads in living rice tissues, is followed by a necrotrophic phase during which rice cell death is induced (Ou, 1985). Such a necrotrophic phase makes a symptomless colonization unlikely and most of the contaminated seeds gave rise to infected seedlings that rapidly died.

The second scenario is the production of inoculum from dead seeds or dead seedlings and the contamination of healthy seedlings by transportation of mycelium or spores. On dead seeds or seedlings on the soil surface, spores and mycelium are produced and transported through the irrigation water to neighbouring rice plants and constitute a possible source of inoculum. This scenario requires that, in irrigated conditions, the fungus grows under water. According to previous studies, blast infection was never detected on seedlings raised under water seeding conditions (Lamey, 1970; Manandhar et al., 1998). However, in this study contaminated seed lots placed within water resulted in diseased seedlings. Although M. oryzae progression was slower than in aerobic conditions, it was present as mycelium or spores on roots and coleoptiles/leaves. Therefore, it seems that M. oryzae can survive, sporulate and infect the seedlings under water and that dead seeds or seedlings produce spores that contaminate healthy plants in irrigated conditions. In the field, rice plants are usually densely sown in a very small area and disease can spread rapidly, resulting in infection of all neighbouring rice plants (if very susceptible) or a proportion of them (if partially susceptible).

Long et al. (2001) showed that M. oryzae could sporulate for several weeks on inoculated seeds, previously autoclaved, when placed on the soil surface and could initiate epidemics in the field. In the same way, Guerber & TeBeest (2006) found rice blast on seedlings from planted or non-germinated seeds and seed coats on the soil surface. Long et al. (2001) also indicated that the development of symptoms was more correlated to inoculum time exposure than to plant age. This means that if the inoculum is already present at the beginning of rice growth, non-infected plants will have more chance to become infected later; this was observed in the present experiments, with symptom development between 2 to 3 months of cultivation. Therefore, a systemic invasion of the stem is not required for blast epidemics to develop from contaminated seeds and for infection of the panicle.

Marcel et al. (2010) proposed an alternative way of overwintering. They presumed that M. oryzae is conserved in the soil as resting structures (microsclerotia) or on rice residues. They speculated that the fungus could enter via the root tissue and spread thereafter in a symptomless mode; this would allow the fungus to continuously feed on healthy rice tissue until reaching the aerial parts for sporulating on leaves. This scenario seems unlikely because no resting structures have been described so far for M. oryzae. Mycelium survival in the soil is also unlikely because of competition with other microorganisms. In addition, no cytological proof of a systemic presence of M. oryzae in entire rice seedlings has been reported until now.

Taken together, the present results and previous results allow reconstruction of the rice blast disease cycle on irrigated rice in temperate regions. Grains are infected by sporulating lesions on the flag leaf probably during heading. The fungus survives on harvested seeds until the next growing season. When seeded under water, infected seeds are colonized by the fungus that, most of the time, kills the seedling. Spores on dead seedlings then serve as primary inoculum to infect healthy plantlets. Lesions on leaves produce spores that will infect new leaves, and so on until the infection of the flag leaf. In this cycle, sowing non-contaminated seeds appears to be the main way to control the disease. Seed treatments have been shown to reduce blast epidemics in temperate regions, supporting the idea that infested seeds are the source of primary inoculum (Ou, 1985).

Acknowledgements

The authors thank Ane Sesma (JIC, Norwich, UK) for kindly providing the FR13-GFP Magnaporthe strain, Marc-Henri Lebrun (CNRS-Bayer, Lyon, France) for giving the plasmid pfC2ORFGFP, and Cécile Ribot and Jean-Benoît Morel (INRA, Montpellier, France) for supplying the CL3.6.7-GFP isolate. Part of this work was supported by a grant from FranceAgrimer. The authors are grateful to Jean-Benoît Morel, Mathilde Sester and Enrico Gobbato for critical reading of the manuscript.

Ancillary