The Role of Snx41-Based Pexophagy in Magnaporthe Development

Pexophagy, the degradation of peroxisomes via selective autophagy, depends on Atg20/Snx42 function in Saccharomyces cerevisiae. Besides its role in selective autophagy, Atg20/Snx42 is also involved in an autophagy-independent endosomal retrieval trafficking, in cooperation with two other sorting nexins, Snx41 and Snx4. Recently, we reported that the sorting nexin MoSnx41, which showed high sequence similarity to yeast Snx41 and Snx42/Atg20 proteins, regulates the gamma-glutamyl cycle and GSH production and is essential for conidiation and pathogenicity in Magnaporthe oryzae. Pexophagy was also found to be defective in Mosnx41Δ mutant. These findings indicate that MoSnx41 likely serves combined functions of Snx42/Atg20 and Snx41 in M. oryzae.. In this study, we performed complementation analyses and demonstrate that MoSnx41 alone serves the dual function of protein sorting (ScSnx41) and pexophagy (ScSnx42/Atg20). To study the potential biological function of pexophagy in fungal pathogenic life cycle, we created deletion mutants of potential pexophagy-specific genes, and characterized them in terms of pexophagy, conidiation and pathogenesis. We identified Pex14 as an essential protein for pexophagy in M. oryzae. Overall, our results show that pexophagy per se is not essential for asexual development or virulence in M. oryzae.

Peroxisomes are unit membrane-bound organelles, containing peroxidase and catalase activities and contribute to several lipid metabolic pathways and redox homeostasis in eukaryotic cells [6,7]. So far 32 PEX genes have been identified and the encoded peroxins are indispensable for peroxisome biogenesis [8,9]. Peroxisome biogenesis can be induced by the presence of fatty acids, and degradation of the organelle in vacuoles (pexophagy) is in response to glucose [5,10,11].
The role of peroxisomal metabolism and pexophagy in fungal pathogenicity is not well understood. Peroxisomal biogenesis protein Pex6 is required for pathogenicity in Colletotrichum orbiculare, the fungal pathogen causing cucumber anthracnose [12,13], and in the rice-blast fungus M. oryzae [14]. On the other hand, pexophagy was found to be required for host invasion by C. orbiculare [15]. The M. oryzae genome contains a subset of ATG and PEX genes, whose orthologs in yeasts are dispensable for non-selective autophagy but serve as critical pexophagy regulators. Thus, such specific ATG or PEX candidate genes are well-suited for studying the biological function of pexophagy in fungal pathogenesis. Atg26 was reported to be essential for pexophagy in Podospora, Pichia [16,17] and in C. orbiculare [18]. It remains unclear whether Atg26 is essential for pexophagy in M. oryzae, although an M. oryzae atg26Δ mutant has been shown to be pathogenic in a recent genome wide study [19]. Atg20, an autophagy protein, was shown to be essential for the Cvt (cytosol-to-vacuole targeting) pathway and pexophagy in yeast [20]. However, previous screening with yeast orthologs showed that the ATG genes specifically involved in the Cvt pathway and/or pexophagy were poorly conserved in filamentous fungi, including M. oryzae [21]. Based on this study, it is believed that there is no Atg20/Snx42 like protein in M. oryzae. In Hansenula polymorpha, a peroxisomal membrane-anchoring Peroxin 14 (Pex14) was implicated in both peroxisome biogenesis and pexophagy [22,23]. Recently, we reported an M. oryzae sorting nexin, Snx41, to be orthologous in sequence to both Snx41 and Snx42/Atg20 from S. cerevisiae. MoSnx41 is dispensable for non-selective autophagy, but essential for pexophagy, which is functionally similar to S. cerevisiae Snx42/Atg20. More importantly, it remains to be investigated whether pexophagy plays a role in M. oryzae conidiation and/or pathogenicity.
In this study, we assessed the role of ATG26 and PEX14, which are single-copy genes in M. oryzae, in pexophagy. Our results indicate that Atg26 is not required for pexophagy in M. oryzae. Whereas, the N-terminal cytosolic domain (amino acid 1-60) of Pex14 was found to be essential for pexophagy in M. oryzae. Gene complementation analysis showed that M. oryzae Snx41 is functionally equivalent to ScSnx42/ScAtg20 in pexophagy, while being similar to ScSnx41 in retrieval trafficking function. However, the biological defects in snx41Δ were more relevant to Snx41-dependent retrieval trafficking, rather than Snx42/Atg20-associated pexophagy [24]. Lastly, we demonstrate that pexophagy does not occur during M. oryzae conidiation. Consistent with this, pexophagy is dispensable for M. oryzae asexual development.

Pexophagy during asexual development in M. oryzae
In order to observe whether pexophagy occurs naturally during M. oryzae conidiogenesis and pathogenesis, the interaction between peroxisomes (labeled by GFP-SRL [14]) and autophagic vesicles (RFP-Atg8 [25]) was examined during conidia and appressoria development. Time-lapse observation of peroxisomes was performed in vegetative hyphae and conidiating cultures grown on Prune Agar (PA) medium (24 h post photo-induction), and during in planta development ( Figure 1A and B). In the vegetative phase, peroxisomes (labeled using GFP-SRL; Peroxisome Targeting Sequence 1) frequently associated with the filamentous or spherical vacuoles that contained RFP-Atg8 ( Figure 1A; left panel; n=100; p < 0.05). However, when conidiation was induced in response to light, peroxisomes and autophagic vesicles were distinctly separate in conidiation-related structures, including the aerial hyphae ( Figure 1A; middle panel; n=100; p < 0.05), the conidiophore and the conidia ( Figure 1A; right panel; n=100; p < 0.05), indicating a likely lack of pexophagy or peroxisome turnover during M. oryzae conidiation. We infer that pexophagy does not occur in the aerial hyphae or conidiophores in the rice-blast fungus.
To monitor pexophagy during appressorium development, conidia were harvested and inoculated on an artificial inductive surface or on leaf explants from barley. Several relevant time points were chosen during appressorium differentiation. Pexophagy was observed in germinating conidium (6-9 hpi), mature appressorium (16-30 hpi) and invasive hyphae (96 hpi) ( Figure 1B). Peroxisomes were delivered to the vacuoles in non-germinating conidial cells during appressorium initiation or during invasive growth ( Figure 1B, arrowheads), indicating that pexophagy likely occurs at these stages during pathogenesis in M. oryzae.
We conclude that pexophagy is undetectable during M. oryzae conidiation, as shown in our time-lapse microscopic observation, but vacuolar degradation of peroxisomes is frequently evident during appressorial development and function. Here, we summarize and schematically illustrate the observed steps in pexophagy during conidial and appressorial development in M. oryzae (Figure 2A).

Generation and functional characterization of pexophagy-deficient mutants in M. oryzae
To study the biological function of pexophagy in M. oryzae, we first created deletion or truncated mutants of pexophagycandidate genes detailed below. Previously, we characterized a Mosnx41Δ mutant, which was blocked in pexophagy and displayed loss of conidiation and pathogenicity [24]. Wild type and snx41Δ strains, both carrying GFP-SRL to visualize peroxisomes, were analysed to assess pexophagy-competent and pexophagy-deficient conditions, as illustrated in Figure 2B. Diffused vacuolar GFP-SRL signal (76.6 ± 3.5 %) was evident in the WT during conidial development ( Figure 2B, arrowheads). In contrast, the snx41Δ conidia lacked vacuolar GFP-SRL signal (0 ± 0 %, p < 0.01 vs WT) at the comparable stage of development ( Figure 2B, arrows) but instead, GFP-SRL punctae accumulated outside and/or at the periphery of the vacuoles.
M. oryzae Snx41 shows extensive similarity to both Snx41 and Snx42/Atg20 from S. cerevisiae. Both ScSnx41 and ScSnx42/Atg20 contain Phox (PX) domain for Phosphatidylinositol 3-Phosphate (PI3P) binding and membrane anchoring ( Figure 2C). BAR (Bin-Amphiphysin-Rvs) domain in conjunction with PX in ScSnx42 may be involved in membrane bending and curvature formation [26,27], while it is absent in ScSnx41 ( Figure 2C). M. oryzae Snx41 possesses PX domain and a region with weak homology to ScSnx42/ Atg20 BAR domain ( Figure 2C), indicating a functional conservation with ScSnx42/Atg20. However, BLASTP analysis with M. oryzae Snx41 showed a higher sequence similarity overall with ScSnx41 [24]. ScSnx42/Atg20 was shown to be essential for pexophagy in S. cerevisiae, while ScSnx41 was not required for turnover or degradation of peroxisomes [20]. Requirement of ScSnx42/Atg20 for pexophagy depends on the BAR domain, which may facilitate membrane curvature for pexophagy-PAS formation [26] or for fusion between pexophagosomes and vacuole [27]. Both ScSnx42/Atg20 and ScSnx41 are involved in non-pexophagy functions too, e.g. endosomal sorting and retrieval trafficking [28], which again rely on BAR domain-dependent membrane bending and/or tubulation. Since ScSnx41 does not possess a BAR domain, it may depend on its partner (ScSnx42/Atg20) for such membrane bending function. M. oryzae Snx41 is essential for pexophagy, and is clearly involved in other vesicular trafficking events, which are unrelated to pexophagy but essential for M. oryzae conidiation and pathogenesis [24]. Since M. oryzae Snx41 contains the BAR domain, it may carry out pexophagy function in a similar fashion as its yeast homolog ScSnx42/ Atg20. Also, for vesicular sorting and trafficking, as it contains a BAR motif of its own, it may not require the formation of an endosomal retromer complex involving Snx4/Snx41/Snx42 [28]. We therefore hypothesize that MoSnx41 combines the functions (retrieval trafficking and pexophagy, respectively) of   (3) and in the germ tube (4). In the newly formed appressorium (5), pexophagy was absent as peroxisomes were not delivered into the vacuoles. In mature appressoria, pexophagy occurs again and most of the peroxisomes were delivered to the lumen of the vacuoles. Right panel depicts pexophagy-competent (WT) or pexophagy-deficient situation in such cell types. Red spherical or filamentous compartments, vacuoles (stained by DND99 Lysotracker-Red); green vesicles, peroxisome labeled with GFP-SRL. (B) Pexophagy-competence and pexophagy-deficiency as illustrated by WT and snx41Δ mutant, respectively. Conidia were inoculated on plastic cover slip for 6-8 h, and co-stained with Lysotracker Red DND99 to label the vacuolar compartments 5 min before confocal microscopy. Arrowheads denote vacuolar GFP-SRL signal, an indication of delivery of peroxisomes to the vacuoles for pexophagy. Arrows denote vacuoles lacking GFP-SRL signal, due to blocked pexophagy. Scale bar equals 10 micron. these two related sorting nexins from budding yeast. To assess functional relevance between MoSnx41 and ScSnx41/42, we expressed ScSNX41 or ScSNX42 in M. oryzae snx41Δ mutant expressing the peroxisomal GFP-SRL fusion protein [24].
Next, we created deletion mutants in ATG26 or PEX14 orthologs in M. oryzae. An atg26Δ strain [19] was created by replacing MGG_03459 with the hygromycin-resistance cassette (HPH1) in the GFP-SRL strain of M. oryzae. M. oryzae Peroxin14 (Pex14) is similar to the H. polymorpha ortholog, which contains a central region composed of a short hydrophobic loop, an SH3 domain and a coiled-coil domain essential for its function as the peroxisomal translocon [23]. The N-and C-terminus of M. oryzae Pex14 (hereafter called Pex14 unless specified otherwise), both predicted to be cytosolic, are rich in Serine and Threonine residues that are potential phosphorylation sites and predicted to act as a "switch" for pexophagy [23] ( Figure 2D). The pex14Δ mutant was generated in M. oryzae (MGG_01028.6; in GFP-SRL background). Two truncated versions of PEX14, PEX14 61-361 and PEX14   (Figure 2D), were individually expressed in the pex14Δ mutant under the native PEX14 promoter.
As expected, GFP-SRL mis-localized to the cytoplasm in the pex14Δmutant ( Figure 3A), likely due to loss of the predicted peroxisomal translocon function of Pex14. Expression ofPEX14 61-361 or PEX14 1-258 as the sole copy in the deletion mutant could restore the punctate localization of GFP-SRL ( Figure 3A), likely due to restoration of peroxisome biogenesis and/or peroxisomal import function of Pex14 via these two truncated Pex14 variants.

Pexophagy is not essential for M. oryzae conidiation or pathogenesis
Next, we investigated the physiological relevance of pexophagy during asexual spore formation (conidiation) and infection, using the pexophagy-deficient mutants and aforementioned Mosnx41Δ complemented with ScSNX41 or ScSNX42/ATG20. Consistent with what was reported earlier [19], the atg26Δ mutant was capable of proper conidiation and pathogenesis ( Figure 4B and C). Despite pexophagy defects, the PEX14 61-361 mutant was fully capable of conidiation and pathogenicity ( Figure 4B and C). This indicated that pexophagy may not be essential for M. oryzae conidiogenesis or pathogenesis. Furthermore, pexophagy and pathogenic defects could be separated in the snx41Δ complemented strains: conidiation and pathogenesis were fully restored in snx41Δ expressing ScSNX41 ( Figure 4B and C), although pexophagy was still blocked (Figures 3B and 4A). In contrast, expression of ScSNX42 restored pexophagy (Figures 3B and 4A) but could not restore conidiation or pathogenesis in the snx41Δ mutant ( Figure 4B and C). We conclude that pexophagy per se is not essential for M. oryzae asexual development or pathogenicity, which likely depend on other vesicular trafficking or sorting pathway(s) that requires Snx41 function.
To conclude, conidiation and pathogenicity were normal in pexophagy-deficient mutants PEX14 61-361 and snx41Δ (ScSNX41), suggesting that pexophagy may not be essential for conidia formation and/or function in M. oryzae.

Discussion
Atg26 was identified as a pathogenic factor in C. orbiculare [15], but does not appear to be important for pathogenesis in    , PEX14 61-361 , snx41Δ and the two complemented strains were grown in peroxisome biogenesis (Pe) or pexophagy induction (Px) conditions. The total lysates from aforementioned strains were subjected to immunoblotting with anti-thiolase (Thi) antibody. The immunoblot was reprobed with anti-Porin antisera as loading control. Percentage denotes reduction of Thiolase in Px condition compared to the Pe condition, as an indicator of pexophagy efficiency. Percentage reduction was calculated using ImageJ [38]  M. oryzae [19]. In our study, we showed that pexophagy was not dependent on Atg26 in M. oryzae, which may explain the dispensability of Atg26 for M. oryzae pathogenesis [19]. Furthermore, our findings clearly contrast to those reported for C. orbiculare [15,18], where pexophagy is essential for host invasion. We found that pexophagy per se is dispensable for M. oryzae pathogenesis.
In M. oryzae and Colletotrichum species, internal turgor pressure is produced in the appressoria to facilitate breaching of the host epidermal cell during invasion [31]. During infectionrelated morphogenesis, the conidia of both M. oryzae and C. orbiculare metabolize the fatty acids stored in lipid bodies through β-oxidation within peroxisomes [13,14,31]. This explains why peroxisome biogenesis and its functional integrity are important for virulence, in these two fungal pathogens. It remains unclear whether the turnover of peroxisomes by pexophagy is equally important. Atg26-mediated pexophagy was shown to be essential for appressorial turgor generation, and the possible reason could be the regulation of pexophagy efficiency by removing redundant peroxisomes in the mature appressoria [15]. Such process appears less important for M. oryzae. Autophagic degradation of peroxisomes occurs in nongerminating conidial cells during M. oryzae appressorial development and maturation ( Figure 1B). However, even in pexophagy-deficient mutants, the clearance/turnover of peroxisomes could be achieved at the late stage of infection (when penetration and in planta growth occur), probably as a consequence of autophagic cell death (ACD) in the conidial cells. ACD is essential for appressorial maturation and function [32], and likely depends on non-selective autophagy, but not on pexophagy and/or other selective subtypes of autophagy in M. oryzae [33].
We also found that Pex14 alone carries out distinct functions in both peroxisome biogenesis and turnover. As a peroxisomal membrane protein, its central region anchors Pex14 on peroxisomes and meanwhile essential for peroxisome biogenesis, by facilitating peroxisomal protein import [34,35] ( Figure 2D). The N-terminal cytosolic region of Pex14 is specifically essential for pexophagy, as shown in our study ( Figure 3B). It was reported recently that N-terminus of Pex14 is also required for peroxisome motility in mammalian cells [36]. Our preliminary observations suggest that it is not the case in M. oryzae (data not shown).
The sorting nexin Snx41 was recently shown to be essential for M. oryzae conidiation and pathogenesis, mainly due to its function in retrieval trafficking of a critical enzyme involved in amino acid metabolism and/or anti-oxidant production [24]. Our previous study revealed that Snx41 was also required for pexophagy [24], but its relevance to M. oryzae pathogenic development was not addressed. In this study, by genetic complementation we demonstrated that retrieval trafficking and pexophagy could be independently carried out by ScSnx41 and ScSnx42, two yeast orthologs related in sequence to M. oryzae Snx41. We further proved that the pexophagy function of Snx41, which was restored by ScSnx42 in snx41Δ mutant, was dispensable for conidiation or pathogenesis. Overall, our study dissects the specific roles of Snx41 in M. oryzae and represents a systematic analysis of pexophagy and its biological relevance in the rice blast pathogen.

Fungal strains, growth conditions and infection assay
M. oryzae wild-type strain B157 (Field isolate, mat1-1) was from the Directorate of Rice Research (Hyderabad, India). M. oryzae strains were propagated on Prune-agar (PA) medium or Complete medium (CM) as described [14,37]. To induce peroxisome biogenesis or pexophagy, vegetative hyphae of respective strains grown in CM liquid medium for 3 d, were shifted to basal medium containing olive oil as the sole carbon source (BM+OL; for peroxisome biogenesis), and then were shifted back to CM (to induce pexophagy). Isolation of nucleic acids, total protein extraction, quantitative analysis of conidiation, and barley leaf explants infection follow the procedure as described [25].

Generation of deletion mutants
For gene deletion of ATG26 or PEX14, genomic DNA fragments (about 1 kb each) representing the 5' and 3' UTR of each gene were amplified by PCR, ligated sequentially so as to flank the HPH1 cassette. A cDNA encoding PEX14 61-361 was generated with AMV reversed transcriptase (Roche, Cat. No. 11495062001) and amplified by PCR, and expressed with PEX14 promoter and TrpC terminator, as a single copy in pex14Δ, with ILV1 cassette as fungal selection marker. Genomic DNA encoding PEX14 1-258 was amplified by PCR and expressed with PEX14 promoter and TrpC terminator, as a single copy in pex14Δ and with the ILV1 cassette as fungal selection marker. ScSNX41 or ScSNX42 coding sequence was PCR amplified and expressed with MPG1 promoter, in the plasmid with ILV1 cassette as fungal selection marker. The complementation construct was transformed into the GFP-SRL expressing snx41Δ mutant. The primers used for gene deletion or complementation are listed in Table 1. Underlined text represents the restriction enzyme site introduced for cloning purposes.

Microscopy and image processing
GFP epifluorescence was observed using a Zeiss LSM510 inverted confocal microscope (Carl Zeiss Inc.) equipped with a 30 mW Argon laser. The objective used was a 100 x Achromat (n.a. 1.25) oil immersion lens. EGFP was imaged with 488 nm wavelength laser excitation, using a 505-530 nm band pass emission filter or a 505 nm long-pass emission filter, while RFP imaging used 543 nm laser and a 560 nm long-pass emission filter.
For quantitative analysis of pexophagy based on fluorescent images, we calculated the percentage of vacuoles containing GFP-SRL out of total vacuoles, and the number represents Mean ± S.E. for three repeats for each strain (n>50). Densitometry analysis with the western blotting was performed with ImageJ, following the procedure as described [38]. Percentage reduction of Thiolase was calculated with three repeats for each strain and the number represents Mean ± S.E..

Biochemistry assays
Total protein extractions follow the reported procedure [24,25]. Protein samples from each extract (100 mg) were fractionated by SDS-PAGE, transferred onto nitrocellulose membrane (Millipore) and immunoblotted with appropriate primary antibody (anti-Thiolase, a gift from Suresh Subramani's lab, generated against ScPot1; or anti-Porin: Invitrogen, Cat. No.459500) at the recommended dilution. Secondary antibody conjugated to horse-radish peroxidase was used at 1:20000 dilution. The Amersham ECL™ Kit (GE, Cat. No. RPN2135) was used to detect the chemiluminescent signals in Western blotting experiments.