Infection-Associated Nuclear Degeneration in the Rice Blast Fungus Magnaporthe oryzae Requires Non-Selective Macro-Autophagy

Background The rice blast fungus Magnaporthe oryzae elaborates a specialized infection structure called an appressorium to breach the rice leaf surface and gain access to plant tissue. Appressorium development is controlled by cell cycle progression, and a single round of nuclear division occurs prior to appressorium formation. Mitosis is always followed by programmed cell death of the spore from which the appressorium develops. Nuclear degeneration in the spore is known to be essential for plant infection, but the precise mechanism by which it occurs is not known. Methodology/Principal Findings In yeast, nuclear breakdown requires a specific form of autophagy, known as piecemeal microautophagy of the nucleus (PMN), and we therefore investigated whether this process occurs in the rice blast fungus. Here, we report that M. oryzae possesses two conserved components of a putative PMN pathway, MoVac8 and MoTsc13, but that both are dispensable for nuclear breakdown during plant infection. MoVAC8 encodes a vacuolar membrane protein and MoTSC13 a peri-nuclear and peripheral ER protein. Conclusions/Significance We show that MoVAC8 is necessary for caffeine resistance, but dispensable for pathogenicity of M. oryzae, while MoTSC13 is involved in cell wall stress responses and is an important virulence determinant. By functional analysis of ΔMoatg1 and ΔMoatg4 mutants, we demonstrate that infection-associated nuclear degeneration in M. oryzae instead occurs by non-selective macroautophagy, which is necessary for rice blast disease.


Introduction
Rice blast disease is a widespread constraint to rice production and therefore poses a persistent threat to global food security [1]. Rice blast infections, caused by the ascomycete fungus Magnaporthe oryzae, are initiated by attachment of a three-celled spore, or conidium, to the rice leaf cuticle. The conidium sticks tightly to the leaf surface by means of an adhesive released from the spore tip during hydration [2]. Once attached, the conidium quickly germinates and forms a single polarized germ tube. Within 4 hours, the germ tube ceases apical extension and terminal hooking of the hypha starts, which represents the initiation of cellular differentiation to form a specialised dome-shaped cell, the appressorium, that is necessary for successful plant infection [3]. A narrow penetration hypha is formed at the base of the appressorium and enters the underlying epidermis, rupturing the cell wall and invaginating the plant plasma membrane [1].
Development of the M. oryzae appressorium requires external cues including a hard, hydrophobic surface and the absence of exogenous nutrients [4]. Multiple cellular signal transduction cascades, such as the cyclic AMP and Pmk1 MAPK signaling pathways, are initiated in response to these external triggers and bring about the terminal differentiation of the germ tube apex into an appressorium [3,5]. The appressorium of M. oryzae ruptures the plant cuticle by application of mechanical force through accumulation of very high concentrations of glycerol, which draws water into the appressorium to create enormous hydrostatic turgor [6]. Autophagic re-cycling of the contents of the conidium is necessary for formation of a functional appressorium [7]. Consistent with this, lipid and glycogen mobilization, under control of the MAPK and cAMP response pathways, have been shown to occur during appressorium development and may provide precursors for glycerol synthesis [8,9].
It is now clear that appressorium development by M. oryzae is genetically controlled by cell cycle progression and that entry of a nucleus in the germinating conidial cell into S-phase is the key step in initiating infection structure development [7,10]. During germination and appressorium development, one nucleus in the conidium undergoes mitosis in the germ tube, after which one daughter nucleus moves into the incipient appressorium and the other returns to the conidium and degenerates [7]. Completion of mitosis leads to collapse and death of the conidium and is necessary for appressorium maturation and plant infection [11]. Systematic deletion of genes encoding each component of the macroautophagy machinery renders M. oryzae non-pathogenic, providing evidence that autophagy is essential for plant infection [7,[11][12][13][14].
Despite evidence to show the importance of autophagy in programmed cell death of the conidium and subsequent appressorium maturation, the molecular machinery responsible for nuclear degeneration in the conidium of M. oryzae remains unknown. Moreover, the factors regulating nuclear degeneration and the destiny of degraded nuclei in the conidium have yet to be characterised. In S. cerevisiae, it has been shown that piecemeal microautophagy of the nucleus (PMN) is a separate process that is necessary for recycling of non-essential portions of the nucleus and is induced by starvation or exposure to rapamycin, an inhibitor of the TOR signalling pathway [15,16,17,18,19,20,21]. PMN occurs constitutively at nucleus-vacuole (NV) junctions, formed through a specific binding interaction of Vac8p on the vacuole membrane and Nvj1p in the outer nuclear envelope [15,16]. During PMN, small teardrop-shaped portions of the nucleus are extruded along NV junctions into invaginations of the vacuolar membrane, which results in formation of tethered blebs that finally release vesicles containing non-essential nuclear material into the vacuole lumen for degradation by resident hydrolases [16,21]. Lipid metabolic proteins Osh1p and Tsc13p have been shown to be recruited and enriched at NV junctions by physical association with Nvj1p and may function in nonvesicular lipid trafficking and biogenesis of a distinctive lipid environment at NV junctions [19,22]. In addition, a spectrum of core autophagy machinery genes is required for the terminal vacuolar enclosure of the invaginated blebs and efficient production of intravacuolar PMN vesicles [20].
In this study, we set out to determine whether there is an identifiable PMN pathway in M. oryzae and to ask whether this process drives nuclear degeneration in the conidium during rice blast infection. Here, we report that MoVAC8 encodes a vacuolar membrane protein, which plays a role in the caffeine response, and that MoTSC13 is necessary for maintaining conidial morphology and for penetration peg development during plant infection. Importantly, we demonstrate that nuclear degeneration in the conidium occurs even in the absence of MoVAC8 and MoTSC13 and that there is no evidence for a discernable PMN pathway in M. oryzae. Instead M. oryzae degrades nuclei using a macroautophagic mechanism, which is a necessary pre-requisite for plant infection.

Results and Discussion
Nuclear degeneration occurs during appressorium development in M. oryzae To investigate nuclear behaviour during appressorium development, we performed live-cell imaging and quantitative analysis of nuclear number in a M. oryzae strain expressing a histone H1enhanced red fluorescent (H1:RFP) protein fusion [10]. Mitosis occurred in the germ tube emerging from the apical cell of the conidium between 4-6 hour post inoculation (hpi) and the daughter nucleus moved into the incipient appressorium, while the mother nucleus returned to the conidium, as shown in Figure 1. After the completion of mitosis and formation of the appressorium, nuclear degeneration occurred in the conidium, during which the nucleus in the basal cell of the conidium collapsed first, followed by the two nuclei occupying the middle cell and apical cell, respectively, as shown in Figure 1. Nuclear degeneration occurred without overt nuclear fragmentation and red fluorescence associated with nuclear material could be observed both in the cytoplasm and in vacuoles within conidia. After 24 h, nuclear degeneration always resulted in a single nucleus, which was present in the mature appressorium ( Figure 1).
Two components of the piecemeal autophagy of the nucleus pathway are present in the M. oryzae genome To identify the molecular machinery involved in nuclear degeneration in M. oryzae, we set out to determine whether the selective PMN pathway, described in S. cerevisiae, participates in degradation and recycling of nuclei during appressorium development by the rice blast fungus. In S. cerevisiae, VAC8, TSC13 and NVJ1 are the three important components of the PMN pathway. We interrogated the M. oryzae genome database using Blastp and identified putative homologues of Vac8p and Tsc13p. VAC8 is a vacuolar membrane-associated protein, which plays important roles in several vacuolar processes in S. cerevisiae, including piecemeal microautophagy of the nucleus (PMN) [15,23,24,25,26]. VAC8 was first identified in a survey of the S. cerevisiae genome for armadillo (ARM) repeat domain-containing proteinsconserved modules involved in mediating protein-protein interactions [23,27,28]. The gene was also identified independently by complementation of a class I vacuole segregation mutant, vac8, which contains multi-lobed vacuoles and arrests early in vacuole inheritance with defects in the cytoplasm to vacuole (Cvt) targeting pathway [29,30]. The myristoylation of glycine and palmitoylation of three cysteine residues inside the N-terminal Src homologue 4 (SH4) domain are critical for Vac8p association with the vacuole membrane [29] and, indeed, palmitoylation at the three cysteines determines the enrichment and function of Vac8p at specific vacuolar membrane sub-domains [25,26,31,32]. Vac8p interacts with different proteins through its ARM repeat domains at discrete vacuole membrane sub-domains specific to each of its distinct functions. Interacting partners include Vac17p in vacuole inheritance, Atg13 in the Cvt pathway, Nvj1p in NV junction formation and Tco89p in caffeine resistance [25,26,27,31]. Homologues of S. cerevisiae VAC8 have been reported to function in glucose-induced pexophagy in Pichia pastoris and in vacuolar inheritance and normal hyphal branching in Candida albicans, respectively [33,34,35,36]. In M. oryzae, MoVac8p shows 85.2% identity to S. cerevisiae Vac8p (Figure S1 A). The predicted MoVac8p coding region has 11 putative ARM repeats and to test this prediction, we designed primers starting at the start codon predicted in the genome database and performed 39 RACE. Unexpectedly, sequencing the 39 RACE amplicon and a subsequent 59RACE product showed that the correct start codon was 303 bp downstream of the predicted start codon within the first predicted intron (Genbank JN977613). The RADAR programme was used to align the ARM repeats of MoVac8p (http://www.ebi.ac.uk/Tools/Radar/index.html) and demonstrated that MoVac8p contains 9 ARM repeats, with repeat 8 and repeat 9 interrupted by 53 amino acids, which contrasts significantly with the 11 continuous ARM repeats in S. cerevisiae Vac8p ( Figure 2). MoVac8p shares similar N-terminal acylation sites to those found in S. cerevisiae Vac8p (Figure 2), consistent with its predicted function. A second major component of the PMN pathway in yeast, the TSC13 gene, encodes enoyl reductase, an enzyme that catalyzes the last step of long-chain fatty acid (C 16 and C 18 ) elongation to produce very-long-chain fatty acids (VLCFAs) [37]. Tsc13p is an integral membrane protein located in the peripheral and perinuclear endoplasmic reticulum (ER), enriched at NV junctions, and is essential for cell viability [15,22,37]. The activity of Tsc13p in the VLCFA elongation cycle has been proposed to contribute to the biogenesis of PMN blebs [22]. There have, however, been no reports of the functions of VAC8 or TSC13 orthologues in any filamentous fungus to date. The M. oryzae MoTsc13p showed 60.1% identity to S. cerevisiae Tsc13p, with six predicted transmembrane domains (Figure 2 C), consistent with the topology of S. cerevisiae Tsc13p [38]. MoTsc13p has conserved amino acids (Figure S1 B), essential for activity of Tsc13p [38] (Genbank JN977614). Importantly, we were unable to find a homologue of S. cerevisiae NVJ1, using either nucleotide or amino acid sequences of NVJ1, based on BLASTP or TBLASTN analysis, in the M. oryzae genome database, or by immunoprecipitation which we used to identify proteins interacting with MoVac8-GFP (data not shown).

MoVac8-GFP localises to the vacuole membrane and Tsc13-GFP to the perinuclear and peripheral ER membrane
To investigate whether MoVac8p and MoTsc13p showed similar sub-cellular localisation patterns to their yeast counterparts (consistent with a PMN function) we generated MoVAC8:GFP and MoTSC13:GFP gene fusion constructs and expressed them under their native promoters in the wild type M. oryzae strain Guy11. MoVac8-GFP showed a membrane-associated distribution pattern in conidia, appressoria and invasive hyphae, as shown in Figure [39].
In hyphae of transformants expressing MoTSC13:GFP, the fusion protein was also membrane-associated, as shown in Figure 3E. To stain nuclei, 2,4,-Diamidino-phenyl-indole (DAPI) was used in hyphae of these transformants and showed that the MoTsc13p is detected predominantly at locations consistent with the peri-nuclear ER membrane and peripheral ER ( Figure 3E). This membrane-associated distribution pattern of MoTsc13p was also found in the conidium, appressorium, penetration peg and invasive hyphae during plant infection ( Figure 3 C and D). During appressorium development, MoTsc13p was detected in the germ tube and differentiating appressorium, indicating that MoTsc13panchored ER moves into the appressorium ( Figure 3C). Taken together, these data revealed that both MoVac8p and MoTsc13p showed sub-cellular distribution patterns consistent with a role in a variety of vacuole and ER functions.

Conservation of MoVAC8 and MoTSC13 function
To determine whether MoVAC8 and MoTSC13 are functional equivalents of S. cerevisiae VAC8 and TSC13, respectively, complementation experiments were performed. Heterologous expression of a MoTSC13 cDNA in a S. cerevisiae tsc13-1 Delo double mutant, was sufficient to restore its ability to grow at 37uC, as shown in Figure 4 A, suggesting that MoTSC13 is the functional homologue of yeast TSC13 enoyl-CoA reductase [40,41]. When we expressed yeast enhanced GFP (yEGFP)-tagged MoTSC13 in tsc13-1 Delo mutants, they also complemented the mutant phenotype and displayed the same perinuclear and peripheral ER membrane-anchoring distribution in yeast cells ( Figure 4E). We conclude that MoTSC13 probably serves an evolutionarily conserved function in catalyzing the fourth reaction of fatty acid elongation to produce VLCFAs in both fungi [37].
In S. cerevisiae, Dvac8 mutants show various vacuole-associated phenotypes, including caffeine hypersensitivity, multi-lobed vacuoles, loss of protein transport from the cytoplasm to vacuoles, an inability of budding daughter cells to inherit vacuoles from the mother cell and, importantly defects in PMN [15,23,24,25,26]. When MoVAC8 cDNA was expressed in a S. cerevisiae Dvac8 mutant BY4741 under control of the GAL1 promoter, growth of yeast was partially restored in the presence of 0.05% or 0.1% caffeine (Figure 4 B), suggesting that VAC8 has conserved functions between S. cerevisiae and M. oryzae in regulating the caffeine response. We used pulse-chase labelling with FM4-64 to track vacuolar morphology and inheritance during budding of the S. cerevisiae strain BY4741 expressing MoVAC8. Vacuoles remained multi-lobed and identical to those observed in the Dvac8 mutant. Moreover, yeast daughter cells failed to inherit vacuoles from mother cells ( Figure 4C and D). To test whether MoVac8p was targeted to the vacuolar membrane of S. cerevisiae, a MoVAC8 cDNA:yEGFP was constructed and introduced into the S. cerevisiae Dvac8 mutant BY4741. Interestingly, MoVac8p was mostly distributed in the cytoplasm and failed to accumulate at the vacuolar membrane ( Figure 4F), indicating that the N-terminal vacuole-membrane anchoring peptide found in M. oryzae is not fully functional in S. cerevisiae. Partial complementation of the yeast Dvac8 mutant by MoVAC8 may reflect the different structural organization of Vac8p between S. cerevisiae and M. oryzae. Taken together, we conclude that MoTSC13 is a direct functional homologue of S. cerevisiae TSC13 while MoVAC8 appears to fulfil a role in the caffeine response but may have diverged in both structure and function in M. oryzae.
MoVAC8 and MoTSC13 are not required for conidial nuclear degeneration during appressorium development To determine the function of both putative PMN proteins in M. oryzae, we generated DMovac8 and DMotsc13 mutants in Guy11 using a split marker method and confirmed targeted gene deletion by Southern blot hybridization ( Figure S2). In order to determine whether MoVAC8 and MoTSC13 are involved in nuclear degeneration, H1:RFP was introduced into both DMovac8 and DMotsc13 mutants to allow live cell imaging of nuclear behaviour. We monitored nuclear numbers during appressorium development and, strikingly, nuclei showed the same behaviour between DMovac8, DMotsc13 and Guy11, as shown in Figure 5A and B. These observations suggest that nuclear degeneration occurs independently of a PMN pathway in M. oryzae because nuclear degeneration was unaffected in either mutant. However, compared to Guy11 and DMovac8, a much higher percentage (,30%) of conidia of the DMotsc13 mutant contained only one or two nuclei, as a consequence of a conidial morphology phenotype that was associated with loss of MoTSC13 ( Figure 5C).  consequently, the nucleus and vacuole were often apposed to one another, but no distinct, regulated physical interaction of vacuoles and nuclei was observed. The typical teardrop-shaped blebs, which in S. cerevisiae originate from NV junctions and release PMN vesicles into the vacuole [15,16], were also absent from conidia undergoing autophagic cell death, as shown in Figure 6A. More importantly, Vac8p was degraded in the conidium at a time when nuclei were still present ( Figure 6 A; 4 h and 8 h timepoints), indicating that vacuole degeneration may proceed before the onset of nuclear degeneration. Because PMN is induced to high levels in S. cerevisiae by starvation [15,16], we also carried out microscopy of the MoVAC8:GFP; H1:RFP strain grown under nitrogen starvation  conditions to determine whether NV junctions were apparent after starvation of M. oryzae. No obvious NV junctions or teardropshaped blebs were detected in M. oryzae following starvation stress, as shown in Figure 6B.
In comparison to the accumulation of S. cerevisiae Tsc13p at NV junctions from peripheral and nuclear ER pools during starvation stress [22,37], we found that MoTsc13-GFP in the conidium of M. oryzae was equally distributed at perinuclear and peripheral ER membranes during appressorium development ( Figure 3C). When the MoTSC13:GFP gene fusion was introduced into Guy11 carrying H1:RFP, MoTsc13p also showed a nuclear and peripheral ER membrane-associated distribution pattern (Figure 6 C), and no enrichment of the MoTsc13p at the nuclear membrane was observed in hyphae grown under starvation conditions (Figure 6 D). Taken together, we conclude that there is no formation of NV junctions, the typical structures of PMN, in M. oryzae either during appressorium development or following nitrogen starvation.
The N-terminal SH4 domain of MoVac8p is required for association of MoVac8p with the vacuolar membrane The N-terminus of Vac8p in S. cerevisae contains a SH4 domain, which serves as a membrane anchoring signal peptide [31]. SH4 domains are normally composed of 18 amino acids and characterised by a myristoylation motif (MGxxxS/Tx) and a palmitoylation site (a cysteine residue) or several basic amino acids [32]. The SH4 domain within Vac8p of S. cerevisae for instance, possesses three palmitoylation sites, which play roles in the localisation of Vac8p in the vacuolar membrane [26,31]. Analysis of the N-terminal sequences of MoVac8p revealed the presence of a myristoylation motif and three potential palmitoylation sites, as shown in Figure 2 A.
To investigate whether MoVac8p contains a functional SH4 domain, we used the first 21 amino acids of MoVac8p to generate a putative SH4 domain:GFP fusion protein. Localisation of the SH4 domain:GFP fusion protein was examined in conidia, appressoria, invasive hyphae and vegetative hyphae by epifluorescence microscopy (Figure 7). The distribution of the SH4 domain:GFP fusion protein in each cell type was coincident with FM4-64 stained membranes and vacuoles, and also overlapped with the CFW stained cell wall and septa, suggesting that SH4:GFP is membrane-associated.
To address whether the myristoylation and palmitoylation sites of the putative SH4 domain are involved in localisation of MoVac8p at the vacuolar membrane, we performed site-directed mutagenesis to generate constructs expressing variants of MoVac8p-GFP, in which glycine and cysteine residues within the SH4 domain were replaced by alanine residues ( Figure S3A). Single point mutations of MoVAC8:GFP, including G2A, C4A, C8A and C9A, did not abolish association of MoVac8p-GFP with the vacuolar membrane (See Figure S3B and Figure S4). However, these single point mutations did result in mislocalisation of MoVac8p-GFP into the septal pore region in vegetative hyphae ( Figure S3B and C), but not in conidia ( Figure S4, at least 50 conidia were examined for each variants). When two of the palmitoylation sites were mutated, including C4A/C8A, C4A/C9A and C8A/C9A, mislocalisation of MoVac8p-GFP in the mycelia septa pore area was further increased ( Figure S3B and C), and MoVac8p-GFP in the C4A/C8A, C4A/C9A variants showed strong cytosolic localisation and loss of association with the vacuolar membrane. The C8A/C9A substitution resulted in an increase in the mislocalisation into the septa pore ( Figure S3 B and C). Similar results were obtained in mutants expressing the double point mutations, C4A/C8A, C4A/C9A and C8A/C9A when conidia were examined ( Figure S4). Moreover, in vegetative hyphae and conidia of the MoVac8p-GFP strain expressing a triple point mutation C4A/C8A/C9A, the association of fusion proteins with the vacuolar membrane was completely disrupted resulting in completely cytosolic localisation and enrichment at the septal pore ( Figure S3B and C, Figure S4). We conclude that both myristoylation and palmitoylation are involved in localisation of MoVac8p.
In S. cerevisae palmitoylation of Vac8 is required for caffeine resistance [26]. To examine the relationship between acylation of the SH4 domain and MoVac8p function, we therefore measured sensitivity of strains expressing mutant alleles of MoVAC8:GFP to caffeine. DMovac8 mutants showed hypersensitivity to 0.1% caffeine, while expression of MoVAC:GFP restored normal growth ( Figure S5). In DMovac8 mutants expressing MoVAC:GFP variants C4A, C8A, C9A, C4A/C8A, C4A/C9A and C8A/C9A, growth on CM containing 0.1% caffeine was restored, but variant G2A only partially restored growth and the triple mutant C4A/C8A/C9A failed to restore full growth ( Figure S5). These results indicate that myristoylation, in particular, and complete palmitoylation of MoVac8p plays a role in caffeine resistance in M. oryzae.
MoVAC8 is necessary for the caffeine response, while MoTSC13 is required for full virulence and cell wall integrity We investigated the functions of MoVAC8 and MoTSC13 by analysis of the phenotypes of each deletion mutant. In view of the role of Vac8 in vacuole inheritance and movement, we investigated movement of vacuoles and endosomes during appressorium development in both Guy11 and DMovac8 mutants by staining with FM4-64. Both DMovac8 mutants and Guy11 showed a similar pattern of vacuole and endosome movement, in which vacuoles in the conidium moved into the germ tube during germination and into the appressorium, during cellular differentiation ( Figure 8A). Moreover, the fusion of vacuoles was not impaired in DMovac8 mutants ( Figure 8A). These results indicate that MoVac8p does not serve roles in vacuole inheritance or vacuole fusion during conidium germination or appressorium development. DMovac8 mutants did, however, show enhanced caffeine sensitivity and slightly increased sensitivity to calcofluor white and high concentrations of Congo red, consistent with a role in cell wall integrity ( Figure S7). Plant infection assays also suggested that MoVAC8 is dispensable for pathogenicity of M. oryzae ( Figure 9B), because DMovac8 mutants caused similar numbers of disease lesions to the isogenic wild type strain Guy11 and appressoria formed normally.
In contrast to the essential function of TSC13 in S. cerevisiae, MoTSC13 is not essential for viability in M. oryzae, but loss of MoTSC13 did reduce vegetative growth and conidiation of M. oryzae and increased sensitivity to osmotic stress and Calcofluor White ( Figure S8A, B and C). Importantly, DMotsc13 mutants were only able to produce very small disease lesions on rice leaves as shown in Figure 9A. DMotsc13 mutants formed appressoria normally ( Figure S8D), implying that neither MoVAC8 nor MoTSC13 serve essential functions in appressorium development.
To determine which stage of infection was impaired in DMotsc13 mutants, we measured appressorium turgor and the frequency of penetration peg formation ( Figure 9B, Figure S8 E). Turgor was unaltered in DMotsc13 mutants ( Figure S8 E). However, penetration peg formation was severly impaired with only 20% of appressoria able to elaborate a penetration peg after 24 h ( Figure 9B). By 36 hpi, invasive hyphae of Guy11 had moved into the second or third rice epidermal cell adjacent to the invasion site, but most DMotsc13 mutant appressoria failed to penetrate, and invasive hyphae were limited to the initial cell at the invasion site ( Figure 9B). Re-introduction of the MoTSC13:GFP fusion construct into a DMotsc13 mutant restored normal vegetative growth, penetration peg formation and pathogenicity on rice leaves ( Figure 9A and B; Figure S8 A). We conclude that MoTsc13 is involved in penetration hypha development during plant infection.

Macroautophagy is required for nuclear degeneration during appressorium development
Given the absence of a discernable PMN pathway, we decided to investigate alternative means by which nuclei might be degraded in M. oryzae. We first expressed the H1:RFP gene fusion in a DMoatg1 mutant to allow in vivo observation of nuclei during appressorium development in a macroautophagy-deficient mutant. Live cell imaging showed that nuclei in the conidium were misshapen and failed to degenerate even after 24 hpi, as shown in Figure 10A. We also examined whether MoATG4 was required for nuclear degeneration in M. oryzae. To achieve this, we performed targeted gene deletion of MoATG4 in a M. oryzae strain expressing both MoVAC8:GFP and H1:RFP (Figure S2 C). We found that MoATG4 was required for conidial collapse and nuclear degeneration during appressorium development ( Figure 10B). We went on to examine nuclear degradation in targeted deletion mutants affecting both macro-autophagy and selected autophagy. We found that mutants in genes associated macro-autophay all showed defects in nuclear degeneration as observed in DMoatg1 and DMoatg4 mutants [11]. By contrast mutants in genes associated exclusively with selective autophagy (ATG11, ATG24, ATG26, ATG27, ATG28, ATG29) did not show any defect in nuclear degeneration (data not shown).
We also investigated the localisation of MoTSC13:GFP and MoVAC8:GFP gene fusion constructs in a DMoatg1 mutant in order to see the effect of arresting autophagy on protein localisation during infection related development We observed that MoTsc13p-GFP accumulated in the conidium until 24 hpi ( Figure  S6), in contrast to the gradual disappearance of MoTsc13p in Guy11 after 4-6 h ( Figure 3C). In Guy11 expressing Mo-VAC8:GFP, vacuole degeneration started after completion of mitosis, and vacuoles were absent from the conidium after 24 hpi ( Figure 3A). While in an DMoatg1 mutant expressing MoVAC8:GFP, vacuoles failed to degenerate even after 24 hpi (Figure S6), suggesting a crucial role for macroautophagy in mediating vacuole degeneration or trafficking from the conidium during appressorium development in M. oryzae. Consistent with this idea, vacuoles also accumulated in the conidium of DMoatg4 mutants, as shown in Figure 10B. When considered together these data suggest that macroautophagy is important for nuclear degeneration, ER degeneration and vacuole degeneration within conidia during plant infection by M. oryzae.

Conclusions
In this study we set out to determine the mechanism by which nuclei are broken down in conidia of the rice blast fungus prior to appressorium formation. Appressorium-mediated plant infection by the rice blast fungus is tightly linked to cell cycle control and conidial cell death and degeneration of nuclei within the spore is an essential pre-requisite to successful plant infection [7,10,11]. In yeast, it is apparent that nuclei are degraded by a selective autophagic process, PMN, in which nuclei bind to vacuoles via nucleus-vacuole (NV) junctions. These NV junctions invaginate and release PMN vesicles containing nuclear material into the lumen of vacuoles for hydrolysis [16,17]. We have demonstrated that M. oryzae possesses two strong candidate PMN genes, MoVAC8 and MoTSC13, but does not possess a NVJ1 homologue and, importantly, does not appear to form NV junctions associated with PMN-mediated nuclear breakdown. Furthermore, we have shown that mutants lacking either MoVAC8 and MoTSC13 still undergo nuclear breakdown and appressorium differentiation, indicating that PMN does not mediate nuclear degeneration in M. oryzae.
Based on yeast complementation experiments, we observed that MoVAC8 fulfils only a sub-set of the functions of its yeast counterpart and was unable to localize correctly when expressed in a yeast Dvac8 mutant. This is likely to be a consequence of its distinct structure with only 9 ARM repeats present in the protein, compared to 11 in Vac8p. It is clear, however, that MoVac8 is a vacuolar membrane protein, which is both myristoylated and palmitoylated [42] in M. oryzae and is involved in the response to caffeine, because DMovac8 mutants show hypersensitivity to caffeine (1,3,7-trimethyl xanthine). This function is also conserved when MoVAC8 was expressed in a yeast Dvac8 mutant. Caffeine sensitivity in S.cerevisiae appears to be associated with the Pkc1/cell integrity pathway because caffeine treatment induces rapid phosphorylation of the Mpk1 MAP kinase and leads to large scale changes in gene expression associated with cell wall stress [43]. However, the similarity in transcriptional response to rapamycin treatment, coupled with the hypersensitivity of Tor1 kinase mutants to caffeine, also point to an effect on the Ras/ cAMP response pathway, and the control of cellular viability, which is coupled with the regulation of autophagy. The hypersensitivity of DMovac8 mutants to caffeine may therefore be associated with an impairment in vacuole transport function, which is consistent with the requirement for myristoylation and palmitoylation for vacuolar membrane localization. Interestingly, the wider reported roles for Vac8p in vacuolar inheritance were  [39]. We can therefore conclude that MoVac8 is a vacuolar protein that is unlikely to serve a role in PMN in the rice blast fungus, but instead plays a role in vacuolar function which may be vital for contending with abiotic stresses such as exposure to caffeine.
In contrast to MoVAC8, MoTSC13 appears to have a highly conserved function as an enoyl reductase that catalyzes the fourth reaction of fatty acid elongation to produce very long chain fatty acids. This function appears to be completely conserved with the role of TSC13p in yeast, but strikingly, MoTSC13 is not essential for cellular viability in M. oryzae and DMotsc13 mutants instead grow well in culture. Furthermore, we found no evidence for a role for MoTsc13p in PMN and there was no distinct localization of the protein at specific NV junctions. Instead, we found that MoTsc13-GFP localized to the perinuclear and peripheral ER. Importantly, we did observe that DMotsc13 mutants are significantly impaired in their ability to cause rice blast disease and that this results as a consequence of a reduced ability to colonize rice epidermal cells following appressorium-mediated penetration of the cuticle. We can conclude that very long chain fatty acid biosynthesis is therefore likely to be important in invasive hyphae development, perhaps pointing to the membrane components of invasive hyphae possessing a distinct lipidic characteristic compared to those of vegetative hyphae-a feature worthy of future investigation.
The final conclusion that can be made from this study is that nuclear degeneration during appressorium formation, which is known to be essential for plant infection [7,10,11], occurs via nonselective macroautophagy. In contrast to yeast, there is no evidence for a separate selective PMN process in M. oryzae. We found that macro-autophagy-associated genes such as MoATG1 or MoATG4 were necessary for nuclear degeneration and their absence rendered the fungus non-pathogenic [11], whereas mutations in genes affecting selective forms of autophagy did not show any difference from the wild type Guy11. Macroautophagy has very recently been reported to mediate nuclear degeneration in Aspergillus oryzae [44], but in that case involved formation of large ring-like autophagosomal structures (1-2 mm) that encircled and mediated degradation of whole nuclei in A. oryzae basal cells. In this study we were only able to detect punctate autophagosomes in both conidia and appressoria of M. oryzae (consistent with [11,12,13]), rather than the much larger, ring-like autophagosome structures reported in A. oryzae [44], suggesting that nuclear breakdown may be performed by distinct macroautophagydependent processes in filamentous fungi. Furthermore, nuclei did not appear to be degraded in their entirety, but rather there was dissolution of nuclear material, which could be observed both sytoplasmically and ithin vacuoles during autophagy. When considered together, we can conclude that conidial cell death and nuclear degeneration, which occur as part of the essential programme for appressorium-mediated plant infection by M. oryzae, both require non-selective autophagy, which re-cycles the contents of these cells, including nuclei, ER and other organelles into the specialized infection structure, prior to plant cuticle rupture and tissue colonization.

Fungal strains, growth conditions, and DNA analysis
The fertile rice pathogenic M. oryzae strain, Guy11, was used in all studies [45]. Culture, maintenance, and storage of M. oryzae isolates, media composition, nucleic acid extraction, and fungal transformation were all as previously described [46]. Yeast strains were manipulated using standard methods. All primers used in this study are described in Table S1. S. cerevisiae strain BY4741 vac8D::KANMX4 (MATa his3D1 leu2D0 met15D0 ura3D0 vac8D::-KANMX4) used for expression of MoVAC8 cDNA was obtained from EUROSCARF. S. cerevisiae strain TDY2058 (MATa elo3::TRP1 tsc13-1 ade2-101 ura3-52 trp1D leu2D) used for expression of MoTSC13 cDNA was kindly provided by Dr. Teresa M. Dunn (Department of Biochemistry, Uniformed Services University of the Health Sciences, Bethesda, Maryland). Gel electrophoresis, restriction enzyme digestion, gel blots, PCR and sequencing were performed using standard procedures [47].
Targeted deletion of MoVAC8, MoVAC8:GFP and MoVAC8SH4:GFP fusion plasmid construction, sitedirected mutagenesis of MoVAC8:GFP and DMovac8 complementation The split-marker recombination method was used for efficient targeted deletion of M. oryzae genes [11,48]. The hph gene, which confers resistance to HygromycinB (HYG) was used as the split marker. The two split hph templates were amplified by primers M13F with HYsplit and M13R with YGsplit, as previously described [11]. A 1 kb sequence flanking either side of the MoVAC8 coding sequence was amplified, with left flanking (LF) sequences amplified by primers vac850.1 and vac8m13f, right flanking (RF) sequences amplified with primers vac830.1 and vac8m13r. The LF sequences were fused with split HY, using primers vac850.1 and HYsplit to form LF-HY, while the RF sequences fused with the split YG fragment using primers YGsplit and vac830.1, to form YG-RF. The resulting amplicons LF-HY and YG-RF were gel-purified and co-transformed into protoplasts of Guy11. The DMovac8 mutants were confirmed by DNA gel blot analysis and two independent mutants selected for further phenotypic analysis. The MoVAC8:GFP construct was made by fusion PCR and standard restriction enzyme-mediated cloning. The MoVAC8 gene (1.7 kbp promoter and 2.1 kbp CDS) was amplified with primers Vac8fusionFor and Vac8GFPRev. Primer Vac8GFPRev contained overhanging sequences at its 59 end, which were complementary to the sGFP sequence. The 1.5 kb sGFP coding region, together with TrpC terminator, was amplified with primers GFPTrpCFor and GFPTrpCRev. The MoVAC8:GFP fusion cassette was then generated with primers Vac8fusionFor and GFPTrpCRev. An XhoI restriction enzyme recognition site was introduced into the 59 end of both primers, Vac8fusionFor and GFPTrpCRev, to facilitate cloning of the MoVAC8:GFP fusion cassette into pCB1532 vector for fungal transformation [49]. The first 21 amino acid section of the N-terminus of MoVac8p, covering the predicted SH4 domain, was fused into the Nterminus of GFP and the yeast recombination method employed to generate a MoVAC8SH4:GFP fusion [50]. The 1.7 kb MoVAC8 promoter sequence was amplified with primer pair, Vac8P-For and Vac8P-Rev, while SH4 domain-coding sequences were amplified with primer pair VAC8SH4-For and VAC8SH4-Rev from cDNA prepared from conidial total RNA. The amplicons were gel-purified and co-transformed into the relevant yeast strain together with plasmid pAGL1:GFP, which was linearized with HindIII and contains the selectable marker gene SUR conferring resistance to chlorimuron ethyl.
Site-directed mutageneses was performed on plasmid pCB1532-MoVAC8:GFP to generate alleles containing replacement of the predicted myristoylation (glycine) or palmitoylation (cysteine) modification sites by alanine residues, including G2A, C4A, C8A, C9A, C4A/C8A, C4A/C9A, C8A/C9A and C4A/C8A/C9A. In brief, a 714 bp region was amplified initially in two fragments, and nucleotide substitutions introduced into the primers, located at the overlapping region of the two fragments that were joined together by fusion PCR using primer pair Vac8mut-For and Vac8mut-Rev. The 714 bp fragment carrying the respective nucleotide substitutions was then digested with FseI and PmlI to release a 320 bp fragment that was used to replace the region spanning FseI and PmlI in plasmid pCB1532-MoVAC8:GFP. DNA sequencing was utilised to confirm successful introduction of each nucleotide substitution. Finally, the plasmid variants of pCB1532-MoVAC8:GFP were transformed into DMovac8 mutants, and at least two independent transformants selected for phenotypic analysis. For complementation of DMovac8 mutants, the pCB1532 vector carrying Mo-VAC8:GFP was transformed into a DMovac8 mutant and at least two independent transformants tested for complementation.
Targeted deletion of MoTSC13, MoTSC13:GFP fusion plasmid construction and DMotsc13 complementation Targeted gene deletion of MoTSC13 was performed with the split-marker recombination method, as described above. The 1.0 kbp LF sequences were amplified with primers tsc1350.1 and tsc13m13f, while the 1.0 kbp RF sequences were amplified with primers tsc1330.1 and tsc13m13r. Generation of DMotsc13 mutants was confirmed by DNA gel blot analysis and two independent mutants selected for further analysis [52]. The MoTSC13:GFP fusion was made using the yeast recombination method, as described above [53]. Briefly, the bialaphos resistance selectable marker gene BAR was amplified using primers BarF and BarR, and a 3.0 kb fragment of the MoTSC13 gene amplified with primers Tsc13GFPFor and Tsc13GFPRev, GFP-TrpC terminator cassette amplified with primers GFPTrpCFor and GFPTrpR [52]. The amplicons were gel-purified and co-transformed into the relevant yeast strain together with vector pNEB-Nat which had been linearized with HindIII and SacI.
Expression of MoVAC8 and MoTSC13 in S. cerevisiae Full-length double-stranded cDNAs of MoVAC8 and MoTSC13 were amplified from 1st strand cDNA using primer pair Vac8yeast50.1 and Vac8yeast30.1 and primer pair Tsc13yeast50.1 and Tsc13yeast30.1, respectively. MoVAC8 cDNA was cloned into KpnI and XbaI sites of yeast expression vector pYES2 (Invitrogen) and introduced into S. cerevisiae strain BY4741 vac8D::KANMX4, while the MoTSC13 cDNA was cloned between HindIII and XbaI sites of pYES2 and introduced into S. cerevisiae strain TDY2058. For expression of yeast-enhanced GFP (yEGFP) tagged MoVAC8 and MoTSC13 in S. cerevisiae, the cDNA of MoVAC8 was amplified by primer pair Vac8yEGFPFor and Vac8yEGFPRev, MoTSC13 amplified by Tsc13yEGFPFor and Tsc13yEGFPRev, and yEGFP amplified by primer pair yEGFPFor and yEGFPRev from plasmid pKT127 (obtained from EUROSCARF). The MoVAC8:yEGFP and MoTSC13:yEGFP fusion cassettes were generated by primer pair Vac8yEGFPFor and yEGFPRev, and primer pair Tsc13yEGFPFor and yEGFPRev respectively, both cloned between EcoRI and SphI sites in pYES2 [53]. The MoVAC8:yEGFP construct was introduced into S. cerevisiae strain BY4741 vac8D::KANMX4, while MoTS-C13:yEGFP was expressed into TDY2058. The cDNA sequences of both MoVAC8 and MoTSC13 in pYES2 were confirmed by DNA sequencing. All yeast transformants were confirmed by PCR and at least two independent yeast transformants chosen for analysis.Sensitivity to caffeine was assessed by spotting a dilution series of yeast cells (10 7 210 4 cells ml 21 ) on synthetic drop-out (SD) medium containing 0.05% or 0.1% caffeine in the presence of galactose. Vacuole morphology and inheritance in S. cerevisiae were observed by staining with FM4-64 (Molecular Probes, Invitrogen) according to [30]. For assessing vacuolar inheritance, pulse-chase labelling with FM4-64 [30] was performed by washing FM4-64 stained cells twice with fresh medium and incubating for an additional 4 h at 30uC. For testing temperature-sensitive lethality of yeast strains TDY2058, a dilution series of yeast cells (10 6 210 4 cells ml 21 ) expressing MoTSC13 were spotted onto synthetic drop-out (SD) medium in the presence of galactose at 37uC, 30uC or 25uC.
Generation of M. oryzae macroautophagy deficient strains carrying either H1:RFP (tdTomato), MoVAC8:GFP or MoTSC13:GFP -gene fusions A H1:RFP fusion construct was introduced into DMovac8 and DMotsc13 mutants for live cell imaging of nuclei. MoVAC8:GFP and MoTSC13:GFP gene fusion constructs were introduced into Guy11 carrying H1:RFP. H1:RFP, MoVAC8-GFP. MoTSC13:GFP gene fusions were also introduced into a DMoatg1 mutant to investigate behaviour of these fusion proteins in macroautophagy-deficient mutants. Transformants were selected by DNA gel blot, and at least two independent transformants investigated for all experiments. For targeted deletion of MoATG4 in Guy11 expressing both H1:RFP and MoVAC8:GFP gene fusions, the split-marker BAR was used. Briefly, the two split BAR templates were amplified by primers M13F with BAsplit and M13R with ARsplit, and MoATG4 LF amplified by primers Atg450.1 and Atg4m13f, MoATG4 RF amplified by primers Atg430.1 and Atg4m13r, as previously described [11]. The LF-BA was obtained with primers Atg450.1 and BAsplit, while AR-RF obtained with primers ARsplit and Atg430.1.

FM4-64 staining of conidia or mycelia in M. oryzae
The lipophilic styryl dye, FM4-64 (N-(3-triethylammoniumpropyl-)-4-(6(4-(diethylamino)phenyl) hexatrienyl) pyridinium dibromide) was used to stain vacuoles and endosomes of conidia or mycelia in M. oryzae (Molecular Probes, Invitrogen). Conidia grown in CM agar plate culture were collected with 4 ml of sterile distilled water, filtered through miracloth (Calbiochem). Approximately 200 ml of conidial suspension, at 1610 6 ml 21 was centrifuged at 6, 000 g for 5 min to precipitate conidia. After washing with 1 ml of sterile distilled water and centrifugation at 6, 000 g for 5 min, the conidial pellet was resuspended in 50 ml of liquid CM with 7.5 mM FM4-64. The suspension was incubated at 26uC for 20 min, and then conidia were recovered by centrifugation. The supernatant was discarded to remove excess FM4-64 and pellet washed twice with 1 ml of sterile distilled water. Conidia were finally resuspended in sterile distilled water at a concentration of 5610 4 ml 21 . Appressorium development was observed on coverslips at indicated time points with epifluorescence microscopy.

Plant pathogenicity and infection structure development assay
Cuticle penetration was assessed by recording the frequency of penetration peg formation from appressoria on onion epidermis. A 50 ml drop of conidial suspension at a concentration of 5610 4 conidia ml 21 was placed on the surface of onion epidermis and incubated in a humid environment at 24uC for 24 h or 48 h. The frequency of cuticle penetration was determined microscopically by counting formation of penetration pegs from at least 100 appressoria in triplicate replications of the experiment. Turgor generation in mature appressoria was measured by a cytorrhysis assay in a series of glycerol solutions of varying molarity, as previously described [6,51]. Rice infections were performed using cultivar CO-39, a dwarf rice cultivar which is very susceptible to M. oryzae [46]. A conidial suspension (5610 4 mL 21 ) was produced by flooding 10-day-old M. oryzae culture plates with 0.2% (v/v) gelatine solution and the suspension spray-inoculated onto 14-day-old rice plants. Plants were placed in plastic bags for 24 h to maintain high humidity and then transferred to controlled environment chambers at 24uC and 90% relative humidity with illumination and 14 h light periods. Plants were incubated until disease symptoms were apparent 96-144 h later.
Conidial germination and development of appressoria were both monitored over time on hydrophobic borosilicate glass cover slips (Fisher Scientific) using a method adapted from [2,46]. Conidial suspensions at 5610 4 conidia mL 21 were inoculated onto cover slips, incubated at 24uC, and all images of conidial germination and appressorium development were recorded using a Zeiss Axioskop 2 microscope (Zeiss).

Light and epifluorescence microscopy
For epifluorescence microscopy of GFP or RFP expressing transformants, conidia were inoculated onto coverslips, incubated at 24uC and collected at indicated time points for observation using an IX81 motorized inverted microscope (Olympus) equipped with an UPlanSApo 1006/1.40 Oil objective (Olympus). Excitation of fluorescently-labeled proteins was carried out using a VS-LMS4 Laser-Merge-System with solid state lasers. The laser intensity was controlled by a VS-AOTF100 System and coupled into the light path using a VS-20 Laser-Lens-System (Visitron System). Images were captured using a Charged-Coupled Device camera (Photometric Cool-SNAP HQ2, Roper Scientific). All parts of the system were under the control of the software package MetaMorph (Molecular Devices) and offline images were analyzed with MetaMorph software and Adobe Photoshop CS2 (Adobe Systems Incorporated).