Re-routing MAP kinase signaling for penetration peg formation in predator yeasts

Saccharomycopsis yeasts are natural organic sulfur auxotrophs due to lack of genes required for the uptake and assimilation of sulfate/sulfite. Starvation for methionine induces a shift to a predatory, mycoparasitic life strategy that is unique amongst ascomycetous yeasts. Similar to fungal plant pathogens that separated from Saccharomycopsis more than 400 million years ago, a specialized infection structure called penetration peg is used for prey cell invasion. Penetration pegs are highly enriched with chitin. Here we demonstrate that an ancient and conserved MAP kinase signaling pathway regulates penetration peg formation and successful predation in the predator yeast S. schoenii. Deletion of the MAP kinase gene SsKIL1, a homolog of the Saccharomyces cerevisiae ScKSS1/ScFUS3 and the rice blast Magnaporthe oryzae MoPMK1 genes, as well as deletion of the transcription factor SsSTE12 generate non-pathogenic mutants that fail to form penetration pegs. Comparative global transcriptome analyses using RNAseq indicate loss of the SsKil1-SsSte12-dependent predation response in the mutant strains, while a methionine starvation response is still executed. Within the promoter sequences of genes upregulated during predation we identified a cis-regulatory element similar to the ScSte12 pheromone response element. Our results indicate that, re-routing MAP-kinase signaling by re-wiring Ste12 transcriptional control towards predation specific genes contributed to the parallel evolution of this predacious behaviour in predator yeasts. Consequently, we found that SsSTE12 is dispensable for mating.


Introduction
A general feature of fungal pathogens is the development of specialized infection structures to force entry into host tissue [1].These include hyphopodia e.g. in Verticillium dahliae or in arbuscular mycorrhizal fungi, infection cushions of Botrytis cinerea and Fusarium graminearum or appressoria, e.g. in Magnaporthe oryzae and Ustilago maydis [2][3][4][5][6][7][8].Appressoria are adhesion structures from which penetration pegs emerge to enter the host tissue [9].Appressoria can be hyaline (Botrytis) or melanized as in Colletotrichum and Magnaporthe [8,10].Invasion, supported by the synthesis of polymer-degrading enzymes such as cutinases, is achieved by high pressure within melanized appressoria of M. oryzae to force penetration pegs through the plant cuticle [11,12].Although the evolutionary origin of appressoria is unknown, molecular details have been obtained by the study of Magnaporthe and the basidiomycetous corn smut fungus Ustilago maydis [10,[12][13][14].In both systems MAP kinase signaling plays a central role in appressorium development and subsequent penetration of host tissue.In Magnaporthe the S. cerevisiae ScKSS1/ScFUS3 homolog MoPMK1 regulates appressorium development.Deletion mutants of MoPMK1 or mutants in which MoPmk1 can be chemically inactivated do not form appressoria [15,16].One of the MoPmk1 targets, the S. cerevisiae ScSTE12 homolog MoMst12, is also required for virulence since mst12 mutants form appressoria but fail to generate penetration pegs [17,18].In U. maydis the MAP kinases Umkpp2 and Umkpp6 are involved in appressorium development; Umkpp6 mutants can form appressoria, but are defective in penetration [19,20].ScSTE12 homologs have been characterized for their role in virulence in several pathogenic fungi [21][22][23].
Saccharomycopsis deviates from the standard nuclear code by translating the CUG codon as serine instead of leucine; these yeasts are therefore closely related to another CTG clade containing e.g.Candida albicans, and to the genus Ascoidea [24][25][26].Another trait of Saccharomycopsis yeasts is their natural methionine auxotrophy due to the loss of several genes for uptake and assimilation of sulfate [27].Nevertheless, the predator yeast S. schoenii appears to be rather successful as it attained a global distribution [28].A striking and unique feature of Saccharomycopsis yeasts within the Saccharomycetales is their necrotrophic mycoparasitism, i.e. the ability of Saccharomycopsis to attack and kill fungal prey cells [28][29][30].Interestingly, the killing behavior may not rely on secreted toxins termed 'killer toxins' that are a widespread amongst budding yeasts [31,32].Rather, killing occurs via specialized structures called penetration pegs.The predacious behavior is instigated by starvation, e.g. by removal of methionine from culture media [26,30].
There is a conspicuous resemblance between penetration pegs generated by appressoria of plant pathogenic fungi and those developed by Saccharomycopsis despite their large evolutionary distance.This led us to characterize the MAP kinase homolog of ScFUS3/ScKSS1/MoPMK1 in S. schoenii.We demonstrate that this MAP kinase is essential for penetration peg formation

PLOS PATHOGENS
Penetration peg formation in predator yeasts and virulence and named the gene SsKIL1, mnemonic for killer kinase 1.Similarly, a potential downstream target of SsKil1, the helix-turn-helix transcription factor of the homeodomain protein family SsSte12, is required for penetration peg formation.Transcriptome analyses allowed us to separate a methionine starvation induced hunger response from the predation response that is regulated by SsKIL1 and SsSTE12.Three cis-regulatory elements were found in differentially expressed genes (DEGs).Our data suggests that within the genus Saccharomycopsis parallel evolution resulted in the re-routing of MAP kinase signaling by rewiring Ste12-output to service predation peg formation and the predation response.

Penetration pegs can be visualized by fluorescence microscopy
It has previously been reported that S. schoenii turns to a predatory lifestyle upon starvation and in the presence of a prey [26], but so far, a more detailed molecular analysis of penetration peg formation has not been reported.S. schoenii cells are elongated while S. cerevisiae cells are ovoid, which allows a clear distinction even of unmarked predator and prey cells.Penetration of a prey cell requires removal of its cell wall.To study predator-prey interactions, we used the fluorescent dye calcofluor white (CW), which binds to chitin in the yeast cell wall (Fig 1).In S. cerevisiae chitin rings at bud scars are brightly stained with CW. Cell wall staining with CW is also brighter in S. cerevisiae cells compared to S. schoenii cells.However, predator yeast cells attacking S. cerevisiae prey cells exhibit strong CW fluorescence of the penetration pegs (Fig 1A).This signal can be used to track predation events and the wild type S. schoenii strain proved to be a very effective predator yeast against S. cerevisiae.Histone H4-GFP-tagged prey cells lost their nuclear GFP signal after an attack by predator yeast cells and after penetration pegs had entered these cells, indicating their demise (Fig 1A).In Fig 1A there is only one S. cerevisiae cell that exhibits the nuclear H4-GFP fluorescence, indicating that only this cell is alive while all other prey cells were killed.Predator yeast cells were capable of producing several penetration pegs in their life time and single prey cells could be attacked simultaneously by several predator yeasts (Fig 1A -1C).Prey cells which were not under attack and which were not marked by penetration pegs remained viable.Interestingly, the base of a penetration peg is stained more broadly and the tip of the penetration peg inside the yeast cell sometimes appears to be balloon-shaped with an increased surface area (Fig 1D) [29].Besides CW, which stains glycosidic bonds, we employed the fluorescent lectin conjugates wheat germ agglutinin (WGA) and concanavalin A (conA) to study penetration peg labelling.Using WGA, which specifically binds to N-acetyl-D-glucosamine found in chitin, only the base of the penetration peg could be labelled but not the region localized within the prey cell (Fig 1E    monitoring predation via time-lapse fluorescence microscopy (n = 29) provided insights into the coordination of both processes.(i) Polarized growth regions of daughter cells were used to generate penetration pegs (24 out of 29) suggesting that polarized growth at the daughter cell can be altered into penetration peg morphogenesis.(ii) Daughter cell growth was halted during predation and resumed at its previous position afterwards indicating maintenance of polarity at the site of bud growth during a predation event (n = 21).(iii) Concomitant with growth arrest of the daughter cell during predation the mother nucleus appeared to be blocked from entry into mitosis suggesting a cell cycle block during predation (n = 9).(iv) This cell cycle arrest clearly indicated that the penetration peg did not receive a nucleus during the predation cycle (n = 29).In fact, we never observed the presence of a nucleus in a penetration peg.This is in contrast to penetration pegs of Magnaporthe that further develop into infection hyphae [33].In Saccharomycopsis penetration pegs were only used in a predation cycle and once formed represent a terminal phenotype, i.e. no outgrowth occurred and penetration pegs did not develop into daughter cells.A new predation cycle inevitably required the targeted development of a new penetration peg.(v) Visible hallmarks of a predacious attack were the onset of penetration peg emergence and the diminishing/disappearance of the nuclear GFP- signal in the prey.A predation event could be as fast as 16 minutes.(vi) Killing of prey mother cells also resulted in killing of their connected daughter cells even after mitosis.(vii) Prey cell death was accompanied by shrinking of the prey cell, indicating transfer of nutrients into the predator cell as was observed previously [26].

MAP kinase signaling regulates penetration peg formation
MAP kinase homologs of the S. cerevisiae FUS3/KSS1 genes have been shown to be essential for fungal development and pathogenicity in a number of filamentous ascomycetes [34,35].To determine whether a Saccharomycopsis FUS3/KSS1-MAP kinase homolog is involved in the predation process we first examined several Saccharomycopsis genomes in order to identify MAP kinase genes.This indicated that Saccharomycopsis species harbor only a single homolog of the FUS3/KSS1 MAP kinases similar to M. oryzae, whereas the more closely related yeast species Candida albicans and Wickerhamomyces anomalus possess two and in the case of Ascoidea rubescens three FUS3/KSS1 homologs (S1 Fig) .The complete open reading frame of the single FUS3/KSS1 homolog in S. schoenii was deleted via gene replacement cassettes with ~1.5 kb flanking homology regions and two independent mutants were verified by diagnostic PCR (S2 Fig).These mutants did not show any growth defects in standard media.However, when tested in predation assays against H4-GFPtagged S. cerevisiae prey cells, the S. schoenii mutants were unable to kill their prey and live prey cells were identified by nuclear GFP-fluorescence (Fig 3A).Mutant Sskil1 predator cells did not form any highly fluorescent CW-labelled penetration pegs and prey cell invasion could not be observed.Based on this non-virulent phenotype and the decisive role of this gene in the killing of prey cells we named this MAP kinase gene SsKIL1 for S. schoenii killer kinase 1.To quantify predation in wild type and Sskil1 mutants we analyzed cell-to-cell predator-prey interactions in situ via fluorescence microscopy.Different from the previous assay, predator and prey cells were incubated on solid medium overnight and were then stained by CW in situ and directly observed.Only cells in close contact were analyzed and the dead/alive ratio of prey cells was determined.This indicated that the wild type was a highly efficient predator on S. cerevisiae prey cells and almost all interactions of starved predator cells with S. cerevisiae cells ended in the penetration and death of the prey cells.In contrast Sskil1 strains were completely non-pathogenic.Upon reintroduction of a wild type SsKIL1 gene in a Sskil1 strain, predation could be restored to wild type levels (Fig 3B and 3C and S4 Table).
In a reverse assay, a plate-based assay was used to analyze the potential of prey cells to establish colonies against a lawn of predator cells.Predator yeasts, which were themselves unable to form colonies on these minimal medium plates based on their methionine auxotrophy were, however, still able to attack prey cells.Against a lawn of S. schoenii wild type cells only very few S. cerevisiae colonies appeared whereas on a lawn of Sskil1 cells a large number of prey colonies were established indicating lack of predation.In the complemented strain virulence was reestablished demonstrating that the defect in predation was caused by SsKIL1 deletion (Fig 3D and 3E and S4 Table ).

Sensing of the prey and positioning of the penetration peg
Starvation is a prerequisite of predation, but the factors that regulate positioning of a penetration peg are unknown.Time lapse microscopy indicated that the tip of a growing daughter cell can be utilized to develop a penetration peg (Fig 2 and S1 Movie), however, penetration pegs were also initiated from central regions of a predator yeast cell (S2 Movie).We found directional polarized growth targeted at prey cells in the S. schoenii wild type (Fig 4A and S3 Movie).Interestingly, such a prey-directed response was also observed in Sskil1 strains suggesting that this thigmotrophic process is upstream and independent of SsKIL1 (Fig 4B and S4 Movie).Sensing in Sskil1 strains, was however, unproductive, did not harm the prey cells and was finally interrupted by cell division.

S. schoenii Ssste12 mutants are avirulent and do not form penetration pegs
In S. cerevisiae, ScSTE12 is a key downstream transcription factor that is regulated by the ScFus3/ScKss1 MAP kinases and controls mating and filamentous growth [36].The S. schoenii SsSTE12 homolog was identified based on its high conservation to ScSte12, which is however, limited to the N-terminal region (amino acids 83-256, carrying the Ste-like transcription factor domain pfam02200 with the putative DNA-binding domain).Two S. schoenii Ssste12 deletion mutants were obtained and verified (S3 Fig) .Ssste12 mutants showed no growth defect compared to the wild type.However, when assayed for predation activity Ssste12 mutant strains were unable to generate penetration pegs and kill prey cells (Fig 3A and S5 Movie).In situ analyses of predator-prey cell-to-cell interactions, similar to Sskil1 strains, showed a complete loss of virulence in Ssste12 mutants (Fig 3B and 3C).The S. schoenii wild type is homothallic.However, the strain we used in this study is a poor sporulator.Therefore, we used marker-assisted breeding to compare mating efficiency of Sskil1 and Ssste12 strains to the wild type.This indicated that wild type and Ssste12 strains generate sexual progeny at a similar level.However, Sskil1 strains, even when mated to a wild type strain, are severely affected in their sexual reproductive capacity (Fig 5 and S5 Table).

Differential expression analysis identifies loss of the predation response in Sskil1 and Ssste12 mutants
Previously, proteomics and transcriptomics were used to identify and quantify the S. schoenii wild type responses during starvation and predation [26].This defined a starvation and a predation response, which resulted in the identification of enriched GO terms for sulfur metabolic processes and transmembrane transport activities in the starvation response and for GO term categories relating to cell wall organization and hydrolysis in the predation response [26].The highly similar non-pathogenic phenotypes of Sskil1 and Ssste12 strains prompted us to determine the global transcriptional responses of these strains under predation conditions (i.e. on minimal medium lacking methionine) and compare these with each other and to the wild type using a similar transcriptome analysis approach.Differentially expressed genes (DEGs) were called with a p value adjusted for a false discovery rate p adj < 0.05 and a log 2 -fold change in the expression of genes of log 2 � 2 to identify upregulated genes (S5 and S6 Tables).This identified 151 DEGs in Sskil1 and 133 DEGs in Ssste12 mutant, which we compared to a set of 228 DEGs in the wild type (Fig 6A ).A core set of 64 genes was upregulated in all strains.This set encompassed genes involved in sulfur amino acid transmembrane transport with two SsYCT1 genes (encoding cysteine transporters), SsMUP1 (a high-affinity methionine permease), the five-member gene family SsSEO1encoding putative sulfur compound permeases and genes involved in sulfur metabolism with SsMET32 (Zinc-finger DNA-binding transcription factor), SsSTR3 (Peroxisomal cystathionine beta-lyase), SsLAP3 (Cysteine aminopeptidase), SsCYS3 (Cystathionine gamma-lyase), SsMET30 (F-box protein, transcriptional regulator), two SsMET2 genes (L-homoserine-O-acetyltransferases), SsMET17 (O-acetyl homoserine-O-acetyl serine sulfhydrylase) as well as genes involved in pyridoxal (vitamin B 6 )-binding, CYS3, MET17 and STR3, which require vitamin B 6 as co-factor in S. cerevisiae (Fig 6A and S6 Table).S. schoenii is a pyridoxal-5'-phosphate auxotroph lacking homologs of the S. cerevisiae SNO/ SNZ genes.In line with this observation, we found that SsTPN1, encoding a vitamin B 6 -transporter, was upregulated in Sskil1 and mildly upregulated in wild type and Ssste12 (S5 Table ).As we used minimal medium lacking amino acids for our predation RNAseq approach this hunger response towards organic sulfur deprivation validated our analyses and underlines the specific need of the methionine auxotrophic predator yeasts.Not only genes for the uptake of sulphur compounds were upregulated during the hunger response but also the expression of genes involved in nitrogen uptake was increased, particularly SsUGA4 (a gamma- We predicted enriched motifs in the promoter sequences of this commonly upregulated gene set and found two enriched motifs that bear resemblance to the S. cerevisiae DNA-binding motifs of the Met31/Met32 and Cbf1 transcriptional regulators (Fig 6B and S8 Table).From this data we conclude that the methionine starvation response executed in all strains is independent of either SsKil1 or SsSte12.
The largest set of DEGs was associated with genes that were only upregulated during predation in the wild type but neither in Sskil1 nor in Ssste12.This set encompassed 147 genes that describe the predation response (Fig 6C).These DEGs were significantly enriched with genes involved in cell wall organization and biogenesis, particularly with hydrolytic activities such as chitinases (8x SsCTS1), glucanases (SsDSE4, 2x SsSCW10 and SsSPR1) and aspartic proteases of the yapsin family (11x SsYPS3), but also three phospholipase genes (2x SsPLB1 and SsPLB3) (Fig 6C and S6 Table).These GO terms were already previously identified as specifically upregulated in the wild type only in the presence of prey [26].Within this set of wild-type predation specific DEGs not found in the Sskil1 and Ssste12 mutants also other transcriptional regulators, namely SsTEC1 and 2x SsNRG1, were identified.Analysis of the promoter regions of predation response genes indicated a highly enriched motif (Motif 3) with the consensus sequence 'TTTCAACATGCA', which was found in 93/147 predation response genes (Fig 6D and S8 Table).This suggests a central role of SsSTE12 in mediating the predation response in S. schoenii.Reporter gene analyses with two predation genes (Ssc08w0210-YPS3 and Ssc02c0119-CTS1) indicted that their expression under predation conditions is dependent on Ste12 (S4 Fig).
The predator-prey interaction period we chose in our experiment was 1.5 h.Under this condition we found only a very small set of 17 downregulated DEGs in the wild type.Nine of these genes were also found to be downregulated in Sskil1 and Ssste12.These genes are involved in metal ion transport and include SsFRE7, a putative ferric reductase and SsCTR1, a high-affinity copper transporter.In contrast, in Sskil1 120 DEGs and in Ssste12 111 DEGs were found to be downregulated with an overlap of 44 shared DEGs.In both mutant strains the dominant sets of DEGs were involved in ribosome biogenesis and rRNA processing (S9 Table ).Interestingly, we did not observe this stress response in the wild type indicative of wild type growth after predation.

Discussion
In this study we have identified the MAP kinase gene SsKIL1 as the key regulator of the predation response in the mycoparasitic yeast S. schoenii.Using phenotypic characterization and global transcript profiling, we determined a large overlap in Sskil1 mutant phenotypes with its conserved and putative downstream target SsSTE12.Both factors orchestrate the activation of genes required for penetration peg formation and the expression of cell wall degrading enzymes supporting the attack of prey cells.Deletion mutants in SsKIL1 and SsSTE12 are, therefore, non-pathogenic.In S. cerevisiae the mating and starvation/filamentation pathways are controlled by the MAP Kinase genes ScFUS3 and ScKSS1, respectively, and a central downstream target of these MAP kinases is the homeodomain transcription factor ScSte12 [37].Furthermore, this MAPK cascade is highly conserved in the fungal kingdom and has been shown to regulate pathogenicity in a wide variety of fungal pathogens [21].
The Magnaporthe ScSte12 homolog MoMst12 is a direct target of MoPmk1, which regulates the formation of melanized appressoria through activation of the transcription factor MoHox7 [18,38].Additionally, STE12 and STE12-like genes (with additional carboxy-terminal C 2 H 2 zinc fingers) have been shown to regulate sexual development, e.g. in Aspergillus and Neurospora, and pathogenicity in a wide variety of fungal systems including, for example, Botrytis cinerea, Candida albicans, Cryptococcus neoformans, Colletotrichum and Fusarium [21][22][23].
Saccharomycopsis species, in contrast to other members of the Saccharomycotina, possess only a single MAP kinase homolog of ScFUS3 and ScKSS1 (S1 Fig) .Our study indicated that S. schoenii SsSTE12 is dispensable for mating and thus dedicated to the predation process.Motif 3 'TTTCAACATGCA' that was present in a large number of predation response gene promoters bears similarity to a homodimeric S. cerevisiae ScSte12-binding site [39].This suggests direct activation of a large part of the predation response genes via SsSTE12.Our study revealed additional transcriptional regulators, SsTEC1 and SsNRG1, that may act as downstream targets of SsKil1/SsSte12 to orchestrate penetration peg formation.In S. cerevisiae the activation of filamentation genes requires both ScSTE12 and ScTEC1 [40].S. schoenii has a single homolog of SsTEC1.SsTEC1 contains one copy of Motif 3 and is strongly upregulated in wild type during predation (log 2 = 3.14), but only mildly upregulated in Sskil1 and Ssste12 (log 2 between 1.15 and 1.46).This suggests that a subset of predation genes may depend on both SsSte12 and SsTec1 for activation in S. schoenii.Another transcriptional regulator that was strongly upregulated in S. schoenii wild type was SsNRG1.There are five SsNRG1 paralogs in S. schoenii, three of these were upregulated in the wild type, but none in Sskil1 (S6 Table ).In S. cerevisiae ScNRG1 is a transcriptional repressor e.g. of filamentous growth [41].
The predation cycle in S. schoenii was more clearly defined by this study.Predation in Saccharomycopsis requires starvation as unstarved cells do not engage in predation even in the presence of prey [26,42].Starvation in the methionine auxotroph predator yeasts can easily be invoked by methionine deprivation as demonstrated.The starvation response we observed in our study centered on the expression of sulfur uptake and metabolism genes.In these genes conserved motifs were found that bear similarity to ScMet31/ScMet32 and ScCbf1 binding sites, which are cis-acting regulatory elements of genes involved in sulfate assimilation and

PLOS PATHOGENS
Penetration peg formation in predator yeasts methionine biosynthesis in S. cerevisiae [43].The methionine starvation response was independent of SsKIL1/SsSTE12.The Sskil1 and Ssste12 mutants showed an additional much stronger response to starvation than the wild type in our assay found in the downregulation of a large set of ribosomal genes similar to the stress response observed in S. cerevisiae, which under nutrient stress conditions also results in the repression of ribosomal protein genes [44].This indicates that the wild type rapidly-within the 1.5 h timeframe of our experiment-benefits from predation and quickly redeems the cost of penetration peg formation.Investigations on how hunger primes predation may provide insight into the upstream regulation of MAP kinase signaling in S. schoenii.In Magnaporthe the G-protein coupled receptor MoPth11 is required for appressorium differentiation induced by hydrophobic signals and acts upstream of the cAMP signaling pathway [45].In predator yeasts sensing and recognition of a prey cell occur and tropism towards a prey cell is apparently independent of SsKIL1.Components acting upstream of SsKIL1 could be involved in nutrient chemotropism, i.e. the sensing of carbon or nitrogen sources linked to a prey cell [46,47].In Candida albicans thigmotropism was shown to be regulated by the Ras-like GTPase CaRSR1 [48].Investigating nutrient transceptors like SsGPR1 and the Ras-like GTPase SsRSR1 in Saccharomycopsis may be helpful to elucidate predator-prey interactions.
Penetration pegs formed by predator yeasts resemble penetration pegs generated by appressoria of plant pathogenic fungi, two groups of fungi that separated more than 400 million years ago [49].This is a remarkable example of parallel evolution and shows the inherent plasticity of the Fus3/Kss1-MAP kinase cascade in fungi.However, predator yeast penetration pegs differ from e.g.Magnaporthe penetration pegs in that they are enriched with chitin, which may act as a shield against the cell wall lytic enzymes used in the attack of the prey.Furthermore, penetration pegs are a dead end: they are a one-off investment in mycoparasitism.Predator yeast penetration pegs do not receive a nucleus and thus do not develop into daughter cells or into nucleated infection hyphae as in Magnaporthe [50].On the contrary, during predation we observed a halt in daughter cell growth and a block of the cell cycle.Both activities were resumed after killing of the prey indicating a coordinated regulation of cell cycle progression and predation.In the corn smut fungus Ustilago maydis a cell cycle arrest in G2 has been described that is required for appressorium formation [13].Saccharomycopsis species are successful in attacking a diverse set of yeasts and filamentous fungi.A broad host range has also been observed in other necrotrophic fungi, for example in Trichoderma [42,51,52].
In summary, we have provided a detailed characterization of the predation cycle in mycoparasitic predator yeasts and demonstrated that S. schoenii predacious interactions are regulated by the MAP kinase SsKil1 and its putative target SsSte12, which act as global regulators of penetration peg formation required for invasion of prey cells.The SsKIL1/SsSTE12 signaling pathway controls predation response genes, which includes large families of cell wall lytic hydrolases, particularly chitinases, glucanases and proteases, but also additional transcription factors.Predator yeasts, therefore, represent a unique and simple unicellular system to study host-pathogen interactions and pathogenicity mechanisms that have evolved in parallel to those of distant filamentous plant-pathogenic ascomycetes.

Generation of targeted deletion mutants
Gene sequences were obtained from the S. schoenii draft genome sequence [26].All primers were obtained from Sigma-Aldrich, Taufkirchen, Germany) and sequences of the primers used are provided in S2 Table .Plasmids used and generated in this study are listed in S3 Table .Gene replacement cassettes were either generated by in vivo recombination in S. cerevisiae or were custom synthesized.To construct the SsKIl1 complete ORF deletion cassette homology regions (HRs) of ~1.5 kb upstream and downstream of the SsKIL1-ORF were PCR-amplified and cloned into pGEM (Promega, Madison, USA) using primers #44 and #45 for the upstream and #46 and #47 for the downstream HR.Adapter sequences and SmaI restriction sites were added to the SsKIL1 HRs by PCR using primers #48 and #49 for the 5'-HR and #50 and #51 for the 3'-HR.These PCR-fragments along with SAK1 marker gene [53] (amplified from pGEM-SAK1 with primers #96 and #97) and the SmaI-linearised pRS415 vector were co-transformed into S. cerevisiae using the LiAc/single-stranded carrier DNA/PEG method with DMSO [54,55].Correct assembly of the plasmid was verified by diagnostic PCR with primers #48 and #49 for 5'-HR, #50 and #51 for 3'-HR and #18 and #7272 for the internal region of SAK1.The SsSTE12 complete ORF deletion cassette was obtained from BioCat (Heidelberg, Germany).All plasmids were propagated in E. coli strain DH5α.Plasmid DNA was extracted using the PureYield Plasmid Midiprep System (Promega, Walldorf, Germany).SsKIl1 disruption cassette was excised via SmaI and SsSTE12 disruption cassette was amplified as full-length fragment from plasmid DNA using primers #1022 and #1023.Targeted gene deletion in predator wild type strain was achieved as described [56].Deletion mutants were verified via diagnostic PCR using G1-G2, G3-G4 and I1-I2 primer pairs #80-#7269, #7270-#81, #82-#83, respectively, for Sskil1 and primer pairs #1020-#1026, #7270-#1021, #1018-#1019 for Ssste12 (S2 and S3 Tables).

Generation of strains for complementation, GFP-expression, marker assisted breeding and promoter analysis
For complementation of the Sskil1 defect a cassette was constructed by in vivo cloning in S. cerevisiae.To this end the pRS415-YES2 plasmid was linearized with SmaI and co-transformed into S. cerevisiae with a PCR product generated from S. schoenii genomic DNA with primers #445 and #447 containing the SsKIL1 gene.Correct assembly was verified by PCR.The obtained plasmid was linearized with EcoRV and transformed into Sskil1 strain B054.Plasmid E196 carrying a synthetic S. schoenii histone H4-GFP fusion gene, obtained from Genscript (Leiden, Netherlands), was linearized by SacI digestion and transformed into S. schoenii.For marker-assisted breeding strains to be mated were equipped with either hygromycin or nourseothricin resistance marker genes to allow for double selection of progeny.To this end PCRfragments of the marker genes were amplified from YES2 (E068) and YES3 (E070) plasmids using primer pairs #285-#287 and #1003-#1004, respectively, and transformed into the selected S. schoenii strains.

Predation assays
An array of assays was used to analyze predacious interactions of S. schoenii strains with its prey S. cerevisiae, ScH4-GFP.Standard predation assays were performed by mixing equal amounts of predator and prey cells spread on SD-plates.Interactions were allowed to proceed for up to 20 h at 22˚C.For subsequent microscopic observations, cells were washed off the plates after 4 h and stained with CW prior to imaging.For direct in situ analyses of predatorprey interactions, objective slides covered with 10 ml SD medium were used onto which predator and prey cells were spotted, and incubated for 20 h at 22˚C.Cells were stained with CW and for each strain 250 predator-prey-interactions were evaluated microscopically to quantify the ratio of prey cells attached to predator cells that were alive (i.e.predator cells formed no penetration pegs and prey cells were alive according to their GFP signal) or penetrated and dead (predator cells formed pegs and prey cells lost their GFP signal).Samples were prepared in triplicates.In plate-based assays, ~600 cells of the prototrophic prey strain S. cerevisiae CEN.PK were trialled against a lawn of predator cells (~1.0x10 7 cells) on SD medium for 3-4 days at 22˚C.Predator yeasts are methionine auxotrophs and thus unable to proliferate on these plates, while the prototrophic prey strain may form colonies if no predation has occurred.Plates were imaged using a Canon EOS 77D camera equipped with an EF-S 18-135 mm objective.Prey CFUs were determined by counting.The assays were performed for three to six biological replicates with five technical replicates and results are provided as the mean ± s.e.m.

Mating assay
In these assays, 2.5 x 10 7 cells of each strain were mixed, plated on corn meal agar (Merck KGaA, Darmstadt, Germany) and incubated for 4-5 days at 22˚C.Then cells were washed off the plates, treated with 1.6 mg/ml zymolyase (Genaxxon bioscience GmbH, Ulm, Germany) for 1 h at 30˚C and then with 0.04 mg/ml Triton X-100 for 30 min at RT to eliminate vegetative cells.Samples were washed three times with H 2 O, resuspended in 750 μl H 2 O of which 1/10 volume was plated on double selective YPD medium containing hygromycin and nourseothricin to select for hybrid progeny expressing both marker genes.Two biological replicas were assayed with three technical replicates for each strain combination.

Microscopic analyses
Microscopic images were acquired either by an Axiovert 200M (Zeiss, Jena, Germany) microscope equipped with an AxioCam MRm camera run by the Axiovision 4 software package (Zeiss, Jena, Germany) or using an Axio Imager with a pco.edge 4.2bi camera controlled by

PLOS PATHOGENS
Penetration peg formation in predator yeasts the VisiView 6 software (Visitron Systems GmbH, Puchheim, Germany).For in vivo timelapse microscopy, predator strains and S. cerevisiae H4-GFP were mixed on deep well slides containing solidified CSM medium supplemented with 5 μg/ml CW.Image acquisition was automated with DIC images taken every 2 min and GFP and CW fluorescence images every 4 min.Appropriate excitation and emission Chroma ET filter sets were used with LED illumination.Image processing was done using ImageJ 1.53c.Images of similar treated cells or all single plane files of a time-lapse series were stacked channel-wise and processed equally.We used the ImageJ plugins 'NMS_fixTranslation_ver1.ijm' for drift correction and 'SPIM_DrawArrowIn-Movie_-14.0.0.jar' to add arrows to a movie.

Staining procedures
Yeast cell walls were stained with the fluorescent dyes calcofluor white (CW) or the fluorescent lectin conjugates Wheat Germ agglutinin (CF-488A Wheat germ agglutinin, WGA) or concanavalin A (CF-488A, conA).Dyes were purchased from VWR International GmbH (Darmstadt, Germany).Cells were treated with 0.01 mg/ml CW or co-stained with CW and up to 0.15 mg/ml WGA or conA.Predator wild type cells form penetration pegs even in the absence of prey when cultivated on corn meal agar plates.These cells were washed off the plates, resuspended and co-stained with CW and 0.1 mg/ml WGA or conA.

Transcript profiling via RNAseq analysis
For transcriptomics, 2.27x10 7 cells of S. schoenii wild type (CBS 7425), Sskil1 (B054 and G238) and Ssste12 (G404 and G398) were cultured either alone on YPD (starvation/predation suppressing medium) or co-cultured with the same amount of S. cerevisiae BY4741 prey cells on SD (starvation/predation inducing medium) for 1.5 h.Cells were collected from these plates and frozen in liquid nitrogen.RNA extraction, quality control, library preparation (PolyA selection with ERCC spike-in), sequencing at 30M paired-end (2x150 bp) reads/sample using Illumina NovaSeq and standard data analysis were performed at Genewiz (Azenta Life Sciences, Leipzig, Germany).Genes with p adj < 0.05 and a 4-fold change in expression level on hunger/predation inducing conditions (SD) compared to suppressing conditions (YPD), i.e. upregulation (log 2 � 2) and downregulation (log 2 � -2), were defined as differentially expressed genes (DEGs).The overlapping sets of DEGs of biological replicates of Sskil1 B054 and G238 as well as Ssste12 G398 and G404, respectively, were compared to DEGs of the wild type.

Gene set analyses
Subsets of unique and co-regulated DEGs displayed in Fig 6 were subjected to enrichment analyses using the generic gene ontology (GO) term finder (https://go.princeton.edu/cgi-bin/GOTermFinder) with default settings.GO terms associated with biological processes,

PLOS PATHOGENS
Penetration peg formation in predator yeasts molecular function and cellular components with a p-value < 1.0E-02 were considered to be significantly enriched.Redundancy among GO terms were reduced using REVIGO v 1.8.1 in default settings and the S. cerevisiae S288C database (http://revigo.irb.hr/).

Motif enrichment analysis
Motif discovery in promoter sequences (defined as up to -1 kb of the start codon) of upregulated DEGs with log 2 � 2 was carried out using MEME (https://meme-suite.org/meme/index.html).Motif searches were limited for sequences ranging between 6 to 24 bases with multiple but non-overlapping occurrences.Letter probability matrices of significantly enriched motifs were subjected to Tomtom v 5.5.5 using YEASTRACT database (https://meme-suite.org/ meme/tools/tomtom).Hits of query motifs to a transcription factor binding site (TFBS) identified in S. cerevisiae with a p-value < 1.0E-04 were reported.

Statistics and reproducibility
Datasets of mating experiment were subjected to Shapiro-Wilk normality test which yielded pvalues > 0.05 indicating normal distribution of data.A one-sided Welch's unpaired t-test was applied and significant results were considered at p-values < 0.05.
Transcriptomes of S. schoenii wild type and its derivative strains analyzed in this study have been submitted to NCBI BioProject database under the accession ID "PRJNA1091439".
All experiments were conducted with at least two biological replicates and technical replicates of an appropriate sample size, estimated based on what is established in the field.The sample sizes, number of biological and technical replicates, and the statistical tests used in each experiment are specified in the figure legends.
bottom left panel), while CW stains the entire stem of the peg (Fig 1E top left pane).This may be due to the size of the lectin that is far bigger than CW.Therefore, we stained cells with free penetration pegs not surrounded by prey cells.Again, CW stained the entire peg but here also WGA colored the base and tip of the penetration peg (Fig1Eright panels).We then compared CW staining (Fig 1F upper panels) with conA staining (Fig 1F lower panels).The lectin conA, which selectively binds to α-gluco-and α-mannopyranosyl residues, uniformly stained the cell walls of predator and prey cells but did neither label free penetration pegs nor pegs inside prey cells (Fig 1F bottom panels, respectively).This suggests that the penetration pegs are specifically enriched with chitin in their cell wall.
Time-lapse microscopy was used to analyze the predation cycle in S. schoenii.Histone H4-GFP-tagged predator and prey cells were mixed on minimal medium agar slides and DIC, PLOS PATHOGENS Penetration peg formation in predator yeasts CW and GFP images were acquired over time (Fig 2 and S1 Movie).Fig 2 depicts key events during predation for both predator and prey cells.At the onset all cells, predator and prey, show H4-GFP-fluroescence indicating that cells were alive (Fig 2A).A S. schoenii mother cell with an elongated daughter cell was in contact with a prey cell connected to a large budded daughter cell after mitosis (Fig 2B at 20').This contact resulted in the formation of a penetration peg (marked by a dagger in Fig 2B at 20').This initiated predation and paused further growth of the daughter cell and further progression of the cell cycle, i.e. the mother nucleus did not enter mitosis.The CW signal at the penetration peg became more intense (Fig 2B at 24'), while the GFP signal of both mother and daughter prey cells disappeared (Fig 2B between 24' and 48').Killing of the prey cells took about 25 minutes from the appearance of a CW signal of the peg until the disappearance of the prey cell nuclear GFP (Fig 2B).Once the prey cells were killed the S. schoenii daughter cell resumed its polar tip growth at its previous position (Fig 2C at 60').The mitosis occurred and cell separation was marked by the appearance of a CW-stained septum between the S. schoenii mother and daughter cells (Fig 2C).Shortly after that the mother cell initiated a new cell cycle (Fig 2D at 144').The tip of the daughter cell contacted a prey daughter cell and a new penetration peg was formed (Fig 2D at 192').Killing of the S. cerevisiae mother and daughter cells occurred within 16 minutes (Fig 2D between 192' and 208').After the disappearance of the nuclear GFP-signal prey cells were observed to shrink indicating material transfer into the predator cell (Fig 2D between 208' and 236').Thus

Fig 1 .
Fig 1. Fluorescence microscopy reveals chitin enrichment of penetration pegs in S. schoenii.(A-D) Live-cell images showing chitin enrichment of calcofluor white (CW)-stained S. schoenii (elongated) and S. cerevisiae cells (ovoid).(A) Several penetration pegs (two marked by daggers; the left one next to a septum separating mother and daughter S. schoenii cells) are located inside S. cerevisiae prey cells.The S. cerevisiae cells carry a H4-GFP tag but only the one cell at the bottom (marked by an arrow) is showing nuclear fluorescence, while all others do not and are thus dead.S. cerevisiae bud scars (two positions marked by arrowheads) stain brightly with CW. (B) A predator yeast cell (center) with two penetration pegs that attacked two S. cerevisiae prey cells to the left and right of it is shown.(C) A S. cerevisiae mother cell connected with a daughter cell (the septum between both cells is marked by an asterisk) were attacked simultaneously by different predator cells; one of the four penetration pegs is marked by a dagger.(D) Two examples of S. schoenii predation pegs inside S. cerevisiae prey cells are shown indicating bulbous tips of the penetration pegs.(E) Micrographs showing co-staining of penetration pegs with CW (upper panels, images obtained with DAPI-filterset) and Wheat Germ Agglutinin (WGA CF-488A, lower panels, images acquired with the GFP filterset) of the same cells.(F) Images of cells co-stained with CW (upper panels) and concanavalin A (conA CF488A, lower panels).(E,F) Panels on the right side show predator-prey cell interactions, while panels on the left depict predator cells without prey cells.(A-F), Scale bars, 5 μm.https://doi.org/10.1371/journal.ppat.1012503.g001

Fig 2 .
Fig 2. Time-lapse fluorescence microscopy defines distinct stages of the predation cycle in S. schoenii.(A) Schematic drawing of a mixture of elongated S. schoenii predator and ovoid S. cerevisiae prey cells at the onset of a time-lapse recording (line drawing on the left based on the DIC + GFP image at 4').The time series consisted of brightfield images (DIC), GFP-images recording the nuclear H4-GFP and CW-images monitoring the appearance of penetration pegs and septa.Upper panels show combined DIC and GFP images, lower panels CW and GFP images.Nuclei of S. schoenii are encircled with oval lines (black solid lines in the upper images, white solid lines in the lower images).Nuclei of S. cerevisiae are marked with dashed circles.(B) Penetration peg formation is marked by a dagger (20'-24').S. schoenii nuclei are marked in the upper panel, while S. cerevisiae nuclei are marked in the lower panel until their disappearance.The position of the tip of the S. schoenii daughter cell is marked by an arrowhead.(C) Progression of the cell cycle after predation in the predator cell.The growing daughter cell bud tip is marked by arrowheads.S. schoenii nuclei are encircled in the upper images indicating mitosis at 96'.After mitosis a septum was generated, which is marked by a '+'.(D) The mother cell entered a new cell cycle marked by bud emergence (arrowhead, 144').The S. cerevisiae prey cell is large-budded, its nucleus undivided (marked by a dashed circle) and deposition of chitin at the bud neck is marked by an asterisk.Contact of the predator yeast daughter cell with the prey cell resulted in penetration peg formation (dagger), disappearance of prey cell H4-GFP and shrinking of the prey cell (daughter cell is marked by black dashed circles at 208' and 236').Underneath the prey cell killed in (B) a predator cell grew (D), which was not involved in the predation of this prey cell and was not the result of further growth of the penetration peg (see also S1 Movie).Scale bar, 5 μm.https://doi.org/10.1371/journal.ppat.1012503.g002

Fig 3 .
Fig 3. SsKIL1 and SsSTE12 are both required for penetration peg formation and predation in S. schoenii.(A) Representative images of predator-prey interactions of the indicated S. schoenii strains with S. cerevisiae, H4-GFP cells.Cells were stained with CW and CW-and GFP-fluorescence images were acquired and overlaid using ImageJ.Several penetration pegs (absent in Sskil1 and Ssste12) are marked by daggers; several prey cells showing nuclear H4-GFP fluorescence are marked by arrows.WT, wildtype; scale bar, 5 μm.(B) Diagram to show the experimental design for in situ analysis of predation.Cells of different S. schoenii strains were mixed with S. cerevisiae H4-GFP cells and spread on glass slides placed in petri dishes and covered with SD medium.Petri dishes were incubated as indicated.Then glass slides were excised from the matrix, cells were stained with CW and evaluated directly under the microscope.Events scored only included prey cells in direct contact with S. schoenii cells; dead cells penetrated by a pegs or alive unpenetrated prey cells were quantified.(C) Enumeration of predator-prey interactions of the indicated strains prepared as described in (B).WT and Sskil1/SsKIL1 strains were assayed in triplicate, the Sskil1 and Ssste12 mutant strains were assessed by 2 biological replicates and 3 technical replicates per biological replicate, Mean ± s.e.m. (D) Experimental design of a plate based predation assay.A lawn of cells of the S. schoenii tester strain was plated on SDplates and ~600 S. cerevisiae prey cells were spread onto this lawn.S. cerevisiae CFUs were counted within the indicated square.(E) Top row shows representative images of petri plates corresponding to the squares shown in (D).S. cerevisiae CFUs that were formed against the lawn of predator cells enumerated as shown in the graph underneath the corresponding plate images.WT and Sskil1 (B054) were assayed with 6 biological replicates and 5 technical replicates.Sskil1 (G238), Sskil1/SsKIL1 and the Ssste12 mutant strains were assessed with 3 biological replicates and 5 technical replicates per biological replicate, Mean ± s.e.m. https://doi.org/10.1371/journal.ppat.1012503.g003

Fig 4 .
Fig 4. Time-lapse microscopy reveals thigmotropism that preceeds predation in S. schoenii.(A) Selected frames of an in vivo time-lapse recording (S3 Movie) showing S. schoenii (SsKIL1 wildtype, elongated cell) predation of an S. cerevisiae prey cell (ovoid cell).Both strains are tagged by histone H4-GFP genes to monitor cell viability.The nucleus of S. schoenii is encircled with oval lines (black solid lines in the upper images, white solid lines in the lower images); the nucleus of the attacked S. cerevisiae cell is marked with a dashed circle in each frame.Penetration peg formation is marked by a dagger (72').Note the chitin rich bud neck in S. cerevisiae (marked by an asterisk at 72') is present before the penetration peg appears.(B) Selected micrographs of a time-lapse recording (S4 Movie) showing the interaction of a Sskil1 cell (elongated) with a S. cerevisiae prey cell (ovoid, displaying a nuclear H4-GFP).(A,B) At each time point (indicated in minutes, top left in each frame) brightfield (DIC), CW and GFP images were acquired.DIC and GFP images (upper panels) and CW and GFP images (lower panels) were stacked using ImageJ.Strains were incubated on SD medium.Scale bars, 5 μm.https://doi.org/10.1371/journal.ppat.1012503.g004

Fig 5 . 4 +
Fig 5. SsSTE12 is dispensable for mating.(A) Experimental design of marker assisted breeding in S. schoenii.Strains were equipped with either hygromycin (Hyg R )or nourseothricin (clonNAT R ) resistance genes and plated on sporulation media.Spores were enriched by zymolyase/Triton X-100-treatment after wash-off and collection in test tubes and then plated on double selective medium to identify progeny harboring both markers.(B) representative images of selection plates of the indicated mating combinations.(C) Plot of CFUs (= the number of hybrids, "N hybrids") identified with the indicated strain combinations.Mating was assessed using 2 biological and three technical replicates for each combination, Mean ± s.e.m. https://doi.org/10.1371/journal.ppat.1012503.g005

Fig 6 .
Fig 6.Comparative global transcriptional profiling defines the roles of SsKIL1 and SsSTE12 in the predation response of S. schoenii.(A) Venn diagram illustrating overlapping DEGs with at least log 2 � 2 differential expression under predation conditions between the wild type and the Sskil1 and Ssste12 mutant strains.Two distinct gene sets were identified: the hunger response common to all strains (a) and the predation response constrained to the wild type (B), each with a set of enriched GO terms.(C,D) Consensus DNA-binding motifs, predicted using MEME, for the hunger and predation response gene sets with best matching transcription factor binding site matches (TFBS) to S. cerevisiae transcriptional regulators.https://doi.org/10.1371/journal.ppat.1012503.g006