GUN Mutants: New Weapons To Unravel Ascospore Germination Regulation in the Model Fungus Podospora anserina

The regulation of ascospore germination in filamentous fungi has been poorly investigated so far. To unravel new genes involved in this regulation pathway, we conducted a genetic screen in Podospora anserina, and we isolated 57 mutants affected in ascospore germination. ABSTRACT In Podospora anserina as in many other Ascomycetes, ascospore germination is a regulated process that requires the breaking of dormancy. Despite its importance in survival and dispersal, ascospore germination in filamentous fungi has been poorly investigated, and little is known about its regulation and genetic control. We have designed a positive genetic screen that led to the isolation of mutants showing uncontrolled germination, the GUN (Germination UNcontrolled) mutants. Here, we report on the characterization of the gun1SG (Spontaneous Germination) mutant. We show that gun1SG is mutated in Pa_6_1340, the ortholog of Magnaporthe oryzae Pth2, which encodes a carnitine-acetyltransferase (CAT) involved in the shuttling of acetyl coenzyme A between peroxisomes and mitochondria and which is required for appressorium development. Bioinformatic analysis revealed that the mutated residue (I441) is highly conserved among Fungi and that the mutation has a deleterious impact on the protein function. We show that GUN1 is essential for ascospore germination and that the protein is localized both in mitochondria and in peroxisomes. Finally, epistasis studies allowed us to place GUN1 together with the PaMpk2 MAPK pathway upstream of the PaNox2/PaPls1 complex in the regulation of ascospore germination. In addition, we show that GUN1 plays a role in appressorium functioning. The pivotal role of GUN1, the ortholog of Pth2, in ascospore germination and in appressorium functioning reinforces the idea of a common genetic regulation governing both appressorium development and melanized ascospore germination. Furthermore, we characterize the second CAT encoded in P. anserina genome, Pa_3_7660/GUP1, and we show that the function of both CATs is conserved in P. anserina. IMPORTANCE The regulation of ascospore germination in filamentous fungi has been poorly investigated so far. To unravel new genes involved in this regulation pathway, we conducted a genetic screen in Podospora anserina, and we isolated 57 mutants affected in ascospore germination. Here, we describe the Germination UNcontrolled One (gun1SG) mutant, and we characterize the gene affected. GUN1 is a peroxisomal/mitochondrial carnitine-acetyltransferase required for acetyl coenzyme A shuttling between both organelles, and we show that GUN1 is a pleiotropic gene also involved in appressorium functioning similarly to its ortholog, the pathogenesis factor Pth2, in the plant pathogen Magnaporthe oryzae. Given the similarities in the regulation of appressorium development and ascospore germination, we speculate that discovering new genes controlling ascospore germination in P. anserina may lead to the discovery of new pathogenesis factors in pathogenic fungi. The characterization of GUN1, the ortholog of M. oryzae Pth2, represents a proof of concept.

ABSTRACT In Podospora anserina as in many other Ascomycetes, ascospore germination is a regulated process that requires the breaking of dormancy. Despite its importance in survival and dispersal, ascospore germination in filamentous fungi has been poorly investigated, and little is known about its regulation and genetic control. We have designed a positive genetic screen that led to the isolation of mutants showing uncontrolled germination, the GUN (Germination UNcontrolled) mutants. Here, we report on the characterization of the gun1 SG (Spontaneous Germination) mutant. We show that gun1 SG is mutated in Pa_6_1340, the ortholog of Magnaporthe oryzae Pth2, which encodes a carnitine-acetyltransferase (CAT) involved in the shuttling of acetyl coenzyme A between peroxisomes and mitochondria and which is required for appressorium development. Bioinformatic analysis revealed that the mutated residue (I441) is highly conserved among Fungi and that the mutation has a deleterious impact on the protein function. We show that GUN1 is essential for ascospore germination and that the protein is localized both in mitochondria and in peroxisomes. Finally, epistasis studies allowed us to place GUN1 together with the PaMpk2 MAPK pathway upstream of the PaNox2/PaPls1 complex in the regulation of ascospore germination. In addition, we show that GUN1 plays a role in appressorium functioning. The pivotal role of GUN1, the ortholog of Pth2, in ascospore germination and in appressorium functioning reinforces the idea of a common genetic regulation governing both appressorium development and melanized ascospore germination. Furthermore, we characterize the second CAT encoded in P. anserina genome, Pa_3_7660/GUP1, and we show that the function of both CATs is conserved in P. anserina. IMPORTANCE The regulation of ascospore germination in filamentous fungi has been poorly investigated so far. To unravel new genes involved in this regulation pathway, we conducted a genetic screen in Podospora anserina, and we isolated 57 mutants affected in ascospore germination. Here, we describe the Germination UNcontrolled One (gun1 SG ) mutant, and we characterize the gene affected. GUN1 is a peroxisomal/mitochondrial carnitine-acetyltransferase required for acetyl coenzyme A shuttling between both organelles, and we show that GUN1 is a pleiotropic gene also involved in appressorium functioning similarly to its ortholog, the pathogenesis factor Pth2, in the plant pathogen Magnaporthe oryzae. Given the similarities in the regulation of appressorium development and ascospore germination, we speculate that discovering new genes controlling ascospore germination in P. anserina may lead to the discovery of new pathogenesis factors in pathogenic fungi. The characterization of GUN1, the ortholog of M. oryzae Pth2, represents a proof of concept.
KEYWORDS mitochondria, Podospora anserina, Fungi, carnitine-acetyl transferase, appressorium, ascospore germination, filamentous fungi, peroxisomes type germination), provided in laboratory conditions in the specific germination G medium, supplemented with yeast extract (YE) to increase germination rate (42). Rather than screening for mutants unable to germinate, we set up a positive genetic screen allowing isolation of mutants producing spontaneously germinating ascospores on M2 medium. In parallel, we screened for suppressors of the germination defect of the DPaNox2 and DPaPls1 mutant strains (8,16) using the same protocol. Self-fertile (mat1/mat-) mycelia of the S, DPaNox2, and DPaPls1 strains were exposed to UV mutagenesis, and mutants producing spontaneously germinating ascospores on standard M2 medium were isolated (see Fig. S1 and Materials and Methods). Mutant screening allowed recovery of 22 suppressors of the DPaNox2 mutant, 16 suppressors of the DPaPls1 mutant, and 19 mutants from the S strain, all of them producing ascospores spontaneously germinating on standard M2 medium (see Table S1). Genetic analysis of these mutants revealed that for all of them, the mutant phenotype was controlled by a single locus. We then checked the germination process in these 57 mutants by microscopic analysis. For all the suppressors of the DPaNox2 and DPaPls1 mutants and 13 S strain mutants, ascospore germination was morphologically abnormal. In these mutants, we could observe germination through the primary appendage, the hyaline and elongated structure at the posterior part of the melanized ascospore that gives the name to the genus "Podo-spora" (Fig. 1; see also Movie S1). The latter selection was important to possibly discard structural mutants in which the ascospore cell wall was impaired, leading to "accidental" germination or mutants in which the cell constituting the primary appendage failed to degenerate. Indeed, in P. anserina the cell constituting FIG 1 Heterokaryotic ascospores morphology and germination. (A) Ascospore morphology. The crosses performed to obtain the different ascospore genotypes shown are indicated in the picture. GUN1/GUN1 (WT), gun1 SG /gun1 SG , and Dgun1 pGUN1/Dgun1 pGUN1 ascospores are fully melanized compared to Dgun1/Dgun1 ascospores which are partially demelanized. The last row shows FDS ascus composed of two GUN1/GUN1 (WT) and two Dgun1/Dgun1 (Dgun1) ascospores. The ascospores were mounted in water. Scale bar, 30 mm. (B) Ascospore germination. GUN1/GUN1 (WT) ascospores require G medium to germinate, while gun1 SG /gun1 SG and Papks1 193 /Papks1 193 ascospores germinate in M2 medium (and in G medium [not shown]). Germination occurs through the germination pore located at the tip of the ascospore. (Third column) The melanized ascospore has the PaPKS1/PaPKS1 genotype, and the nonmelanized ascospore has the Papks1 193 /Papks1 193 genotype. Scale bar, 10 mm.
Characterization of the gun1 SG Mutant in P. anserina Microbiology Spectrum the primary appendage degenerates; otherwise, a germ tube arises from this cell, leading to spontaneous germination. Only 6 of the mutants isolated from the wild-type S strain showed morphologically "normal" germination proceeding through the germination pore (see Table S1 and Movie S1). We speculated that these six mutants represent mutants of genes involved in the signaling pathway that controls germination and were named GUNx SG for Germination Uncontrolledx Spontaneous Germination , where "x" stands for the mutant number. Among these mutants, one had the particular characteristic to germinate on agar plates devoid of carbon and nitrate sources, while the five others did not (data not shown). We therefore started the characterization of this mutant that we named the gun1 SG mutant. This mutant differentiated ascospores with a normal shape (Fig. 1A) and with visually normal melanization and spontaneous germination through the germination pore (Fig. 1B). We determined the germination rate of this mutant on M2 medium, as well as on G1YE medium, and compared it to the wild type. Throughout our experiments, we never observed germination of wild-type ascospores when sown on M2 medium. In contrast, when gun1 SG /gun1 SG heterokaryotic ascospores were transferred onto M2 medium to estimate the germination rate, 54/100 germinated. In the same experiment, 94/100 germinated on G1YE, a rate comparable to wild-type (WT) ascospores (92/100). Once germination was initiated, development of the mycelium produced by the gun1 SG mutant was identical to the wild type: the gun1 SG mutant showed wild-type vegetative growth and mycelium morphology, fertility, ascospore production and appressorium formation. However, despite this apparently normal appressorium formation within cellophane, the gun1 SG mutant was delayed by 1 day compared to the wild type for breaching the cellophane layer ( Fig. 2 and Table 1; see also Fig. S2 and S3 in the supplemental material). GUN1 encodes a carnitin-acetyltransferase. The gene mutated in gun1 SG was identified through whole-genome sequencing. To that end, this mutant was backcrossed five times with the wild-type strain beforehand in order to eliminate most of the mutations generated during UV mutagenesis but not genetically linked to the mutation responsible for the mutant phenotype. The analysis of the gun1 SG whole-genome sequence revealed the presence of six silent mutations, and three missense mutations, one in Pa_1_13700, which encodes a putative protein of unknown function, another in Pa_5_7800, encoding a putative phosphoketolase, and the last one in the Pa_6_1340 CoDing Sequence (CDS), where an isoleucine was changed into an asparagine (I441N), caught our attention (Fig. 3). Transcriptome sequencing (RNA-seq) data indicated that Pa_6_1340 was strongly induced (fold change = 56) during ascospore germination, underlining the involvement of this gene during ascospore germination (A. Demoor, unpublished data). This CDS encodes a putative peroxisomal/mitochondrial carnitine-acetyltransferase (CAT) of 643 amino acids (43)(44)(45)(46). Homologs of this gene have previously been studied in S. cerevisiae, Aspergillus nidulans, Giberella zeae, Sclerotinia sclerotiorum, and M. oryzae, where they are involved in acetate/acetyl coenzyme A (acetate/acetyl-CoA) metabolism and more particularly in pathogenicity and appressorium development in the phytopathogenic species mentioned (36,37,39,(47)(48)(49). Regarding the role of acetate and peroxisomes in the control of germination in P. anserina and given the induction of Pa_6_1340 during ascospore germination, this gene emerged as a particularly good candidate for further study.
In order to explore the role of Pa_6_1340 in the ascospore germination process, a gene replacement was performed, in which the Pa_6_1340 CDS was substituted by a hygromycin B resistance marker (see Fig. S4). The gene disruption construct was introduced into a Dmus51::phleoR strain impaired for NHEJ (50). Two independent hygromycin B-resistant (hygR) transformants were obtained. In order to purify DPa_6_1340:: hygR from the Dmus51::phleoR mutation, we crossed both primary transformants with the S strain. Interestingly, we observed partially demelanized ascospores in the progeny of both crosses. Furthermore, when homokaryotic ascospores were sown on G1YE germination medium, only half of the progeny germinated, and those were only (hygS) melanized ascospores. This suggested that the (hygR) DPa_6_1340::hygR ascospores were the partially demelanized ones and that they were not able to germinate.
These crosses were repeated on M2 medium supplemented with tricyclazole, a fungicide impairing melanin synthesis and provoking spontaneous germination of ascospores in P. anserina (23). We collected ascospores directly projected on M2 medium supplemented with hygromycin B, and we isolated (hygR) germinating thalli. These thalli were fragmented and homokaryotic (hygR, phleoS) DPa_6_1340::hygR strains of each mating type were purified (see Materials and Methods). Deletion of Pa_6_1340 CDS (Dgun1) in these strains was verified by Southern blot analysis (see Fig. S4), and mutant phenotypes were characterized. In homozygous DPa_6_1340::hygR Â DPa_6_1340::hygR crosses, ascospores exhibited demelanization and completely lost their ability to germinate on G1YE medium, a phenotype opposite to that of the gun1 SG mutant (Dgun1 in Fig. 1). Genetic analyses of heterokaryotic ascospores showed that this phenotype due to DPa_6_1340::hygR deletion was recessive and that DPa_6_1340::hygR segregated with a second division segregation (SDS) rate of 55% (a detailed genetic analysis is provided in Materials and Methods under "tetrad analysis in the gun1 SG and in the Dgun1 strains"). Fertility in the DPa_6_1340::hygR strain was affected too: perithecium production in a DPa_6_1340::hygR Â DPa_6_1340::hygR (Dgun1) cross was slightly reduced, and ascospore production was significantly diminished compared to a wild-type cross (see Fig. S2). It has been shown that the formation of the appressorium in M. oryzae and the germination of melanized ascospores in P. anserina are two processes sharing common regulatory elements (16). We found that in the DPa_6_1340::hygR mutant, breaching of cellophane (a process involving appressorium development in P. anserina) was delayed compared to the wild type (Table 1). However, microscopic observations did not detect any morphological defect of appressorium development (Fig. S3). Importantly, introduction of the wild-type allele of Pa_6_1340 carried on the pGUN1 plasmid (see Materials and Methods) into the DPa_6_1340::hygR genome restored wild-type phenotypes, thus showing that the deletion of Pa_6_1340 (Dgun1) was responsible for the mutant phenotypes ( Fig. 1, 2, and 4; see also Fig. S2).
Finally, we addressed the question of whether the phenotypes in the gun1 SG mutants were due to the I441N mutation in Pa_6_1340 or not, by testing whether DPa_6_1340::hygR and gun1 SG are alleles of the same gene by means of a complementation test. First, we genetically determined that spontaneous germination in gun1 SG was a recessive trait, a prerequisite for the complementation test. These genetic analyses also showed that gun1 SG segregated with a SDS rate of 54%, very similar to that of DPa_6_1340::hygR (55%), suggesting that DPa_6_1340::hygR and gun1 SG mutants could be allelic (see "Tetrad analysis in the gun1 SG and in the Dgun1 strains" in Materials and Methods). We then crossed gun1 SG with DPa_6_1340::hygR strains, reasoning that if DPa_6_1340::hygR and gun1 SG strains are allelic, the SDS asci of the F 1 progeny will be composed of four DPa_6_1340::hygR/gun1 SG heterokaryotic ascospores showing no functional complementation: these ascospores can germinate spontaneously on M2 medium supplemented with hygromycin B (note that, since the gun1 SG phenotype is not fully penetrant, the number of ascospores germinating per ascus can vary); in contrast, if gun1 SG and DPa_6_1340::hygR strains are not allelic (i.e., they are alleles belonging to two different loci: the "Pa_6_1340" locus and a The listed strains were cultured for 5 days on M2 medium topped with a cellophane layer at 27°C. After 2, 3, 4, and 5 days of culture, the cellophane layer was removed to check fungal growth in the medium underneath. The day the cellophane was breached is indicated. Three replicates were made for each day of culture. This experiment was repeated twice. the "GUN1" locus), functional complementation leading to restoration of wild-type phenotype (no spontaneous germination) is expected in SDS asci. In the F 1 progeny of the gun1 SG Â DPa_6_1340::hygR cross, we observed in SDS asci that up to four heterokaryotic ascospores germinated spontaneously on M2 medium supplemented with hygromycin B (no functional complementation), showing that gun1 SG and DPa_6_1340 strains were allelic. This demonstrated that Pa_6_1340 was the gene mutated in the gun1 SG mutant responsible for the spontaneous germination phenotype. We therefore named Pa_6_1340, GUN1. This evidence was confirmed when we showed that the gun1 SG mutant was also complemented by ectopic integration of a wild-type copy of Pa_6_1340/GUN1 carried by the pGUN1 plasmid (Table 2; see also Materials and Methods). (B) GUN1 amino acid sequence. The MTS is highlighted in blue, the histidine of the catalytic site is highlighted in green, the I441 is highlighted in magenta, and the AKI PTS1 is highlighted in yellow. (C) 3D structure of murine CAT (PDB 2H3P) (left) and 3D modelization of GUN1 using the I-TASSER modeling tool (right). The isoleucine, I441, in GUN1 and the proline P422 (numbered P396 in the PDB 2H3P crystallized protein) corresponding to the proline aligned to the isoleucine 441 in the murine CAT are indicated by an arrow.
Characterization of the gun1 SG Mutant in P. anserina Microbiology Spectrum P. anserina possesses two CATs. A BLASTP search on the P. anserina predicted CDS database (http://podospora.i2bc.paris-saclay.fr/) identified a second CAT encoded by the Pa_3_7660 putative CDS. Compared to GUN1, this putative enzyme did not harbor any localization signal. Similarly to other Fungi, P. anserina may be endowed with two types of CAT, one located in peroxisomes and in mitochondria (GUN1) and one remaining in the cytoplasm (Pa_3_7660). To confirm this hypothesis, we undertook a phylogenetic analysis of CATs in Fungi and searched the homologs of GUN1 in the genome sequence of representative fungal species (see Materials and Methods). In these species, two CATs were always identified, except for S. cerevisiae in which three CATs were identified as previously shown (49). The protein sequences were aligned, and the corresponding phylogenetic tree was built (see Fig. S5). The phylogeny of CATs in Fungi clearly indicated that there are two main types of CATs in Fungi: the putative peroxisomal/mitochondrial CAT, including GUN1, A. nidulans AcuJ, and M. oryzae Pth2/ Crat1, and the putative "cytoplasmic" CATs, including P. anserina Pa_3_7760, A. nidulans FacC, and M. oryzae Crat2. Careful analysis of protein sequences indicated that proteins of the former type all contained the appropriate sequence signals to locate in peroxisomes and in mitochondria. In addition, search for orthologs through OrthoDB (51) did not identify orthologs for either GUN1 or Pa_3_7760 in plants or in bacteria.
Pa_3_7760 putative CDS was renamed GUP1 for GUN1 Paralog 1. To determine the function of this second CAT and assess its role in ascospore germination, we undertook FIG 4 Relative CAT enzymatic activity in mycelial extracts. CAT activity was assayed through the spectrophotometric measure of CoA-SH produced per min per mg of protein in cell extracts in the presence of carnitine. Activities are reported as the activity ratio of the wild-type (WT) S strain. The CAT activity means and standard deviations have been calculated based on four to seven biological replicates (N is indicated for each genotype). An exact two-sample Fisher-Pitman permutation test was used to compare CAT activities. *, CAT activities significantly different from the WT (P , 0.05); **, CAT activities in complemented strains significantly different from the activity in the respective mutant strains (P , 0.05).  Fig. S6). The Dgup1 strain was purified from the Dmus51 mutation by crossing primary transformants with the wild-type S strain followed by the selection of (phleoR genS) homokaryotic ascospores of the Dgup1 genotype in the progeny. Dgup1 ascospores had no melanization defect and germinated on G1YE medium as the wild type. Importantly, when Dgup1 homokaryotic ascospores were sown on M2 medium, no spontaneous germination was observed, leading us to conclude that GUP1 deletion had no effect on ascospore melanization and germination in P.
anserina. In addition, the Dgup1 strain differentiated wild-type mycelium and exhibited wild-type fertility, ascospore production, appressorium development, and cellophane breaching ( Fig. 2 and Table 1; see also Fig. S2 and S3). We then constructed the Dgun1 Dgup1 double mutant (see Materials and Methods) and observed that Dgun1 Dgup1 homokaryotic ascospores exhibited impaired melanization and lack of germination similarly to the Dgun1 ascospores. GUN1 and GUP1 functions are conserved in Fungi. Previous studies in Fungi have demonstrated that CATs play an important role in primary metabolism and carbon source utilization. In particular, it has been shown that knockout strains of the cytoplasmic CAT do not grow on acetate, whereas mutant strains of the peroxisomal/mitochondrial CAT do not grow on acetate and on media containing fatty acids, such as oleic acid (36,39,(47)(48)(49). To assess the role of GUN1 and GUP1 in carbon source utilization, we monitored the growth of the different mutant strains on acetate and oleic acid. As shown in Fig. 2, the wild-type S strain was able to grow on all the tested media, including the Tween 40 control (Tween 40 is necessary to solubilize oleic acid), indicating that P. anserina was able to use this detergent as a carbon source. Remarkably, the gun1 SG strain, as well as the gun1 SG pGUN1 complemented strain, grew as the wild type. The Dgun1, Dgup1, and the double Dgun1 Dgup1 mutants grew as the wild type on dextrin (M2 medium), and the Dgun1 and Dgun1 Dgup1 strains exhibited slightly reduced growth on Tween 40, almost no growth on oleic acid, and no growth on acetate, whereas Dgup1 growth was impaired only on acetate. The Dgup1 pGUP1 strain had restored wild-type growth on acetate, indicating that wild-type GUP1 complemented Dgup1 deletion. This confirmed that GUP1 function was required in P. anserina on acetate. Similarly, wild-type growth was restored on Tween 40 and oleic acid in both Dgun1 pGUN1 and Dgun1 Dgup1 pGUN1 complemented strains, indicating that GUN1 function was required for Tween 40 and oleic acid utilization. Strikingly, both the Dgun1 and the Dgun1 Dgup1 strains could grow on Tween 40 but not on oleic acid medium, although the latter contained the same amount of Tween 40 (0.5%; see Materials and Methods). This suggested that impaired growth on oleic acid for Dgun1 and Dgun1 Dgup1 strains was due to a toxic effect of oleic acid in these mutant strains. Overall, these data showed that the roles of both main types of CATs were conserved in P. anserina: CATs of the AcuJ/Pth2/Crat1/GUN1 type are required for growth on acetate, as well as on long-chain fatty acids, whereas CATs of the FacC/Crat2/GUP1 type are required for growth on acetate.
Structure prediction analysis of gun1 SG loss of function. MAFFT alignment with the protein sequences encoded by the fungal orthologs of GUN1, including the ortholog from Homo sapiens, revealed that the isoleucine 441 mutated in the gun1 SG mutant (I441N) was highly conserved in Fungi: it is conserved in Pezizomycotina, Saccharomycotina, Mucoromycota, and Basidiomycota (see Fig. S7). Using I-TASSER, we carried out threedimensional (3D) structure prediction of GUN1, and we compared this to the 3D structure of the murine CAT (PDB 2H3P, 34% identity). As can be seen in Fig. 3C, the overall structure of both enzymes was well conserved, showing that the 3D modeling of GUN1 was congruent. Based on this GUN1 3D model, we could localize the isoleucine 441 near the extremity of an a-helix. Since the hydrophobicity of amino acids is paramount in a-helix formation, we wondered whether the substitution of aliphatic isoleucine 441 by polar asparagine could destabilize the a-helix and/or the whole protein. We determined the Gibbs freeenergy Gap (DDG) induced by the I441N substitution in gun1 SG with STRUM (52). The calculated DDG of 2.11 kcal mol 21 was indeed indicative of a destabilization of the gun1 SG mutant protein, but this low DDG value (,6 kcal mol 21 ) was indicative of a local destabilization of the protein rather than a complete destabilization (53). In line with this, we did not notice any stability issues in both chimera reporters gun1 SG -mCherry and gun1 SG -mCherry-AKI compared to GUN1-mCherry and GUN1-mCherry-AKI, respectively, in our microscopic observations (see below). We also used the PROVEAN (Protein Variation Effect Analyzer) analysis tool to determine the impact of the I441N substitution on GUN1 function. PROVEAN is a software tool predicting the potential deleterious effect of a point mutation (usually amino acid substitutions or indel). In keeping with the recessive nature of the gun1 SG mutation, the calculated PROVEAN score of 26.63, far below the predefined cutoff of 22.5, was predictive of a "deleterious" loss-of-function effect of the I441N substitution on GUN1 function (54,55).
CAT activity decreases in gun1 SG and Dgun1 strains. We measured the CAT activity in different mutant strains in protein extracts from mycelium grown on M2 medium (Fig. 4). It is worth mentioning that we failed to measure CAT activity in ascospores, and we could only obtain reliable results in mycelia (see Materials and Methods). Compared to the wild type, the CAT activity in the mycelium was greatly reduced in Dgun1, Dgun1 Dgup1, and gun1 SG mutants. Significantly, CAT activity was restored to the wild-type level in the Dgun1 pGUN1 and Dgun1 Dgup1 pGUN1 complemented strains, confirming that lack of GUN1 was responsible for reduced CAT activity. In the complemented gun1 SG pGUN1 strain, the CAT activity was even higher than in the wild-type pointing to a role of CAT activity increase in the restoration of the wild-type phenotype in gun1 SG pGUN1. Interestingly, CAT activity in the Dgup1 strain was similar to wild-type CAT activity, a result in line with previous observations in M. oryzae showing that CAT activity in Dcrat2 (the ortholog of GUP1) mutants was not altered (37). These results, showing a decreased CAT activity in gun1 SG , and hence a loss of function of gun1 SG , were congruent with the modelized "deleterious" effect of the I441N mutation on gun1 SG function and the recessivity of the gun1 SG phenotype. However, the fact that similar reduced CAT activity was measured in gun1 SG and Dgun1 (P , 0.05) strains suggested that the CAT activity measured in the mycelium did not account for the difference in phenotype between gun1 SG and Dgun1 mutants (i.e., germination and melanization of ascospores, acetate, and acid oleic growth).
Subcellular localization of GUN1 and gun1 SG proteins. Previous studies carried out on GUN1-type CATs in other fungal species have indicated that these enzymes could be localized in peroxisomes and in mitochondria (39,56). Analysis of GUN1 protein sequence using wolfPSORT showed that GUN1 is probably located in both peroxisomes and mitochondria. Accordingly, scanning GUN1 protein sequence with MitoFates allowed us to identify a mitochondrial targeting sequence (MTS), and we manually identified the "AKI" tripeptide at the C-terminal end of the protein sequence as a type 1 peroxisomal targeting sequence (PTS1) (57) (Fig. 3). In order to investigate GUN1 subcellular localization, as well as the impact of the gun1 SG mutation on its subcellular localization, we tagged both GUN1 and gun1 SG proteins with mCherry and with mCherry-AKI, a modified mCherry version bearing the putative PTS1 peroxisome targeting signal of GUN1 in the C terminus (Fig. 3). As previously mentioned, GUN1 is specifically induced in ascospores during germination (Demoor, unpublished). To ensure that expression of the fusion proteins was under the control of the native GUN1 regulatory sequences, we tagged the endogenous GUN1 alleles (at the GUN1 locus) through insertion of the mCherry and mCherry-AKI coding sequences in 39 of GUN1 and gun1 SG CDS by homologous recombination (see Fig. S8 and Materials and Methods). Four tagged strains were obtained: GUN1-mCherry, gun1 SG -mCherry, GUN1-mCherry-AKI, and gun1 SG -mCherry-AKI. Importantly, GUN1-mCherry and GUN1-mCherry-AKI tagged strains germinated like the wild type, whereas gun1 SG -mCherry and gun1 SG -mCherry-AKI tagged strains germinated spontaneously on M2 medium like the gun1 SG mutant. This suggested that the tagging by mCherry and mCherry-AKI did not modify GUN1 and gun1 SG functions during ascospore germination. Each strain was crossed with strains expressing green fluorescent protein (GFP) markers tagging either mitochondria (mito-GFP) or peroxisomes (GFP-SKL) to obtain double-tagged strains in the progeny (58,59). Importantly, the presence of the mito-GFP or the GFP-SKL reporter genes did not modify ascospore germination. These double tagged strains allowed us to observe subcellular localization of GUN1 and gun1 SG in mycelium ( Fig. 5) but not in melanized ascospores (data not shown). To that end, each strain was crossed with the Papks1 136 mutant producing partially demelanized ascospores in order to obtain all the double-tagged strains in a Papks1 136 genetic background in the progeny. In contrast to the Papks1 193 mutation, which leads to spontaneous ascospore germination (Fig. 1B), the Papks1 136 mutation did not modify ascospore germination. All the Papks1 136 double-tagged strains produced partially demelanized ascospores, allowing fluorescence microscopic observations within ascospores, and were isolated in both mating types (mat1 and mat-) in order to proceed to homozygous crosses for ascospore production and observation. Ascospores were observed either in M2 liquid medium (noninduction condition) or in G liquid medium for the induction of ascospore germination (Fig. 6). We observed in ascospores and in mycelium that GUN1-mCherry-AKI and gun1 SG -mCherry-AKI could colocalize with both mito-GFP and GFP-SKL, showing that both GUN1-and gun1 SG -mCherry-AKI reporter proteins could be found in mitochondria and in peroxisomes (Fig. 5). Colocalization of GUN1 and gun1 SG -tagged proteins with mitochondria and peroxisomes was qualitatively determined by visual analysis of images and quantified by calculating the Pearson's correlation coefficient (PCC) between the green and the red fluorescence signals. PCCs range from 1 for two images whose fluorescence signals are perfectly linearly related to 21 for two images whose fluorescence signals are perfectly, but inversely, related to one another, with intermediate values indicating partial colocalization. Values near zero reflect distributions of signals that are uncorrelated with one another (60). Careful comparison of the GFP-tagging pattern and the mCherry tagging pattern indicated that GUN1-mCherry-AKI and gun1 SG -mCherry-AKI could be absent in some mitochondria or in some peroxisomes. This partial colocalization of both tagged proteins with peroxisomes and mitochondria was quantified by PCCs comprised between 0.21 (GUN1-mCHerry-AKI with mito-GFP in noninduced ascospores and gun1 SG -mCherry-AKI with GFP-SKL in induced ascospores) and 0.58 (GUN1-mCherry-AKI with mito-GFP in the mycelium) ( Fig. 5C and 6C). It should be stressed that the low PCC of 0.21 calculated for GUN1-mCHerry-AKI with mito-GFP in noninduced ascospores corroborated their evident lack of colocalization observed in Fig. 6B. In line with this, comparably low PCCs (0.15 and 0.23, respectively) were calculated for GUN1-mCherry and gun1 SG -mCherry with GFP-SKL in the mycelium (Fig. 5C). Accordingly, Fig. 5A clearly shows that GUN1-mCherry and gun1 SG -mCherry did not seem to colocalize with GFP-SKL. Compared to GUN1-mCherry-AKI, gun1 SG -mCherry-AKI was the protein colocalizing the less with GFP-SKL in mycelium and in ascospores. Inversely, gun1 SG -mCherry-AKI colocalization with mito-GFP in spontaneously germinating ascospores (gun1 SG -mCherry-AKI ascospores germinate in M2 and G media) was more important ( Fig. 6B and C). In comparison, GUN1-mCherry-AKI colocalization with mito-GFP was significantly lower in ascospores under induced (PCC = 0.26) and noninduced (PCC = 0.21) conditions for germination ( Fig. 6B and C). Altogether, these data showed that the distribution of gun1 SG -mCherry-AKI in mitochondria and in peroxisomes was different from the one of GUN1-mCherry-AKI, especially in ascospores where gun1 SG -mCherry-AKI localized significantly more in mitochondria and less in peroxisomes than GUN1-mCherry-AKI.
We observed that GUN1-mCherry and gun1 SG -mCherry colocalized only with the mito-GFP reporter in ascospores (see Fig. S9), as well as in mycelium (Fig. 5B). Similarly to the mycelium, a clear discrepancy in the localization of both GUN1-and gun1 SG -mCherry-tagged proteins and the GFP-SKL marker was observed in ascospores (see Fig. S9A), and low PCCs (from 0.01 to 0.06) were measured. This observation was consistent with previous studies on the localization of G. zeae CAT1-GFP (the ortholog of GUN1) showing that adding the GFP in C terminus of the protein masked the PTS1 signal of CAT1 (48). More generally, it has been shown that adding a tag after a C-terminal PTS1 signal abolishes import into peroxisomes, suggesting that GUN1-mCherry and gun1 SG -mCherry could be mislocalized (61,62). It is worth noting that ascospores in the GUN1-mCherry tagged strain germinated as wild type and ascospores in gun1 SG - Characterization of the gun1 SG Mutant in P. anserina Microbiology Spectrum mCherry tagged strain germinated spontaneously, strongly suggesting that mislocalization of GUN1-or of gun1 SG -mCherry did not affect ascospore germination. In contrast, whereas GUN1-mCherry-AKI and gun1 SG -mCherry-AKI strains grew as the wild type on the different media tested, the GUN1-mCherry and gun1 SG -mCherry strains did not grow on oleic acid (see Table S3). This result suggested that peroxisomal localization of GUN1 (and gun1 SG ) was required for oleic acid metabolism. Cross-talk between GUN1 and the MAPK PaMpk2 pathway upstream of the PaNox2/PaPls1 complex. We have shown in previous studies that the MAPK PaMpk2 pathway, PaNox2, its regulatory subunit PaNoxR, and PaPls1 are essential for ascospore germination in P. anserina: the deletion of these genes blocks ascospore germination; inversely, constitutive phosphorylation/activation of PaMpk2 in the PaMKK2 c mutant carrying a constitutively active allele of PaMKK2, triggers spontaneous ascospore germination (Fig. 7) (8-10, 17). We took advantage of the spontaneous germination phenotype of the gun1 SG mutant to conduct epistasis studies in order to place GUN1 in the regulatory cascade triggering ascospore germination. These studies were carried out by crossing gun1 SG with the DPaMpk2, DPaPls1, and DPaNox2 strains. A total of 48 homokaryotic spores were sown on M2 medium for each cross, but none of the germinated spores presented the DPaMpk2, DPaPls1, or DPaNox2 deletions (Table 3). This clearly showed that PaMpk2, PaPls1, and PaNox2 were required for germination in gun1 SG ascospores. In keeping with this, we found  that PaMpk2 was constitutively phosphorylated in gun1 SG ascospores. In absence of germination induction, PaMpk2 was phosphorylated at an even higher level in gun1 SG ascospores than in wild-type ascospores induced for germination (Fig. 7). We then crossed the Dgun1 strain with the PaMKK2 c mutant showing spontaneous ascospore germination (17). In the progeny of this cross, 305 homokaryotic ascospores were sown on M2 medium and 208 on G 1 YE medium (Table 4). Strikingly, no (hygR) Dgun1 bearing ascospore germinated on either medium. In addition to the very low germination rate observed (see below), the very low number of wild-type ascospores germinating on G 1 YE medium suggested that genetic linkage decreased recombination between the PaMKK2 c transgene insertion site and the GUN1 locus, leading to a reduction in the expected number of Dgun1 PaMKK2 c ascospores in the progeny. To test this hypothesis independently of the germination defect due to Dgun1 deletion, we crossed the PaMKK2 c strain with the GUN1-mCherry-AKI strain obtained by homologous recombination at the GUN1 locus (see Materials and Methods, Fig. S8). 84/123 homokaryotic ascospores germinated on G 1 YE medium (Table S4). In this cross, we observed a reduced number of progeny with recombined genotype (x 2 -test, P value = 1.5 Â 10 26 ) arguing for genetic linkage between the PaMKK2 c transgene insertion site and the GUN1 locus (recombination frequency, r = 0.202; genetic distance = 20.2 cM). This recombination frequency (r = 0.202) was used in the analysis of the Dgun1 Â PaMKK2 c cross to calculate the "expected" size of each genotypic category in the progeny (Table 4).
In particular, 21 Dgun1 PaMKK2 c ascospores were expected on G 1 YE and 30.5 on M2 (Table 4). We also determined that the germination rate of PaMKK2 c ascospores was 49% (41/83) on G 1 YE and 22% (27/122) on M2. Because spontaneous ascospore germination on M2 due to PaMKK2 c was 22% in this cross, we estimated that if Dgun1 had no influence on germination, a maximum of 6.8 ascospores of the Dgun1 PaMKK2 c genotype might FIG 7 Western blot analysis of PaMpk2 phosphorylation in ascospores. Proteins were extracted from genetically homogeneous ascospores. Anti-p44 and anti-phospho-p44 antibodies recognizing both PaMpk1/PaMpk2 and p-PaMpk1/p-PaMpk2, respectively, were used. For ascospore germination induction (WT ind.), ascospores were placed on G1YE medium for 2 h before protein extraction (see Materials and Methods). PaMKK2 c constitutive mutation induces PaMpk2 phosphorylation and spontaneous ascospore germination. The original blots were spliced, and noncontiguous lanes are separated. have germinated. Finally, the lack of (hygR) ascospores in the progeny showed that Dgun1 impaired both ascospore germination on G 1 YE as well as PaMKK2 c -induced spontaneous ascospore germination on M2. Altogether, these data clearly indicated that gun1 SG triggered spontaneous ascospore germination through the activation of the PaMpk2 MAPK pathway and they also showed that GUN1 function was required when spontaneous ascospore germination was triggered by the constitutive activation of PaMpk2 (Fig. 8).
gun1 SG -induced ascospore germination requires PEX5 and PEX13. Peroxisomes are key organelles for germination. It has been shown that mutants of the peroxisomal matrix protein import machinery (such as Dpex5, Dpex7, and Dpex13 mutants) produce partially demelanized ascospores impaired for germination (29,33,63). We crossed gun1 SG with the Dpex5 Dpex7 double mutant and with the Dpex13 mutant and checked whether Dpex5 gun1 SG , Dpex7 gun1 SG , and Dpex13 gun1 SG double mutants could germinate when sown on G1YE germination medium (Tables 5 and 6). None of the germinated ascospores carried Dpex5 or Dpex13, showing that Dpex5, Dpex13, Dpex5 gun1 SG , and Dpex13 gun1 SG ascospores did not germinate. The fact that Dpex5 and Dpex13 mutants were epistatic over the gun1 SG strain indicated that the function of both genes was required for gun1 SG -induced ascospore germination.
PEX7 is located approximately 810,000 bp away from GUN1 on chromosome 6. Therefore, genetic linkage between both loci should lead to a reduced number of recombinant Dpex7 gun1 SG progeny compared to Dpex7 parental progeny, regardless of any genetic interaction between Dpex7 and gun1 SG strains (i.e., suppression or epistasis). Interestingly, only 2 Dpex7 ascospores germinated, a really low number compared to the 27 recombinant Dpex7 gun1 SG ascospores that germinated. This result, opposite to what FIG 8 Regulation of ascospore germination. Breaking of dormancy is initiated by germination triggers: ammonium acetate and Bacto peptone in vitro. The activities of the PTS1 and PTS2 import receptors PEX5 and PEX7, as well as that of the importomer component Pex13, are required for germination. We speculate that GUN1-driven acetyl-CoA shuttling to mitochondria activates the Mpk2/FUS3 MAPK pathway, which in turn activates the NADPH oxidase complex Nox2/Pls1/NoxR. Reciprocally, the Mpk2/ FUS3 MAPK module requires GUN1 function to activate ascospore germination. Whether activation of the Nox2 complex sets up a septin ring and actin cytoskeleton rearrangement at the germination pore in a similar manner as in M. oryzae appressorium remains to be addressed.
Characterization of the gun1 SG Mutant in P. anserina Microbiology Spectrum could be expected in case of genetic linkage only, suggested that Dpex7 ascospores were strongly impaired for germination and that the gun1 SG strain could suppress this defect in the recombinant Dpex7 gun1 SG ascospores. However, it was previously shown that in a heterozygous cross, the germination of Dpex7 ascospores is only moderately affected (29). Hence, the fact that only 2 Dpex7 ascospores germinated could reveal complex genetic interactions between Dpex7, Dpex5, and gun1 SG strains and, mostly, a non-cell-autonomous effect of the Dpex5 mutation on the germination of ascospores as previously observed (29). We therefore directly tested the genetic interaction between Dpex7 and gun1 SG strains by crossing the Dpex7 strain with the gun1 SG strain (Table 7). Only 50/90 ascospores germinated on G + YE medium. We performed a x 2 test that determined that the different genotypic categories obtained were not statistically different in size (P = 0,059). In conclusion, no genetic interaction between Dpex7 and gun1 SG strains was observed, gun1 SG had no suppressive effect on Dpex7 ascospore germination defect. Accordingly, Dpex7 gun1 SG ascospores had a similar melanization defect (gray color) as Dpex7 ascospores, showing that gun1 SG had no suppressive effect on the melanization defect of Dpex7 ascospores.

DISCUSSION
Despite its importance in the fungal life cycle, the regulation of ascospore germination in filamentous fungi has not been thoroughly researched. In this study, we aimed at uncovering new actors of this regulation pathway in P. anserina. To that end, we conducted a direct genetic screen, a particularly powerful approach to identify new genes. A majority of the mutants isolated, including all suppressors of DPaNox2 and DPaPls1, showed spontaneous but abnormal ascospore germination. However, for six mutants, germination proceeded as in the wild type through the germination pore at the tip of the ascospore. We hypothesized that in these mutants, ascospore dormancy was broken and that these were mutants specifically impaired in the control of ascospore germination. Here, we characterize the first of these six Germination UNcontrolled-GUN mutants (gun1 SG ). Although most of the spontaneous germination analyses were performed on M2 medium containing dextrin, the gun1 SG mutant germinated on medium lacking carbon and nitrate sources, indicating that in this mutant, control of dormancy escaped  Characterization of the gun1 SG Mutant in P. anserina Microbiology Spectrum possible nutrient stimuli (data not shown). Through whole-genome sequencing and genetic analyses, we showed that the gene mutated in gun1 SG is the Pa_6_1340 putative CDS. This CDS encodes a peroxisomal/mitochondrial carnitine-acetyltransferase (CAT), a key metabolic enzyme involved in acetyl-CoA shuttling between peroxisomes and mitochondria (45,64). Gene expression of this CAT is significantly induced during ascospore germination, highlighting the pivotal role of this enzyme during this process (Demoor, unpublished). The mutation in gun1 SG CDS is a substitution of the conserved isoleucine 441 by an asparagine (I441N). 3D modelization of GUN1 and of the mutated gun1 SG protein combined with in silico analyses of the stability of gun1 SG predict a moderate effect of the I441N substitution on the overall stability of the protein but a deleterious effect on its activity. This prediction correlates with the recessive nature of the gun1 SG mutation, as well as the reduced CAT activity measured in gun1 SG , both pointing to a loss of function of the gun1 SG allele. Hence, GUN1 may act as an ascospore germination inhibitor. However, we also show that deletion of GUN1 leads to a complete lack of germination and a defect in melanin synthesis, indicating that GUN1 function is required for both processes and suggesting that the gun1 SG allele is hypomorphic compared to Dgun1, which is a null allele. Impairment of both germination and melanization are frequently observed in mutants of peroxisomal import machinery, as well as in mutants of peroxisomal/mitochondrial b-oxidation (28,58,63). Indeed, in P. anserina, ascospores in the DechA mutant impaired for mitochondrial b-oxidation and in the Dfox2 mutant impaired for peroxisomal b-oxidation show a reduced rate of germination (28). Interestingly, the Dgun1 strain exhibits a complete lack of germination, suggesting that GUN1 and acetyl-CoA shuttling between peroxisomes and mitochondria is essential during activation of ascospore germination. As with many other fungi, we found a second CAT encoded in the P. anserina genome. Deletion of this second CAT encoding gene, Pa_3_7660/GUP1 (GUN1 Paralog 1), does not show any ascospore germination defect nor melanization defect, suggesting that this CAT has no role in both processes. Furthermore, CAT activity is reduced in the Dgun1 mutant and in the gun1 SG strain but not in the Dgup1 mutant (in mycelium). This finding is similar to what has been observed in M. oryzae where Pth2 mutants (GUN1 ortholog) show reduced CAT activity and reduced pathogenicity, while the Dcrat2 knockout strain (GUP1 ortholog) exhibits wild-type CAT activity and pathogenicity (37). As previously found in Fungi (M. oryzae, A. nidulans, G. zeae, and S. cerevisiae), we show that in P. anserina, the peroxisomal/mitochondrial CAT GUN1 is required for growth on acetate and oleic acid while the cytosolic CAT GUP1 is required for growth on acetate only, showing strong conservation of the function of both enzymes in fungal primary metabolism (36,39,41,(47)(48)(49). Nonetheless, GUN1 is required for growth on fatty acids, but we also show that oleic acid is toxic for Dgun1 mutants, as previously demonstrated for pex2 mutants impaired for peroxisomal import (58). Indeed, both Dgun1 and Dgun1 Dgup1 strains grow better on Tween 40 control medium than on oleic acid medium containing similar amount of Tween 40. In contrast to peroxisomal import mutants such as pex2 mutants, which are sterile in homozygous cross, fertility and ascospore production are only moderately decreased in Dgun1 and Dgun1 Dgup1 mutants.
To understand how the mutation in gun1 SG triggers breaking of dormancy, we explored both gun1 SG CAT activity and gun1 SG subcellular localization. The gun1 SG strain shows a loss in CAT activity in the mycelium similar to that of the Dgun1 mutant. However, Dgun1 and gun1 SG mutant strains exhibit noticeable phenotype discrepancies: the Dgun1 mutant produces nongerminating demelanized ascospores, while the gun1 SG mutant produces spontaneously germinating melanized ascospores; also, the gun1 SG mutant grows as the wild type on both oleic acid and acetate but not the Dgun1 mutant. We were not able to measure CAT activity in ascospores and therefore to properly address the question of CAT activity during germination in gun1 SG and Dgun1. The fact that both Dgun1 and gun1 SG mutants have a reduced CAT activity but different phenotypes is intriguing, and one cannot exclude that spontaneous germination in the gun1 SG mutant could be due to an as-yet-unknown activity of GUN1, not related to acetyl-CoA shuttling but specifically involved in the control of dormancy in ascospores.
To explore how gun1 SG causes spontaneous germination, both the wild-type GUN1 protein and the mutant gun1 SG protein were tagged with mCherry or mCherry-AKI (AKI is the PTS1 peroxisomal import signal present in C terminus of GUN1), and colocalization studies with GFP-tagged peroxisomes and GFP-tagged mitochondria were performed. In the mycelium, as well as in ascospores, although GUN1 is found both in peroxisomes and in mitochondria, our data show that GUN1 is preferentially located in peroxisomes. The dual localization of GUN1 in mitochondria and in peroxisomes is in agreement (i) with the predicted localization of this CAT, bearing both a mitochondrial targeting sequence (MTS) at the N terminus and the peroxisomal targeting sequence (PTS1) AKI at the C terminus and (ii) with studies of GUN1 orthologs in other Fungi such as A. nidulans and G. zeae (39,48). In contrast, M. oryzae Pth2 was shown to only localize in peroxisomes (36). Attention must be drawn to the fact that the main difference between GUN1 and gun1 SG is their respective distribution between mitochondria and peroxisomes, the gun1 SG mutant protein being predominantly present in mitochondria, while the GUN1 wild-type protein is preferentially located in peroxisomes. Furthermore, GUN1 seems to be almost exclusively located in peroxisomes in dormant ascospores not subjected to a germination trigger. These data suggest that the breaking of dormancy in wild-type and gun1 SG ascospores may involve shuttling of gun1 SG from peroxisomes to mitochondria. We show that, in keeping with the role of mitochondria in ascospore germination, the mislocalization of GUN1-mCherry (and gun1 SG -mCherry) solely in mitochondria does not impair the germination or melanization of ascospores. However, GUN1-mCherry and gun1 SG -mCherry strains do not grow on oleic acid, in the same way as the Dgun1 strain. A comparable effect has been reported in A. nidulans, where mislocalized AcuJ in the cytoplasm impairs growth on oleic acid but not on acetate (39). These data point to a central role of peroxisomal localization of GUN1 in oleic acid utilization independent of ascospore germination. However, it has been shown that peroxisomes are essential for germination in P. anserina, and our data confirm that Dpex5 and Dpex13 strains lacking PTS1-dependent peroxisomal import machinery cannot germinate (29,33,63). How ascospores carrying mislocalized GUN1 germinate is an open question. Given that GUN1 bears a PTS1-AKI at the C terminus (and no internal PTS2) and that the Dpex5 mutant is epistatic over gun1 SG , it is highly likely that GUN1 (and gun1 SG ) peroxisomal import is dependent on the PTS1 and not on the PTS2 import pathway. In this study, we could not determine any genetic interaction specifically between Dpex7, a deletion impairing PTS2-dependent import machinery, and the gun1 SG mutation. However, we found that Dpex7 gun1 SG ascospores germinated at a higher rate than Dpex7 ascospores in the progeny when the Dpex5 Dpex7 double mutant was crossed with the gun1 SG strain. However, this result was specifically observed when the Dpex5 mutation was present in the parental strain, confirming, as previously shown, that the Dpex5 mutation has a noncell-autonomous effect on ascospore germination (since it affects the germination of ascospores without the Dpex5 deletion) (29). The Dpex7 mutation has been shown to suppress some Dpex5 mutant phenotypes during sexual reproduction (29); inversely, our results suggest that the Dpex5 mutation could increase Dpex7 germination defect and that gun1 SG would suppress this Dpex5-dependent Dpex7 germination defect. The complex genetic interactions taking place between these three genes suggest that they have intricate functions in peroxisomes and further studies will be required to clarify how the product of these genes interact in peroxisomes during ascospore germination.
Through the characterization of the GUN mutants, we sought to discover new genes that control dormancy in ascospores. Genetic approaches in P. anserina and, in particular, the amenability to perform epistasis studies makes this model fungus a powerful system for deciphering regulation pathways such as the one controlling ascospore germination. Here, we investigated the relationships between GUN1 and already-known actors of ascospore germination: the PaMpk2 MAPK pathway, the tetraspanin Pls1, and the NADPH oxidase Nox2 complex (4,10,17). We show that gun1 SG requires PaMpk2, PaNox2, and PaPls1 functions to induce the spontaneous germination of ascospores. Furthermore, we show that gun1 SG controls the PaMpk2 pathway by activating PaMpk2 phosphorylation. Altogether, these data demonstrate that on the one hand GUN1 acts upstream of the PaMpk2 pathway and the PaNox2/PaPls1 complex in the regulatory cascade controlling ascospore germination. But our data also show that on the other hand, GUN1 function is required when spontaneous germination is triggered by the constitutive activation of the PaMpk2 pathway (Fig. 8).
Melanin is a major component of appressorium cell wall in M. oryzae where it is involved in turgor pressure generation (26). M. oryzae mutants, such as the Pth2 mutant, showing impairment of melanin synthesis cannot build up the turgor pressure necessary in the appressorium for host penetration and are therefore nonpathogenic. As observed in the Papks1 193 mutant devoid of melanin biosynthesis, melanin in ascospores is important to avoid uncontrolled "accidental" germination (23). The melanization defect exhibited by the Dgun1 ascospores is likely due to the lack of acetate supply to the dihydroxynaphthalene pathway involved in melanin biosynthesis, a defect shared by the DechA and Dfox2 P. anserina mutants, impaired in mitochondrial and in peroxisomal b-oxidation, respectively (23,28). However, unlike DechA and Dfox2 ascospores that germinate spontaneously, Dgun1 ascospores do not, suggesting that the melanization defect in Dgun1 ascospores is not sufficient to induce spontaneous germination. Accordingly, tricyclazole, a fungicide inhibiting melanin biosynthesis or the Papks1 193 mutation blocking melanin production (23), suppress the germination defect of Dgun1 ascospores. Indeed, Papks1 193 Dgun1 ascospores germinate spontaneously (data not shown). Hence, it is likely that in Dgun1 ascospores residual melanin in the cell wall is enough to avoid "accidental" spontaneous germination.
The fact that tricyclazole and Papks1 193 null mutation trigger germination of Dgun1 ascospores suggests that these ascospores are competent for the formation of the germination peg and eventually hyphal growth but that they might be blocked in the formation of the germination pore. The formation of a pore in melanized ascospores is a process sharing similarities with the formation of the pore in M. oryzae appressorium (16). In this pathogenic fungus, the formation of the appressorial pore is preceded by the formation of a septin ring required for actin cytoskeleton remodeling and appressorial pore solidity. In-depth studies in M. oryzae have deciphered the intricate genetic signaling, culminating in the setting up of this septin ring (14,65). Among other regulatory components, the FUS3/PMK1/PaMpk2 pathway and the NoxB (Nox2)/Pls1 complex play a key role in septin ring assembly. The fact that both pathways are also essential for ascospore germination in P. anserina, N. crassa, and S. macrospora, three species producing melanized ascospores, leads us to hypothesize that a similar process involving septin ring assembly and cytoskeleton remodeling may take place to initiate the formation of the germination pore. Given the similarities in the regulation of appressorium functioning and ascospore germination when those are melanized, we speculate that studying and discovering new genes controlling ascospore germination in P. anserina may lead to the discovery of new pathogenesis factors controlling appressorium development in pathogenic fungi. The identification and the characterization of GUN1, the ortholog of M. oryzae Pth2 represents a proof of concept. Indeed, we show here that appressorium development is delayed by 1 day in the hypomorphic gun1 SG mutant and by 2 days in the Dgun1 mutant, demonstrating that, similarly to Pth2, GUN1 is also involved in appressorium functioning in P. anserina. The characterization of the other GUN mutants and, in the future, the isolation of new GUN mutants will be of great interest for better understanding ascospore germination as well as discovering new pathogenic factors, killing two birds with the same stone.

MATERIALS AND METHODS
Strains and culture conditions. The strains used in this study are all listed in Table 8. All of these P. anserina strains derive from the wild-type S strain, ensuring a homogeneous genetic background (66,67). The Papks1 193 mutant for the polyketide synthase encoding gene acting at the first step of melanin synthesis is described elsewhere (23). Standard culture conditions, media, and genetic methods for P. anserina were described previously (68) and can be found in the data of Silar (7) and on the Podospora database (http://podospora.i2bc.paris-saclay.fr/). The compositions of the M0 and M3 media are similar to that of the M2 medium, except that dextrin is replaced by glucose in the M3 medium (5.5 g L 21 ), while no carbon source is added in the M0 medium. This M0 medium was used as a basis for the development of media in which the only carbon source was sodium acetate (60 mM) or oleic acid (Sigma- Aldrich) (6 mM) dissolved in Tween 40 (0.5%). A control medium M0 with only Tween 40 (0.5%) was also used. The germination medium (CH 3 COONH 4 , 4.4 g L 21 ; Bacto peptone, 15 g L 21 ) used in this study was supplemented with yeast extract (G1YE) at 5 g L 21 . In order to allow germination in strains producing ascospores unable to germinate, crosses were set up on M2 medium supplemented, when required, with tricyclazole (1 mg mL 21 ), a fungicide impairing melanin synthesis in P. anserina ascospores (23). Genetic screening of constitutively germinating mutants. It has been shown that wild-type P. anserina ascospores do not germinate on standard M2 medium and that the DPaNox2 and DPaPls1 mutant strains produce ascospores unable to germinate on all tested media (8,10). To isolate mutants producing spontaneously germinating ascospores, UV mutagenesis was performed on self-fertile mat-/mat1, wild-type S, DPaNox2, and DPaPls1 strains. The selection process of the germination mutants is summarized in Fig. S1. The mycelia of the strains mutagenized were fragmented (as described in "Mycelium Fragmentation and Strain Purification" below) and spread on M2 plates at 1,000 CFU. Shortly after UV exposure (UV 254 nm, 250 J/m 2 ), followed by a 1-day culture in the dark to prevent repair by photoreactivation, the mutated strains were grown on standard M2 medium for 1 week until they developed mature ascospore-producing perithecia. In order to recover spontaneously germinating ascospores, M2 medium plates were put on top of the plates bearing perithecia, which allowed ascospores to be harvested in bulk on M2 medium. In P. anserina, most of the progeny is composed of heterokaryotic mat1/mat-ascospores leading to self-fertile mycelium upon germination (42). The thalli produced by the spontaneously germinating ascospores were incubated until they formed mature perithecia projecting their ascospores. Ascospores produced by these perithecia were individually collected with a needle and transplanted onto M2 medium. To ensure their independence, a single mutant (i.e., a single spontaneously germinating homokaryotic ascospore) per initial plate was selected for further analyses. For every mutant, progeny analysis of mutant Â wild-type S crosses showed that a single mutated locus was responsible for the mutant phenotype (i.e., spontaneous germination of ascospores). For mutants recovered with the DPaNox2 and DPaPls1 strains, genetic analyses showed that the mutations enabling germination were unlinked to the DPaNox2 and DPaPls1 mutations, respectively. Homokaryotic matand mat1 mutant strains were isolated from the progenies, and homozygous mutant crosses were performed to check for the fertility/sterility of the mutants. For every isolated strain, microscopic observations were also performed to determine whether the ascospores germinated through the germination pore at the tip of the ascospore or through any other part of the ascospore (data not shown).
Tetrad analysis in the gun1 SG and in the Dgun1 strains. (i) Demonstration of the recessivity of the gun1 SG allele. P. anserina produces mainly asci containing four heterokaryotic/dikaryotic ascospores, allowing nonordered tetrad analysis of first division segregation (FDS) asci and second division segregation (SDS) asci. More detailed explanations on tetrad analysis in P. anserina can be found in Grognet et al. (69). To determine the dominance/recessivity of the gun1 SG allele, we crossed the gun1 SG mutant with the WT. In 30 asci of the progeny, we found that 16 asci contained four heterokaryotic ascospores unable to germinate on M2 medium (4 [nongerminating] ascospores) and 14 asci contained different number of ascospores germinating spontaneously: 3 (nongerminating), 1 (germinating) or 2 (nongerminating), and 2 (germinating) ascospores. Above all, we never observed asci containing more than two ascospores germinating on M2 medium. Assuming some ascospore germination failure (due to their manipulation or incomplete penetrance of the gun1 SG phenotype), especially the ones germinating spontaneously, we concluded that (i) asci of the first type were SDS asci containing four gun1 SG /GUN1 ascospores of the WT phenotype (nongerminating on M2 medium) and (ii) asci of the second type were FDS asci containing two gun1 SG /gun1 SG of the GUN (Germination UNcontrolled phenotype; germinating on M2 medium) and two GUN1/GUN1 ascospores of the WT phenotype. The gun1 SG allele was thus recessive against the wild-type GUN1 allele (i.e., only gun1 SG / gun1 SG ascospores germinated on M2 medium but not the gun1 SG /GUN1 ones).
(iii) Complementation test between the DPa_6_1340::hygR (Dgun1) and the gun1 SG alleles. We crossed gun1 SG with DPa_6_1340::hygR, and we reasoned that if DPa_6_1340::hygR and gun1 SG are allelic, no functional complementation in heterokaryotic ascospores in SDS asci is expected for the spontaneous germination of gun1 SG : DPa_6_1340::hygR/gun1 SG ascospores germinate spontaneously on M2 medium. In contrast, if gun1 SG and DPa_6_1340::hygR are not allelic, the theoretical genetic cross can be written Pa_6_1340::hygR GUN1 Â Pa_6_1340 1 gun1 SG , and several genetic combinations can be generated in SDS and in FDS asci. However, more importantly, none of these combinations can in theory lead to the spontaneous germination of more than two heterokaryotic ascospores. As shown above, concerning the Pa_6_1340::hygR locus, both kinds of asci were obtained in the F 1 progeny: the SDS asci (54%; n = 50) were composed of four (hygR, melanized) ascospores, and the FDS asci were composed of two (melanized, hygS) and two (nongerminating, demelanized, hygR) ascospores. These asci were characterized by sowing them on G1YE and testing them for hygromycin B resistance. Functional complementation was tested by sowing 27 SDS asci on M2 medium supplemented with hygromycin B. Five types of asci were obtained ranging from four germinating ascospores to no germination of the four ascospores. We counted six asci with four germinating ascospores, seven asci with three germinating ascospores, four asci with two germinating ascospores, and seven asci with one germinating ascospore, and three asci showed no germination. Taking into account the partial penetrance of the gun1 SG phenotype, we interpreted that three or four spontaneous germination on M2 medium plus hygromycin B in the same ascus could only occur if the four ascospores in the ascus were of the DPa_6_1340::hygR/gun1 SG genotype. This led us to conclude that functional complementation does not occur in SDS asci, showing that gun1 SG and DPa_6_1340 are allelic. Eventually, this also demonstrated that Pa_6_1340 was the gene mutated in the gun1 SG mutant responsible for the spontaneous germination phenotype. gun1 SG genome sequencing and analysis. In a first step toward identifying the gene mutated in gun1 SG through whole-genome sequencing, we backcrossed the mutant for five generations with the parental wild-type S strain to eliminate any mutation unrelated to the mutant phenotype. The gun1 SG genomic DNA was extracted as described previously (70). The genomic DNA was then subjected to complete sequencing using Illumina technology at the Imagif facility, Gif-sur-Yvette, France (CNRS, I2BC Sequencing Facility, https://www.i2bc.paris-saclay.fr/sequencing/). Custom-made libraries had 300bp inserts, and sequencing was 76-bp paired end. Coverage was 80-fold. The sequence reads were then mapped onto the latest version of the reference genome of the S strain (69). Potential mutations were detected using SAMtools and bcftools on the Galaxy web server (https://usegalaxy.org/).
Deletion of GUN1 and GUP1 and construction of the Dgun1 Dgup1 strain. (i) Dgun1 strain. The deletion of Pa_6_1340/GUN1 and its paralog Pa_3_7660/GUP1 was performed using deletion cassettes made of two overlapping PCR fragments (see Fig. S4 and S6) (17). This method is based on the generation of two DNA PCR fragments carrying a resistance marker flanked by either 59 or 39 flanking sequences of the targeted gene. For Pa_6_1340/GUN1, we first amplified the 803-bp 59 and 486-bp 39 flanking regions of the S strain DNA by PCR with the primer pairs 1340_1/1340_2 and 1340_3/1340_4, respectively (see Table S2). At the same time, the hygromycin B resistance marker was amplified with 1340_MkF and 1340_MkR (see Table S2) from the pBC-hygR vector (71). In a second PCR round, using the primers 1340_1 and 1340-MkR and the primers 1340-MkF and 1340_4, the resistance marker was fused with the 59 and 39 flanking regions, respectively. Both PCR products were used to transform a Dmus51::phleoR strain, in which the mus51 gene involved in the NHEJ repair system is replaced by a phleomycin resistance gene (phleoR), allowing a high rate of homologous recombination (50). Two hygromycin B-resistant (hygR) transformants were obtained. Each one was crossed with the wild-type S strain. We observed in the progeny that homokaryotic (hygR) ascospores did not germinate. Consequently, crosses were performed on M2 medium supplemented with tricyclazole, leading to the spontaneous germination of ascospores (23). The (hygR) thalli coming from spontaneously germinating ascospores were selected on M2 medium supplemented with hygromycin B, fragmented and (hygR, phleoS) DPa_6_1340::hygR (Dgun1) homokaryotic mycelia of each mating type were isolated. Deletion of Pa_6_1340 was verified by Southern blotting (see Fig. S4). Only one strain was selected for further analyses.
(ii) Dgup1 strain. The same protocol was performed to produce the deletion cassettes for Pa_3_7660/GUP1. Using the primers pairs: 7660_F1/7660_R2 and 7660_R3/7660_R4 (see Table S2), the 1,104-bp 59 and 1,011-bp 39 Pa_3_7660 flanking regions were PCR amplified, while the phleomycin resistance marker was amplified with the primers 7660_MkF and 7660_MkR (see Table S2) from a pBC-phleoR plasmid (71). In a second PCR round, using the primers 7660_F1 and 7660_MkR and the primers 7660_MkF and 7660_R4, the resistance marker was fused with the 59 and 39 flanking regions. Both PCR products were used to transform a Dmus51::genR strain. A total of 26 phleomycin-resistant (phleoR) transformants were obtained, and two independent (phleoR, genS) Dgup1::phleoR strains were selected from the progeny of a cross with the wild-type S strain (germination of Dgup1::phleoR ascospores was as the wild type). Deletion in these two independent strains was verified by Southern blotting (see Fig. S6). Only one strain was selected for further analyses.
Plasmid constructions for complementation of gun1 SG , Dgun1, and Dgup1 strains. (i) Construction of pGUN1. The Pa_6_1340/GUN1 CDS, its 803-bp 59 upstream and 486-bp 39 downstream sequences were amplified by PCR from wild-type S genomic DNA using 1340_1 and 1340_4 primers (see Table S2). The PCR product obtained was cloned blunt end into pBC-genR plasmid carrying a Geneticin resistance marker digested by EcoRV to produce the pBC-GUN1-genR plasmid (renamed pGUN1 for the sake of simplicity). The insert was verified by sequencing (data not shown). This plasmid was used to transform the Dgun1::hygR deletion strain. Two (genR) transformants were obtained and checked for the restoration of wild-type phenotypes in ascospores. To that end, two of these transformants were crossed with the Dgun1::hygR strain. In the progeny of both crosses, (hygR, genR) melanized homokaryotic ascospores germinated (on G1YE germination medium), allowing us to purify Dgun1 pGUN1 mat1 and mat-homokaryotic strains and to show that wild-type GUN1 complemented the Dgun1 mutation. The (hygS, genR) GUN1 pGUN1 homokaryotic ascospores were also isolated in the progeny. These ascospores germinated as the wild type. The pGUN1 was also used to transform the gun1 SG mutant. Three (genR) transformants were obtained and crossed with gun1 SG to assess restoration of wild-type germination in the progeny. For one transformant, we observed that gun1 SG pGUN1 progeny showed wild-type ascospore germination, i.e., gun1 SG pGUN1 spores did not germinate spontaneously on M2 medium but germinated on G1YE germination medium, showing that ectopic wild-type GUN1 complemented the gun1 SG mutant ( Table 2).
(ii) Construction of pGUP1. The Pa_3_7660/GUP1 CDS, its 1,104-bp 59 upstream and 1,011-bp 39 downstream sequences were amplified by PCR from wild-type S genomic DNA using the primers 7660_F1 and 7660_R4 (see Table S2). The PCR product obtained was cloned blunt-end into pBC-nouR (carrying a nourseothricin resistance marker) digested by EcoRV to produce the pBC-GUP1-nouR plasmid (renamed pGUP1). The pGUP1 plasmid was used to transform the Dgup1::phleoR deletion strain. A total of 17 (nouR) transformants were obtained and checked for the restoration of growth on acetate (ace 1). Fourteen of them were (ace 1), showing that wild-type GUP1 complemented Dgup1 mutation. Two of these complemented transformants were crossed with the wild-type S strain and (phleoR nouR) Dgup1 pGUP1, as well as (phleoS nouR) Dgup1 pGUP1 homokaryotic mat1 and mat-strains were purified.
Plasmid construction for GUN1-and gun1 SG -mCherry/mCherry-AKI tagging. Two kinds of tagging were undertaken: one with the mCherry-AKI reporter protein, carrying the AKI PTS1 peroxisome targeting signal present in C terminus of GUN1, added in C terminus of the mCherry, and a second with the standard mCherry without the AKI PTS1 signal. To this end, we constructed plasmids allowing integration of the mCherry-AKI CDS or the mCherry CDS in 39 (and in frame) of GUN1 or of gun1 SG CDS at the endogenous GUN1 locus by homologous recombination (see Fig. S8). To achieve this, the 621-bp region upstream of the stop codon (but downstream of the mutation present in the gun1 SG allele) was PCR amplified with primers 1340GFP_F2 and 1340GFP_R1 (see Table S2) designed to incorporate the ApaI and XhoI restriction sites in the sequence, respectively. The PCR product was cloned blunt end into the pBC-genR plasmid previously digested with EcoRV. The insert was then sequenced, digested with the restriction enzymes ApaI and XhoI, and gel purified to be finally cloned upstream of the mCherry CDS into the pBC-mCherry-hygR plasmid digested by ApaI and XhoI (see Table S2). This pBC-GUN1-mCherry-hygR plasmid (renamed pGUN1-mCherry for the sake of simplicity) was sequenced and transformed into both Dmus52::genR and gun1 SG Dmus52::genR, and (hygR) transformants were selected. We obtained 1 (hygR) transformant for Dmus52::genR and 2 (hygR) transformants for gun1 SG Dmus52::genR. Every transformant showed red fluorescence under the microscope. Correct GUN1-mCherry and gun1 SG -mCherry gene fusions were verified by sequencing (data not shown). One transformant of each genotype was selected and crossed with the S strain to purify (hygR, genS) GUN1-mCherry and gun1 SG -mCherry homokaryotic mat-and mat1 strains in the progeny. Finally, we observed that GUN1-mCherry ascospores germinated as the wild-type and that gun1 SG -mCherry ascospores germinated spontaneously on M2 medium. To construct the pBC-GUN1-mCherry-AKI-nouR plasmid, we amplified by PCR the insert present in pGUN1-mCherry with the primers 1340GFP_F2 and mCH_AKIR1, the latter primer allowing addition of the AKI coding sequence at the end of the mCherry (see Fig. S8). This 1,341-bp PCR fragment was cloned blunt end into pBC-nouR digested with EcoRV to give the pBC-GUN1-mCherry-AKI-nouR plasmid (renamed pGUN1-mCherry-AKI). The construction was sequenced and transformed into the Dmus52:: genR strain. We selected one (nouR) transformant showing red fluorescence under the microscope, and we crossed it with the wild-type S strain to purify (genS, nouR) GUN1-mCHerry-AKI mat1 and matstrains. This GUN1-mCHerry-AKI strain showed red fluorescence under a microscope, and correct GUN1-mCHerry-AKI gene fusion was verified by sequencing (data not shown). This GUN1-mCherry-AKI strain was then crossed with the gun1 SG mutant to generate the gun1 SG -mCherry-AKI strain by meiotic recombination between the gun1 SG mutation and the mCherry-AKI-nouR insertion. In the progeny of this cross, we isolated 13 thalli from spontaneously germinating (nouR) ascospores on M2 medium supplemented with nourseothricin. All of these isolates were heterokaryotic self-fertile (mat1/mat-). Since the gun1 SG mutation is recessive, we hypothesized that these isolates arose from gun1 SG -mCherry-AKI/gun1 SG -mCherry-AKI recombinant heterokaryotic ascospores. One of these isolates was fragmented, and (nouR) homokaryotic gun1 SG -mCherry-AKI ascospores of each mating type were purified establishing the gun1 SG -mCherry-AKI strain. Red fluorescence in this final strain was verified, and the presence of the gun1 SG mutation, as well as correct mCherry-AKI integration, was verified by sequencing (data not shown). Germination of gun1 SG -mCherry-AKI ascospores was spontaneous on M2 medium as for the gun1 SG mutant.
Mycelium fragmentation and strain purification. For strains carrying the Dgun1 deletion and which therefore cannot germinate, homozygous crosses were performed on M2 medium supplemented with tricyclazole (1 mg mL 21 ), leading to the spontaneous germination of ascospores. The purification of homokaryotic strains was performed through mycelium fragmentation as follows. A small implant of the thallus of interest (0.5 cm 2 ) was set in a 2-mL tube containing 500 mL of H 2 O and ground using a FastPrep (TeSeE; Bio-Rad, Hercules, CA) for 20 s at 5,000 rpm. Then, 100 mL of the fragmented mycelium was spread on an agar plate, and small hyphal fragments were isolated using a stereomicroscope (magnification, Â40). These isolates were then placed on M2 medium and cultured for 2 days at 27°C. Homokaryotic (mat1 or mat-) versus self-fertile heterokaryotic (mat1/mat-) genotypes of isolates were determined through a mat-type test using wild-type S mat1 and mat-strains.
Fertility assay. Fertility was assayed by generating self-fertile mat1/mat-heterokaryons of the tested strains. To this end, implants (0.5 cm 2 ) of each mating type of the strains of interest were placed in 2-mL tubes containing 500 mL of H 2 O and ground using a FastPrep (TeSeE) for 20 s at 5,000 rpm. Next, 10 mL of each heterokaryon was dropped onto M2 medium and cultured for 10 days at 27°C with light. A qualitative evaluation of perithecium formation directly on the plates and of ascospore production projected on the petri plate lids during 4 days was performed. Every heterokaryon were analyzed in duplicate.
Quantification of ascospore production. At day 9 (ascospore projection starts at day 7), for the homozygous crosses-WT Â WT (WT), Dgun1 Â Dgun1 (Dgun1), and Dgun1 pGUN1 Â Dgun1 pGUN1 (Dgun1 pGUN1)-ascospores projected for 30 min on circle surfaces 2 cm in diameter were counted under a stereomicroscope. Eleven measures were made for each cross (n = 11). A Student statistical t test was performed to compare the three measures.
Cellophane penetration assay. An implant of each tested strain was placed on a cellophane layer (Bio-Rad). After 2, 3, 4, and 5 days of growth at 27°C, the cellophane layer was removed, and the presence of mycelium in the medium checked with a stereomicroscope to determine whether the strain had breached the cellophane layer. In parallel, the presence or absence and the morphology of appressoria in the cellophane layer were observed under a microscope as described previously (9, 72) (see Fig. S3).
Microscopic observations. Ascospores observations were performed in Ibidi eight-well chamber microslides (Gräfelfing, Germany). Each well was filled with 200 mL of either M2 or modified G liquid medium: the quantity of Bacto peptone was divided by 2 compared to standard G medium in order to reduce the fluorescing background noise due to Bacto peptone. Germination induction of this modified medium was not altered (data not shown). Crosses were performed on standard M2 medium. Once perithecia were mature, agar plugs bearing the perithecia were cut and placed above the microscopic chambers upside-down. Perithecia were left to project their ascospores into the well for 5 h. For mycelium observations, small squares of medium (1 cm 2 ) with grown mycelium on them were cut at the edge of the thallus and placed upside-down in water on the coverslip of a microscopic chamber. Images were taken with an inverted microscope Zeiss spinning disk CSU-X1 (Oberkochen, Germany) using four lasers (405, 488, 561, and 640 nm) for fluorescence observations, their associated filters, and a sCMOS PRIME95 (Photometrics) camera at the Imagoseine Imaging Facility (https://www.ijm.fr/plateformes-et-services/ plateformes/imagoseine/). The images were analyzed with Fiji (73).
Colocalization quantification. For every genotype analyzed, the Pearson's correlation coefficient (PCC) was calculated in the mycelium and in ascospores using the Coloc 2 tool of the Fiji software. PCC estimates colocalization (codistribution) of both fluorescent signals (green and red). PCC values range from 1 for two images whose fluorescence signals are perfectly linearly related to 21 for two images whose fluorescence signals are perfectly, but inversely, related to one another, with intermediate values indicating partial colocalization. Values near zero reflect distributions of signals that are uncorrelated with one another (60). For mycelium analysis, from four to seven different hyphae (of different images) were analyzed. In hyphae, PCCs have been measured on ;20-mm regions of interest (ROIs) on one focal plan for a total number of measures ranging from 12 to 25 per genotype. The numbers of analyzed ROIs for each genotype are indicated in Fig. 5C. A Student t test was applied to compare the PCCs of the different genotypes analyzed. For ascospore analyses, the whole surface of one focal plan per ascospore was divided in several ROIs. The sizes and numbers (ranging from four to eight per ascospore) of ROIs were designed depending on the presence or not of lipid droplets, so that the latter could be excluded. For the mCherry-AKI tagging experiment (Fig. 6), the number of analyzed ascospores ranged from four to seven for a total number of PCC measures ranging from 21 to 34 per genotype and condition. These data are indicated in Fig. 6C. In order to account for the within-unit correlation induced by repeated observations on the same ascospores, a mixed-effect linear model was used to assess the statistical significance of the medium and strain on quantified values. A total of 21 statistical units were considered in the GFP-SKL experiment, and 19 statistical units in the case of mito-GFP. Post hoc pairwise comparisons were performed using a Bonferroni-corrected t test with the Satterwaithe approximation (unequal variances), considering a family-wise error rate of 5%. For the mCherry tagging experiment (see Fig. S9), the number of analyzed ascospores ranged from two to three for a total number of analyzed ROIs ranging from 10 to 15. A Student t test was used to assess the pooled values per genotype to compare the PCCs.
Phylogenetic analysis. Fungal genes homologous to GUN1 were searched by using BLAST at the GenBank and MycoCosm databases (74,75), using the default parameters with the GUN1 protein sequence as a query. For a selection of Ascomycota, Basidiomycota, and Mucoromycota species, hits with an E value lower than 10 25 were selected. Research for other homologs was carried out for each selected species on OrthoDB (51). The alignment was performed with MAFFT (76) and manually refined using Jalview (77). A phylogenetic tree was built using the maximum-likelihood method (PhyML 3.1 software using the default parameters) (78). The tree was visualized on the iTOL server (79). Bootstrap values of 100 replicates are indicated (see Fig. S5). The GUN1 I441 residue conservation was assessed by visualizing the alignment on Jalview (see Fig. S7) (77).
CAT activity assay. (i) Mycelium protein extraction. Mycelia growing on M2 medium (2 confluent plates per strain) were harvested after 2 days; placed in 2-mL tubes, each containing a tungsten bead (diameter, 3 mm) and 1 mL of potassium phosphate buffer (pH 7.4) supplemented with 2 mM EDTA; and ground in a TissueLyser II apparatus (Qiagen, Hilden, Germany) at 30 rpm/s for 4 min at 4°C. The lysate was centrifuged at 4°C for 20 min at 17,000 Â g, and the supernatant was separated from the pellet (cell debris). The protein concentration in the supernatant was measured by the spectrophotometric Bradford method (Sigma Chemical Co., St. Louis, MO).
(ii) Ascospore protein extraction. Each strain was crossed (in homozygous crossing) on at least five M2 medium plates. When perithecia were mature, the projected ascospores were collected on agar plates topped with a cellophane layer and pulled together in a 2-mL tube for each strain, each containing a tungsten bead (diameter, 3 mm). Immediately after harvesting, each tube was flash frozen in liquid nitrogen. Ascospores were dry-crushed in a TissueLyser II apparatus (Qiagen) at 30 rpm/s for 4 min at 280°C. Ascospores were maintained at -80°C in precooled TissueLyser blocks. The crushed ascospores were resuspended in 200 mL of potassium phosphate buffer (pH 7.4) supplemented with 2 mM EDTA. The lysate was centrifuged at 4°C for 20 min at 17,000 Â g, and the supernatant was separated from the pellet (cell debris). The protein concentration in the supernatant was estimated by the spectrophotometric Bradford method (Sigma Chemical Co.).
(iii) Carnitine-acetyltransferase assay. CAT activity was assayed as described previously (88). The reaction was monitored spectrophotometrically at room temperature by following the release of CoA-SH from acetyl-CoA using the thiol reagent 5,59-dithiobis-nitrobenzoic acid (DTNB; Sigma Chemical Co.). The reaction mixture contained 100 mM Tris-HCl buffer (pH 7.8), 0.05 mM acetyl-CoA (Sigma Chemical Co.), 0.1 mM DTNB, 22 mM DL-carnitine chloride (Sigma Chemical Co.), and protein extract in final volume of 1.0 mL. The reaction was initiated by adding a volume of the protein extract, and the increase in absorbance was monitored at 412 nm. CAT activities were determined by measuring the initial velocity of the CoA-SH production reaction and then reported as the activity ratio of the wild-type S strain. Standard deviations and statistical analyses were calculated from four to seven biological replicates. Eventually, CAT activities were compared using a Fisher-Pitman permutation test.
Western blot analysis. PaMpk2 phosphorylation was assessed as described previously (17). Ascospores produced by homozygous crosses of the S strain, the PaMKK2 c mutant and the gun1 SG mutant were harvested on agar plates topped with a cellophane layer to facilitate the ascospore harvesting process. For germination induction, a cellophane layer with wild-type ascospores was transferred on G1YE medium for 2 h. Once collected, ascospores were flash frozen in liquid nitrogen and then dry disrupted in a Micro-Dismembrator (Sartorius) at 2,600 rpm for 1 min at 280°C. Crushed ascospores were resuspended in Laemmli buffer, placed 5 min at 100°C, and then centrifuged for 15 min at 14,000 rpm. Samples were placed on a 1-mm-thick 15% SDS-PAGE gel and migrated for 3 h at 130 V, 25 mA/gel, and 25 W. The gel was then transferred onto a polyvinylidene difluoride membrane. Hybridization with the anti p44/p42 or anti-phospho p44/42 antibodies (Cell Signaling Technology), diluted to 1/1,000, was carried out overnight at 4°C. Hybridization with the second antibody coupled to peroxidase (GE Healthcare), diluted to 1/1,000, was carried out for 1 h at room temperature. For the revelation, an Immobilon chemiluminescence kit (Millipore) was used according to the supplier's recommendations. A relative quantification of the ratio P-PaMk2/PaMk2 has been done using Fiji software. The mean pixel intensity of a defined ROI, including the band of interest, was measured for P-PaMpk2 and PaMpk2 signals, and the background intensity mean was subtracted. The ratio P-PaMk2/PaMk2 was then normalized with the "wild-type noninduced" ratio (ratio = 1) to give the induction fold compared to this reference condition (Fig. 7).

SUPPLEMENTAL MATERIAL
Supplemental material is available online only. SUPPLEMENTAL FILE 1, AVI file, 0.6 MB. SUPPLEMENTAL FILE 2, PDF file, 6.6 MB.