A fungal member of the Arabidopsis thaliana phyllosphere antagonizes Albugo laibachii via a GH25 lysozyme

Much like the ‘good bacteria’ that live in our guts, many microscopic organisms can co-exist with and even benefit the plants they live on. For instance, the yeast Moesziomyces bullatus ex Albugo (MbA for short) can shield the leaves of its plant host against white rust, a disease caused by the organism Albugo laibachii. Studies have started to unveil how the various microbes at the surface of leaves interact and regulate their own community, yet the genetic mechanisms at play are less well-known.

To investigate these processes, Eitzen et al. examined the genes that were switched on when MbA cells were in contact with A. laibachii on a leaf. This experiment revealed a few gene candidates that were then deleted, one by one, in MbA cells. As a result, a gene emerged as being key to protect the plant from white rust. It produces an enzyme known as the GH25 hydrolase, which, when purified, could reduce A. laibachii infections on plant leaves.

Bacteria, fungi and other related microorganisms cause many diseases which, like white rust, can severely affect crops. Chemical methods exist to prevent these infections but they can have many biological and ecological side effects. A solution inspired by natural interactions may be safer and more effective at managing plant diseases that affect valuable crops. Harnessing the interactions between microbes living on plants, and the GH25 enzyme, may offer better disease control.

Plants are colonized by a wide range of microorganisms. While some microbes enter the plant and establish endophytic interactions with a broad range of outcomes from beneficial to pathogenic, plant surfaces harbor a large variety of microbial organisms. Recent research has focused largely on the importance of the rhizosphere microbiota in nutrient acquisition, protection from pathogens, and boosting overall plant growth and development (Ritpitakphong et al., 2016; Walker et al., 2003; Bulgarelli et al., 2013). However, the above ground parts of the plant including the phyllosphere are colonized by diverse groups of microbes that also assist in plant protection and immunity (Busby et al., 2016; Mikiciński et al., 2016). The environment has a major impact on the microbial communities of the leaf surface, ultimately influencing their interactions with the host (Stone et al., 2018).

Scale-free network analysis was performed with the leaf microbial population of Arabidopsis thaliana (Agler et al., 2016). The majority of the interactions between kingdoms, e.g. fungi and bacteria, were found to be negative, consistent with the fact that rather the antagonistic interactions stabilize a microbial community (Coyte et al., 2015). Phyllosphere network analysis of A. thaliana identified a small number of microbes as ‘hub’ organisms, i.e. influential microbes that have severe effects on the community structure. The major hub microbe in the A. thaliana phyllosphere is the oomycete Albugo laibachii, which is a pathogenic symbiont biotrophic of Arabidopsis (Agler et al., 2016). This pathogen has been shown to significantly reduce the bacterial diversity of epiphytic and endophytic leaf habitats. Since bacteria generally comprise a large proportion of the phyllosphere microbiome (Vorholt, 2012), phylogenetic profiling of A. thaliana was also directed toward identifying a small group of bacteria that frequently colonize A. thaliana leaves. The analysis helped to develop a synthetic community of bacteria for experiments in gnotobiotic plants.

Besides bacteria and oomycetes, the microbiota of the A. thaliana leaf also comprises a broad range of fungi. Among those fungi, basidiomycete yeasts are frequently found and the most frequent ones are the epiphytic basidiomycete genus Dioszegia (Agler et al., 2016), as well as an anamorphic yeast associated with A. laibachii infection belonging to the Ustilaginales. This order includes many pathogens of important crop plants, for example corn smut and loose smut of oats, barley, and wheat are caused by Ustilago maydis, U. avenae, U. nuda, and U. tritici, respectively. Generally, the pathogenic development of smut fungi is linked with sexual recombination and plant infection is only initiated upon mating, when two haploid sporidia form a dikaryotic filament (Brefort et al., 2009). Ustilaginales Pseudozyma sp. yeasts, however, are found mostly in their anamorphic stage in nature. They tend to epiphytically colonize a wide range of habitats, where an infrequent sexual recombination might occur when they meet on a susceptible host (Kruse et al., 2017). Phylogenetic reconstruction (Kruse et al., 2017; Wang et al., 2015) showed that the smut pathogen of millet, Moesziomyces bullatus and four species of Pseudozyma, namely P. antarctica, P. aphidis, P. parantarctica, and P. rugulosa form a monophyletic group. The latter do represent anamorphic and culturable stages of M. bullatus and, hence, can be grouped to this genus. Moesziomyces strains have been reported in a number of cases to act as microbial antagonists. A strain formerly classified as Pseudozyma aphidis (now M. bullatus) inhibited Xanthomonas campestris pv. vesicatoria, X. campestris pv. campestris, Pseudomonas syringae pv. tomato, Clavibacter michiganensis, Erwinia amylovora, and Agrobacterium tumefaciens in vitro and also led to the activation of induced defense responses in tomato against the pathogen (Barda et al., 2015). It was reported that P. aphidis can parasitize the hyphae and spores of Podosphaera xanthii (Gafni et al., 2015). Pseudozyma churashimaensis was reported to induce systemic defense in pepper plants against X. axonopodis, Cucumber mosaic virus, Pepper mottle virus, Pepper mild mottle virus, and broad bean wilt virus (Lee et al., 2017).

In the present study, we explored the antagonistic potential of an anamorphic Ustilaginales yeast within the leaf microbial community of A. thaliana. We show that Moesziomyces bullatus ex Albugo on Arabidopsis (which will be referred to as MbA from further on) prevents infection by the oomycete pathogen A. laibachii and identified fungal candidate genes that were upregulated in the presence of A. laibachii, when both microbes were co-inoculated to the host plant. A knockout mutant of one of the candidates, which belongs to the glycoside hydrolase – family 25 (GH25) – was found to lose its antagonistic activity toward A. laibachii, providing mechanistic insights into fungal-oomycete antagonism within the phyllosphere microbiota. Functional characterization of GH25 will be an important step toward establishing MbA as a suitable biocontrol agent.

In a previous study we isolated a basidiomycetous yeast from Arabidopsis thaliana leaves infected with the causal agent of white rust, Albugo laibachii (Agler et al., 2016). This yeast was tightly associated with A. laibachii spore propagation. Even after years of subculturing in the lab and re-inoculation of plants with frozen stocks of A. laibachii isolate Nc14, this yeast remained highly abundant in spore isolates. Phylogenetic analyses based on fungal ITS-sequencing identified the yeast as Pseudozyma sp. Those yeasts can be found across the family of Ustilaginaceae, being closely related to pathogens of monocots like maize, barley, sugarcane, or sorghum (Zuo et al., 2019). Based on phylogenetic similarity to the pathogenic smut Moesziomyces bullatus which infects millet, several anamorphic Pseudozyma isolates were suggested to be renamed and grouped to M. bullatus (Wang et al., 2015). Since the Pseudozyma sp. that was isolated from A. laibachii spores groups into the same cluster, we classified this newly identified strain as MbA (Moesziomyces bullatus ex Albugo on Arabidopsis).

Based on the identification of MbA as having a significant effect on bacterial diversity in the Arabidopsis phyllosphere, we tested its interaction with 30 bacterial strains from 17 different species of a synthetic bacterial community (SynCom, Supplementary file 1) of Arabidopsis leaves in one-to-one plate assays. This experiment identified seven strains being inhibited by Moesziomyces, as indicated by halo formation after 7 days of co-cultivation (Figure 1—figure supplement 1). Interestingly, this inhibition was not seen when the pathogenic smut fungus U. maydis was co-cultivated with the bacteria, indicating a specific inhibition of the bacteria by MbA (Figure 1—figure supplement 1).

The primary hub microbe in the Arabidopsis phyllosphere was found to be the pathogenic oomycete A. laibachii, which was isolated in direct association with Moesziomyces (Agler et al., 2016). To test if both species interfere with each other, we deployed a gnotobiotic plate system and quantified A. laibachii infection symptoms on Arabidopsis. In control experiments, spray inoculation of only A. laibachii spores on Arabidopsis leaves led to about 33% infected leaves at 14 days post infection (dpi) (Figure 1). When the bacterial SynCom was pre-inoculated on leaves 2 days before A. laibachii spores a significant reduction of A. laibachii infection by about 50% was observed (Figure 1). However, if Moesziomyces was pre-inoculated with the bacterial SynCom, A. laibachii spore production was almost completely abolished. Similarly, the pre-inoculation of only MbA resulted in an almost complete loss of A. laibachii infection, independently of the presence of a bacterial community (Figure 1). The antagonistic effect of MbA toward A. laibachii was further confirmed using Trypan blue staining of A. laibachii infected A. thaliana leaves. A. laibachii forms long, branching filaments on Arabidopsis leaves at 15 dpi. Contrarily, in the presence of MbA, we observed mostly zoospores forming either no or very short hyphae, while further colonization of the leaf with long, branching was not observed (Figure 1—figure supplement 2). Together, our findings demonstrate that MbA holds a strong antagonistic activity toward A. laibachii, resulting in efficient biocontrol of pathogen infection. Thus, MbA is an important member of the A. thaliana phyllosphere microbial community, with a strong impact on its quantitative composition. However, despite several reports of the basidiomycete yeasts acting as antagonists, genome information of this group is rather limited. To enable a molecular understanding of how MbA acts on other members of the phyllosphere community, MbA genome information is required. We therefore sequenced the genome of MbA and established molecular tools including a protocol for stable genomic transformation to allow functional genetic approaches.

Figure 1 with 2 supplements see all

Download asset Open asset

Figure 1—source data 1

Quantification of A. laibachii infection on A. thaliana.

https://cdn.elifesciences.org/articles/65306/elife-65306-fig1-data1-v2.xlsx

Download elife-65306-fig1-data1-v2.xlsx

The genome of MbA

Genome sequence of MbA was analyzed by Single Molecule Real-Time sequencing (Pacific Biosciences, Menlo Park, CA), which lead to 69,674 mapped reads with an accuracy of 87.3% and 8596 bp sub-read length. Sequence assembly using the HGAP-pipeline (Pacific Biosciences) resulted in 31 contigs with an N₅₀ Contig Length of 705 kb. The total length of all contigs results in a predicted genome size of 18.3 Mb (Table 1). Gene prediction for the MbA genome with Augustus (Stanke et al., 2004) identified 6653 protein coding genes, of which 559 carry a secretion signal. Out of these 559, 380 are predicted to be secreted extracellularly (i.e. they do not carry membrane domains or cell-wall anchors) (Table 1). The small genome size and high number of coding genes result in a highly compact genome structure with only small intergenic regions. These are features similarly found in several pathogenic smut fungi such as U. maydis and S. reilianum (Table 1). Remarkably, both MbA and Anthracocystis flocculosa, which is another anamorphic and apathogenic yeast, show a similarly high rate of introns, while the pathogenic smut fungi have a significantly lower intron frequency (Table 1).

Table 1

	MbA	U. bromivora	S. scitamineum	S. reilianum	U. maydis	U. hordei	M. pennsylvanicum	A. flocculosa
Assembly statistics
Total contig length (Mb)	18.3		19.5	18.2	19.7	20.7	19.2	23.2
Total scaffold length (Mb)		20.5	19.6	18.4	19.8	21.15	19.2	23.3
Average base coverage	50×	154×	30×	20×	10×	25×	339×	28×
N50 contig (kb)	705.1		37.6	50.3	127.4	48.7	43.4	38.6
N50 scaffold length (kb)		877	759.2	738.5	817.8	307.7	121.7	919.9
Chromosomes	21	23		23	23	23
GC-content (%)	60.9	52.4	54.4	59.7	54	52	50.9	65.1
Coding (%)	62.8	54.4	57.8	62.6	56.3	54.3	54	66.3
Coding sequence
Percentage CDS (%)	69.5	59.8	62	65.9	61.1	57.5	56.6	54.3
Average gene size (bp)	1935	1699	1819	1858	1836	1708	1734	2097
Average gene density (gene/kb)	0.36	0.35	0.34	0.37	0.34	0.34	0.33	0.30
Protein-coding genes	6653	7233	6693	6648	6786	7113	6279	6877
Exons	11,645	11,154	10,214	9776	9783	10,907	9278	19,318
Average exon size (bp)	1091	1101	1191	1221	1230	1107	527	658
Exons/gene	1.75	1.5	1.5	1.47	1.44	1.53	1.48	2.8
tRNA genes	150	133	116	96	111	110	126	176
Noncoding sequence
Introns	9333	3921	3521	3103	2997	3161	2999	12,427
Introns/gene	1.40	0.54	0.53	0.47	0.44	0.44	0.48	1.81
Average intron length (base)	163	163	130.1	144	142	141	191.4	141
Average intergenic distance (bp)	769	1054	1114	929	1127	1186	1328	1273
Secretome
Protein with signal peptide	559		622	632	625	538	419	622
Secreted without TMD	380				467			737
– with known domain	260				264			554

1. [Brefort et al., 2014; Nadal and Gold, 2010].

To gain better insights in the genome organization of MbA, we compared its structure with the U. maydis genome, which served as a manually annotated high-quality reference genome for smut fungi (Kämper et al., 2006). Out of the 31 MbA contigs, 21 show telomeric structures and a high synteny to chromosomes of U. maydis, with three of them displaying major events of chromosomal recombination (Figure 2A). Interestingly, the Moesziomyces contig 2, on which also homologs to pathogenic loci like the U. maydis virulence cluster 2A (Kämper et al., 2006) can be found, contains parts of three different U. maydis chromosomes (Chr. 2, 5, and 20) (Figure 2—figure supplement 1). The second recombination event on contig six affects the U. maydis leaf-specific effector protein See1, which is required for tumor formation (Redkar et al., 2015). This recombination event is also found in the genome of the maize head smut S. reilianum, wherein the U. maydis chromosomes 5 and 20 recombined in the promoter region of the see1 gene (Figure 2B). In this respect it should be noted that S. reilianum, although infecting the same host, does not produce leaf tumors as U. maydis does (Schirawski et al., 2010).

Figure 2 with 1 supplement see all

Download asset Open asset

Also the third major recombination event, affecting MbA contig 8, changes the genomic context genes encoding essential virulence factors in U. maydis (stp1 and pit1/2), as well as the A mating type locus, which is important for pheromone perception and recognition of mating partners (Bölker et al., 1992). Based on the strong antibiotic activities of MbA, we mined the genome of MbA for the presence of secondary metabolite gene clusters. Using AntiSMASH, we were able to predict 13 of such clusters, of which three can be assigned to terpene synthesis, three contain nonribosomal peptide synthetases, and one cluster has a polyketide synthase as backbone genes (Figure 3—figure supplement 1). Interestingly, the secondary metabolite cluster that is involved in the production of the antimicrobial metabolite ustilagic acid in other Ustilaginomycetes is absent in MbA (Figure 3—figure supplement 1B). On the contrary, we could identify three MbA-specific metabolite clusters that could potentially be involved in the antibacterial activity of MbA (Figure 3—figure supplement 1C).

A previous genome comparison of the related Ustilaginales yeast A. flocculosa with U. maydis concluded that this anamorphic strain had lost most of its effector genes, reflecting the absence of a pathogenic stage in this organism (Lefebvre et al., 2013). In contrast, MbA contains 1:1 homologs of several known effectors with a known virulence function in U. maydis (Table 2). We previously found that Moesziomyces sp. possess functional homologues of the pep1 gene, a core virulence effector of U. maydis (Sharma et al., 2019), suggesting that such anamorphic yeasts have the potential to form infectious filamentous structures by means of sexual reproduction (Kruse et al., 2017). To assess the potential virulence activity of MbA effector homologs, we expressed the homolog of the U. maydis core effector Pep1 in an U. maydis pep1 deletion strain (SG200∆01987). This resulted in complete restoration of U. maydis virulence, demonstrating that MbApep1 encodes a functional effector protein (Figure 3—figure supplement 2).

Table 2

Name	Homologue	Query cover	E-value	Identity (%)	U. maydis knockout phenotype	Reference
g1653	UMAG_01987 (Pep1)	82%	3-e56	60.96	Complete loss of tumor formation – blocked in early stages of infection	Doehlemann et al., 2009
g1828	UMAG_01829 (Afu1)	99%	0.0	71.57	Organ-specific effector – reduced virulence in seedling leaves	Schilling et al., 2014
g2626	UMAG_12197 (Cce1)	98%	1e-48	60.16	Complete loss of tumor formation – blocked in early stages of infection	Seitner et al., 2018
g2765	UMAG_11938 (Scp2)	100%	1e-73	93.44	Reduced in virulence	Krombach et al., 2018
g2910	UMAG_02475 (Stp1)	32%	3e-42	60.71	Complete loss of tumor formation – blocked in early stages of infection	Schipper, 2009
g3652	UMAG_02239 (See1)	43%	9e-11	54.90	Organ-specific effector – reduced virulence in seedling leaves	Redkar et al., 2015
g3113	UMAG_01375 (Pit2)	*	*	*	Complete loss of tumor formation – blocked in early stages of infection	Doehlemann et al., 2011
g3279	UMAG_03274 (Rsp3)	10%	5e-20	70.11	Strong attenuation of virulence – reduced tumor size and number	Ma et al., 2018
g5296	UMAG_05731 (Cmu1)	98%	3e-70	43.84	Reduced virulence	Djamei et al., 2011
g6183	UMAG_06098 (Fly1)	100%	0.0	81.85	Reduced virulence	Ökmen et al., 2018
g5835	UMAG_05302 (Tin2)	87%	8e-24	37.81	Minor impact on tumor formation – reduced anthocyanin biosynthesis	Brefort et al., 2014

A hallmark of the U. maydis genome structure is the presence of large clusters with effector genes, the expression of which is only induced during plant infection (Kämper et al., 2006). To assess the presence of potential virulence clusters in MbA, we compared all U. maydis effector gene clusters to the MbA genome, based on homology. This revealed that the 12 major effector clusters of U. maydis are present in MbA. However, while many of the clustered effector genes are duplicated in pathogenic smut fungi, MbA carries only a single copy of each effector gene. This results in the presence of ‘short’ versions of the U. maydis gene effector clusters (Figure 3—figure supplement 3). This gets particularly obvious for the biggest and most intensively studied virulence cluster of smut fungi, the effector cluster 19A (Schirawski et al., 2010; Brefort et al., 2014; Dutheil et al., 2016). In MbA, only 5 out of the 24 effector genes present in U. maydis are conserved in this cluster (Figure 3). Interestingly, some anamorphic yeasts like Kalmanozyma brasiliensis and A. flocculosa completely lost virulence clusters, while another non-pathogenic member of the Ustilaginales, Pseudozyma hubeiensis, shows an almost complete set of effectors when compared to U. maydis (Figure 3).

Figure 3 with 3 supplements see all

Download asset Open asset

Genetic characterization of MbA

To perform reverse genetics in MbA, we established a genetic transformation system based on protoplast preparation and polyethylene glycol (PEG)-mediated DNA transfer. In preliminary transformation assays, we expressed a cytosolic GFP reporter-gene under control of the constitutive o2tef-Promoter (Figure 4A). For the generation of knockout strains, a split marker approach was used to avoid ectopic integrations (Figure 4B). To allow generation of multiple knockouts, we used a selection marker-recycling system (pFLPexpC) that allows selection marker excision at each transformation round (Khrunyk et al., 2010).

Figure 4

Download asset Open asset

We decided to apply the transformation system to study the MbA mating type loci in more detail. Although phylogenetically closely related to U. hordei, which has a bi-polar mating system, MbA owns a tetrapolar mating system whereby both mating type loci are physically not linked. This situation is similar to the mating type structure in the pathogenic smut U. maydis (Figure 4A). The a-locus, which encodes a pheromone receptor system that is required for sensing and fusion of compatible cells, is located on contig 6. The b-locus can be found on contig 1. This multiallelic mating locus contains two genes (b-East and b-West), which code for a pair of homeodomain transcription factors. Upon mating of compatible cells, pathogenic and sexual development are triggered by a heterodimeric bE/bW complex (Brefort et al., 2009). Since the MbA genome is completely equipped with mating type genes, we first deployed a screen for potential mating partners. To this end, we screened wild M. bullatus isolates to find a suitable mating partner, but we could not observe any mating event (Figure 5—figure supplement 1). To test if MbA is able to undergo pathogenic differentiation in the absence of mating, we generated a self-compatible strain (CB1) that carries compatible b-mating alleles: to construct the CB1 strain, we used compatible alleles of the b-East and b-West genes of the barley smut U. hordei, a pathogen that is the phylogenetically most closely related to MbA and amenable to reverse genetics. The native MbA locus was replaced by the compatible U. hordei b-East and b-West gene alleles via homologous recombination (Figure 5B).

Figure 5 with 1 supplement see all

Download asset Open asset

Figure 5—source data 1

Appressoria quantification of MbA.

https://cdn.elifesciences.org/articles/65306/elife-65306-fig5-data1-v2.xlsx

Download elife-65306-fig5-data1-v2.xlsx

Incubation of the MbA CB1 on charcoal plates led to the formation of aerial hyphae with the characteristic fluffy phenotype of filamentous strains like the self-compatible, solopathogenic U. maydis SG200 strain (Figure 5C). A second established method to induce filament formation in smuts is on hydrophobic parafilm (Mendoza-Mendoza et al., 2009). Quantification after 18 hr incubation of MbA CB1 on parafilm resulted in the formation of filaments comparable to those of the U. maydis SG200 strain (Figure 5D). While about 17% of MbA wild-type cells showed filaments, the CB1 strain with compatible b-genes showed 38% filamentous growth.

Formation of appressoria is a hallmark of pathogenic development in smut fungi (Mendoza-Mendoza et al., 2009). While the switch from yeast-like growth to filamentous development is the first step in the pathogenic development of smut fungi, host penetration is accompanied by the formation of a terminal swelling of infectious hyphae, termed ‘appressoria’. Induction of appressoria formation in vitro can be induced by adding 100 µM of the cutin monomer 16-hydroxyhexadecanoic acid (HDD) to the fungal cells prior to cell spraying onto a hydrophobic surface (Mendoza-Mendoza et al., 2009). In the absence of HDD, only about 8% of the U. maydis SG200 cells and 14% of the MbA cells formed appressoria on parafilm 24 hr after spraying (Figure 5E). Addition of 100 µM HDD resulted in a significant induction of appressoria in both U. maydis and MbA, demonstrating that MbA does hold the genetic repertoire to form infection structures in vitro. Together, the analysis of the recombinant CB1 strain indicates that MbA can sense pathogenesis-related surface cues and produce penetration structures to a similar level as that seen for the pathogenic model organism U. maydis.

Identification of microbe–microbe effector genes by RNA-Seq

To study the transcriptomic response of MbA to different biotic interactions, RNA sequencing was performed. The MbA transcriptome was profiled in five different conditions (Figure 6A; cells in axenic culture versus cells on-planta, on-planta + SynCom, on-planta + A. laibachii, on-planta + SynCom + A. laibachii). Inoculations of A. thaliana leaves were performed as described above for A. laibachii infection assays (Figure 6A). For MbA RNA preparation, the epiphytic microbes were peeled from the plant tissue by using liquid latex (see Materials and methods section for details).

Figure 6 with 2 supplements see all

Download asset Open asset

Figure 6—source data 1

Gene expression data of candidate effector genes.

https://cdn.elifesciences.org/articles/65306/elife-65306-fig6-data1-v2.xlsx

Download elife-65306-fig6-data1-v2.xlsx

The libraries of the 15 samples (five conditions in three biological replicates each) were generated by using a poly-A enrichment and sequenced on an Illumina HiSeq4000 platform. The paired end reads were mapped to the MbA genome by using Tophat2 (Kim et al., 2013). The analysis revealed that MbA cells on A. thaliana leaves (on-planta) downregulated 1300 and upregulated 1580 genes compared to cells in axenic culture (Figure 6B, Supplementary file 2). Differentially expressed genes were determined with the ‘limma’-package in R on ‘voom’ (Figure 6—figure supplement 1) using a False discovery rate threshold of 0.05 and log2FC > 0. A gene ontology (GO) terms analysis revealed that, among the downregulated genes, 50% were associated with primary metabolism (Figure 6—figure supplement 2). In the two conditions in which A. laibachii was present, we observed upregulation of 801 genes. Among these genes, 411 genes were specific to co-incubation of MbA with A. laibachii and SynCom while 174 were specific to incubation with A. laibachii only. A set of 216 genes was shared in both conditions (Figure 6B).

In the presence of A. laibachii, mainly metabolism- and translation-dependent genes were upregulated, which might indicate that MbA can access a new nutrient source in the presence of A. laibachii (Figure 6—figure supplement 2). Among all A. laibachii-induced MbA genes, 18 genes encode proteins carrying a secretion signal peptide and having no predicted transmembrane domain (Figure 6C). After excluding proteins being predicted to be located in intracellular organelles, nine candidate genes remained as potential microbe–microbe-dependent effectors, i.e. MbA genes that are induced by A. laibachii, show no or low expression in axenic culture, and encode for putative secreted proteins (Figure 6C). Interestingly, four of these genes encode putative glycoside hydrolases. Furthermore, two genes encode putative peptidases, one gene likely encodes an alkaline phosphatase and two encode uncharacterized proteins (Figure 6C).

To directly test the eventual antagonistic function of those genes toward A. laibachii, we selected the two predicted glycoside hydrolases-encoding genes g5 and g2490 (GH43 and GH25) and the gene encoding the uncharacterized protein g5755 for gene deletion in MbA. The respective mutant strains were tested in stress assays to assess, whether the gene deletions resulted in general growth defects. Wild-type and mutant MbA strains were exposed to different stress conditions including osmotic stress (sorbitol, NaCl), cell wall stress (calcofluor, Congo-red), and oxidative stress (H₂O₂). Overall, in none of the tested conditions we observed a growth defect of the deletion mutants in comparison to wild-type MbA (Figure 7—figure supplement 1). To test an eventual impact of the deleted genes in the antagonism of the two microbes, the MbA deletion strains were each pre-inoculated on A. thaliana leaves prior to A. laibachii infection. Deletion of g5 resulted in a significant but yet marginal increase of A. laibachii disease symptoms, while deletion of g5755 had no effect on A. laibachii. We therefore considered these two genes being not important for the antagonism of MbA toward A. laibachii. Strikingly, the MbA Δg2490 strain almost completely lost its biocontrol activity toward A. laibachii. This phenotype was reproduced by two independent g2490 deletion strains (Figure 7A). To check if this dramatic loss of microbial antagonism is specific to the deletion of g2490, in-locus genetic complementation of strain Δg2490_1 was performed via homologous recombination. The resulting strain MbA Δg2490/compl regained the ability to suppress A. laibachii infection, confirming that the observed phenotype specifically resulted from the deletion of the g2490 gene (Figure 7B). Together, these results demonstrate that the biocontrol of the pathogenic oomycete A. laibachii by the basidiomycete yeast MbA is determined by the secretion of a previously uncharacterized GH25 enzyme, which is transcriptionally activated specifically when both microbes are co-colonizing the A. thaliana leaf surface.

Figure 7 with 5 supplements see all

Download asset Open asset

Figure 7—source data 1

Infection data of A. laibachii on A. thaliana.

https://cdn.elifesciences.org/articles/65306/elife-65306-fig7-data1-v2.xlsx

Download elife-65306-fig7-data1-v2.xlsx

Healthy plants in natural habitats are extensively colonized by microbes, therefore it has been hypothesized that the immune system and the microbiota may instruct each other beyond the simple co-evolutionary arms race between plants and pathogens (Vannier et al., 2019). Community members as individuals or in a community context have been reported to confer extended immune functions to their plant host. Root endophytic bacteria for example were found to protect A. thaliana and stabilize the microbial community by competing with filamentous eukaryotes (Durán et al., 2018). A large inhibitory interaction network was found in the leaf microbiome of A. thaliana and genome mining was used to identify over 1000 predicted natural product biosynthetic gene clusters (BGCs) (Helfrich et al., 2018). In addition, the bacterium Brevibacillus sp. leaf 182 isolate was found to inhibit half of the 200 strains isolated from A. thaliana phyllosphere. Further analysis revealed that Brevibacillus sp. leaf 182 produces a trans-acyltransferase polyketide synthase-derived antibiotic, macrobrevin along with other putative polyketide synthases (Helfrich et al., 2018).

In this study, we describe the role of the basidiomycete yeast MbA, which we previously co-isolated with the oomycete pathogen A. laibachii and now characterized as an antagonistic driver in the A. thaliana phyllosphere. A. laibachii inhibits in vitro growth of seven members of a bacterial leaf SynCom and, most strikingly, strongly suppresses disease progression and reproduction of the pathogenic oomycete A. laibachii on A. thaliana. MbA is a member of the Ustilaginales, which had previously been classified into the group of pathogenic smut fungi of the Moesziomyces bullatus species (Kruse et al., 2017). Our genome analysis identified the anamorphic yeasts M. rugulosus, M. aphidis, and M. antarcticus, which had previously been classified as ‘Pseudozyma sp.’, as the closest relatives of MbA. Anamorphic Ustilaginales yeasts are long known and have been used for biotechnological applications and also biocontrol (Boekhout, 2011). Mannosylerythritol lipids produced by M. antarcticus are known to act as biosurfactants and are of great interest for pharmaceutical applications (Kitamoto et al., 1990; Morita et al., 2007). Glycolipids like flocculosin produced by A. flocculosa or ustilagic acid characterized in the smut fungus U. maydis have antifungal activity. Those compounds destabilize the membrane of different fungi and thus serve as biocontrol agents against powdery mildews or gray mold (Cheng et al., 2003; Mimee et al., 2005; Teichmann et al., 2007).

We identified 13 potential secondary metabolite gene clusters in MbA, including non-ribosomal peptide synthases and polyketide synthetase. Interaction among microbes within the same habitat is believed to have given rise to a variety of secondary metabolites (Schroeckh et al., 2009; Rutledge and Challis, 2015). The presence of Streptomyces rapamycinicus was shown to activate an otherwise silent polyketide synthase gene cluster, fgnA, in Aspergillus fumigatus. The resultant compound proved to be a potent fungal metabolite that inhibited the germination of S. rapamycinicus spores (Stroe et al., 2020). Therefore, secondary metabolite gene clusters and their corresponding products may confer a competitive advantage to fungi over the bacteria that reside in the same environment.

What is still under investigation is the relation of anamorphic yeasts with the related pathogenic smuts. Many smut fungi including the model species U. maydis are dimorphic organisms. In their saprophytic phase they grow as haploid non-pathogenic yeast cells. Only on appropriate host surfaces, haploid cells switch to filamentous growth and expression of pathogenicity-related genes is only activated upon mating in the filamentous dikaryon. A prime prerequisite for pathogenic development is therefore the ability of mating (Bölker, 2001; Nadal and Gold, 2010). Our genome analysis identified a tetrapolar mating system with a complete set of mating genes in MbA. Looking more closely on the phylogeny of different mating genes it appears that all sequenced Moesziomyces strains have the same pheromone receptor type (Figure 7—figure supplement 3). Together with our unsuccessful mating assays, this suggests that all sequenced strains of this species have the same mating type and, therefore, are unable to mate. Mating type bias after spore germination was reported for Ustilago bromivora, which leads to a haplo-lethal allele linked to the MAT-2 locus (Rabe et al., 2016). In this case, an intratetrad mating event rescues pathogenicity in nature as the second mating partner is not viable after spore germination. Together with the observation that anamorphic Moesziomyces yeasts are ubiquitous in nature, one could hypothesize that these fungi are highly competitive in their haploid form and antagonism might have led to the selection of one viable mating type. This eventually adapted to the epiphytic life style. At the same time, one should not dismiss the possibility that in the presence of a viable mating partner and suitable host surface, the yeasts can undergo potential sexual reproduction and thereby revitalizing the gene pool from time to time.

Transcriptome analysis showed that epiphytic growth of MbA on A. thaliana leads to massive transcriptional changes particularly in primary metabolism, which might reflect adaptation to the nutritional situation on the plant surface. Moreover, MbA showed specific transcriptional responses to a bacterial community, as well as to A. laibachii when being co-inoculated on plant leaves. The presence of A. laibachii resulted in the induction of primary metabolism and biosynthesis pathways, which might reflect enhanced growth of MbA in the presence of A. laibachii.

A set of MbA genes encoding secreted hydrolases was induced by A. laibachii and one of these genes which encodes a putative GH25 hydrolase with similarity to Chalaropsis type lysozymes appeared to be essential for the biocontrol of A. laibachii. Initially discovered in the fungi Chalaropsis sp., this group of proteins is largely present in bacteria as well as phages, for example the germination-specific muramidase from Clostridium perfringens S40 (Chen et al., 1997). The bacterial muramidase, cellosyl from Streptomyces coelicolor (Rau et al., 2001), also belongs to the Chalaropsis type of lysozyme. These proteins are proposed to cleave the β-1,4-glycosidic bond between N-acetylmuramic acid (NAM) and N-acetylglucosamine (NAG) in the bacterial peptidoglycan. Specifically, the β-1,4 N,6-O-diacetylmuramidase activity allows the Chalaropsis type lysozyme to degrade the cell wall of Staphylococcus aureus, in contrast to the commercially available HEWL (Rau et al., 2001). Despite differences in structure and molecular weight from HEWL, the GH25 of MbA has lysozyme activity against the gram positive bacterium Micrococcus lysodeikticus in a fluorogenic assay. This highlights the overall biochemical functionality of the recombinant glycoside hydrolase. The glycoside Hydrolase 25 family is predicted to have an active site motif DXE that is highly conserved across the fungal kingdom (Figure 7—figure supplement 4). The structure of glycoside hydrolase family 25 from Aspergillus fumigates was characterized and the presence of N-terminal signal peptide was considered to indicate an extracellular secretion of the protein with possible antimicrobial properties (Korczynska et al., 2010). The role of the secreted hydrolase in the fungal kingdom is not completely explored yet. The presence of such hydrolases has in many cases been hypothesized to be associated with hyperparasitism of fungi parasitizing fungi (Hyde et al., 2019) or oomycetes parasitizing oomycetes (Horner et al., 2012). Our results might therefore indicate a cross kingdom hyperparasitism event between a fungus and an oomycete. Previous work on microbial communities has indicated that negative interactions stabilize microbial communities. Hyperparasitism is such a negative interaction with a strong eco-evolutionary effect on pathogen–host interactions and therefore on community stability (Parratt and Laine, 2016). MbA might therefore regulate A. laibachii infection and reduce disease severity. The qPCR evaluation of oomycete biomass strongly points toward the idea that A. laibachii is a direct target of antagonism for MBA. Since we observed reduced formation of A. laibachii in the presence of MbA, we also tested if the GH25 lysozyme would suppress zoospore germination. However, we could not detect a significant reduction of A. laibachii zoosporangia germination upon treatment with active GH25 lysozyme (Figure 7—figure supplement 5), suggesting that the GH25 lysozyme interferes with A. laibachii at a later stage of infection. As A. laibachii has been shown to reduce microbial diversity (Agler et al., 2016), MbA might increase diversity through hyperparasitism of A. laibachii. At the same time this increased diversity might have caused the need for more secondary metabolites to evolve in the MbA genome to defend against niche competitors. Through its close association with A. laibachii, MbA could be a key regulator of the A. thaliana microbial diversity and therefore relevant for plant health beyond the regulation of A. laibachii infection.

In conclusion, the secreted hydrolase we identified as a main factor of A. laibachii inhibition has great potential to act as antimicrobial agent. The isolated compound is not only valuable per se in an ecological context. It can further lay the grounds for exploring other microbial bioactive compounds that mediate inter-species and inter-kingdom crosstalk. A main goal of our future studies will be to understand on the mechanistic level, how the GH-25 suppresses A. laibachii, and at which developmental step the oomycete infection is blocked. It will be particularly interesting to elucidate if and how the GH25 enzyme activity directly interferes with the A. laibachii cell wall. While the canonical lysozyme substrate is found in bacterial cell walls, a detailed biochemical characterization of substrate specificity will be required to pinpoint potential target sites in the oomycete cell wall. Also, to the best of our knowledge there is no detailed information on the A. laibachii cell wall structure and composition. One could speculate if GH25 activity directly affects cell wall integrity of A. laibachii, or if a modification of the cell wall structure interferes with pathogenic development, e.g. by interfering with cellular differentiation, blocking signal perception, or by triggering a host defense response.

Since the GH-25 enzyme is well conserved among Ustilaginales including pathogenic species, it will also be tempting to elucidate whether the species-specific antagonism identified here is broadly conserved among Ustilaginales fungi and oomycetes. We further will investigate potential responses by the host plant and how this impacts A. laibachii growth upon MbA colonization. Functional investigation of these interactions can provide meaningful insights as to why certain yeasts prefer to colonize specific environments. At the same time, it will be worth exploring how the basidiomycete yeasts influence the bacterial major colonizers of the phyllosphere.

Genome information and RNA sequencing have been submitted to NCBI Genbank and are available under the following links: https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE148670.

1. 1. Eitzen K
   2. Doehlemann G
   3. Kemen E

   (2021) NCBI Gene Expression Omnibus

   ID GSE148670. Transcriptome of Moesziomyces albugensis in response to different biotic factors (A. laibachii & SynCom) on A. thaliana leaves.

   https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE148670

1. 1. Agler MT
   2. Ruhe J
   3. Kroll S
   4. Morhenn C
   5. Kim ST
   6. Weigel D
   7. Kemen EM

   (2016) Microbial hub taxa link host and abiotic factors to plant microbiome variation

   PLOS Biology 14:e1002352.

   https://doi.org/10.1371/journal.pbio.1002352

   • PubMed
   • Google Scholar
2. 1. Barda O
   2. Shalev O
   3. Alster S
   4. Buxdorf K
   5. Gafni A
   6. Levy M

   (2015) Pseudozyma aphidis induces Salicylic-Acid-Independent resistance to Clavibacter michiganensis in tomato plants

   Plant Disease 99:621–626.

   https://doi.org/10.1094/PDIS-04-14-0377-RE

   • PubMed
   • Google Scholar
3. 1. Boekhout T

   (2011) Pseudozyma bandoni emend: boekhout (1985) and a comparison with the yeast state of Ustilago maydis (De candolle) Corda (1842)

   The Yeasts 2011:1857–1868.

   https://doi.org/10.1016/B978-0-444-52149-1.00153-1

   • Google Scholar
4. 1. Bölker M
   2. Urban M
   3. Kahmann R

   (1992) The a mating type locus of U. maydis specifies cell signaling components

   Cell 68:441–450.

   https://doi.org/10.1016/0092-8674(92)90182-C

   • PubMed
   • Google Scholar
5. 1. Bölker M

   (2001) Ustilago maydis--a valuable model system for the study of fungal dimorphism and virulence

   Microbiology 147:1395–1401.

   https://doi.org/10.1099/00221287-147-6-1395

   • PubMed
   • Google Scholar
6. 1. Brefort T
   2. Doehlemann G
   3. Mendoza-Mendoza A
   4. Reissmann S
   5. Djamei A
   6. Kahmann R

   (2009) Ustilago maydis as a pathogen

   Annual Review of Phytopathology 47:423–445.

   https://doi.org/10.1146/annurev-phyto-080508-081923

   • PubMed
   • Google Scholar
7. 1. Brefort T
   2. Tanaka S
   3. Neidig N
   4. Doehlemann G
   5. Vincon V
   6. Kahmann R

   (2014) Characterization of the largest effector gene cluster of Ustilago maydis

   PLOS Pathogens 10:e1003866.

   https://doi.org/10.1371/journal.ppat.1003866

   • PubMed
   • Google Scholar
8. 1. Bulgarelli D
   2. Schlaeppi K
   3. Spaepen S
   4. Ver Loren van Themaat E
   5. Schulze-Lefert P

   (2013) Structure and functions of the bacterial Microbiota of plants

   Annual Review of Plant Biology 64:807–838.

   https://doi.org/10.1146/annurev-arplant-050312-120106

   • PubMed
   • Google Scholar
9. 1. Busby PE
   2. Peay KG
   3. Newcombe G

   (2016) Common foliar fungi of Populus trichocarpa modify Melampsora rust disease severity

   New Phytologist 209:1681–1692.

   https://doi.org/10.1111/nph.13742

   • Google Scholar
10. 1. Chen Y
    2. Miyata S
    3. Makino S
    4. Moriyama R

    (1997) Molecular characterization of a germination-specific muramidase from Clostridium perfringens S40 spores and nucleotide sequence of the corresponding gene

    Journal of Bacteriology 179:3181–3187.

    https://doi.org/10.1128/JB.179.10.3181-3187.1997

    • Google Scholar
11. 1. Cheng Y
    2. McNally DJ
    3. Labbé C
    4. Voyer N
    5. Belzile F
    6. Bélanger RR

    (2003) Insertional mutagenesis of a fungal biocontrol agent led to discovery of a rare cellobiose lipid with antifungal activity

    Applied and Environmental Microbiology 69:2595–2602.

    https://doi.org/10.1128/AEM.69.5.2595-2602.2003

    • PubMed
    • Google Scholar
12. 1. Coyte KZ
    2. Schluter J
    3. Foster KR

    (2015) The ecology of the microbiome: networks, competition, and stability

    Science 350:663–666.

    https://doi.org/10.1126/science.aad2602

    • PubMed
    • Google Scholar
13. 1. Djamei A
    2. Schipper K
    3. Rabe F
    4. Ghosh A
    5. Vincon V
    6. Kahnt J
    7. Osorio S
    8. Tohge T
    9. Fernie AR
    10. Feussner I
    11. Feussner K
    12. Meinicke P
    13. Stierhof YD
    14. Schwarz H
    15. Macek B
    16. Mann M
    17. Kahmann R

    (2011) Metabolic priming by a secreted fungal effector

    Nature 478:395–398.

    https://doi.org/10.1038/nature10454

    • PubMed
    • Google Scholar
14. 1. Doehlemann G
    2. van der Linde K
    3. Assmann D
    4. Schwammbach D
    5. Hof A
    6. Mohanty A
    7. Jackson D
    8. Kahmann R

    (2009) Pep1, a secreted effector protein of Ustilago maydis, is required for successful invasion of plant cells

    PLOS Pathogens 5:e1000290.

    https://doi.org/10.1371/journal.ppat.1000290

    • PubMed
    • Google Scholar
15. 1. Doehlemann G
    2. Reissmann S
    3. Assmann D
    4. Fleckenstein M
    5. Kahmann R

    (2011) Two linked genes encoding a secreted effector and a membrane protein are essential for Ustilago maydis-induced tumour formation

    Molecular Microbiology 81:751–766.

    https://doi.org/10.1111/j.1365-2958.2011.07728.x

    • PubMed
    • Google Scholar
16. 1. Durán P
    2. Thiergart T
    3. Garrido-Oter R
    4. Agler M
    5. Kemen E
    6. Schulze-Lefert P
    7. Hacquard S

    (2018) Microbial interkingdom interactions in roots promote Arabidopsis survival

    Cell 175:973–983.

    https://doi.org/10.1016/j.cell.2018.10.020

    • PubMed
    • Google Scholar
17. 1. Dutheil JY
    2. Mannhaupt G
    3. Schweizer G
    4. M K Sieber C
    5. Münsterkötter M
    6. Güldener U
    7. Schirawski J
    8. Kahmann R

    (2016) A tale of genome compartmentalization: the evolution of virulence clusters in smut fungi

    Genome Biology and Evolution 8:681–704.

    https://doi.org/10.1093/gbe/evw026

    • PubMed
    • Google Scholar
18. 1. Gafni A
    2. Calderon CE
    3. Harris R
    4. Buxdorf K
    5. Dafa-Berger A
    6. Zeilinger-Reichert E
    7. Levy M

    (2015) Biological control of the cucurbit powdery mildew pathogen Podosphaera xanthii by means of the epiphytic fungus Pseudozyma aphidis and parasitism as a mode of action

    Frontiers in Plant Science 6:132.

    https://doi.org/10.3389/fpls.2015.00132

    • PubMed
    • Google Scholar
19. 1. Helfrich EJN
    2. Vogel CM
    3. Ueoka R
    4. Schäfer M
    5. Ryffel F
    6. Müller DB
    7. Probst S
    8. Kreuzer M
    9. Piel J
    10. Vorholt JA

    (2018) Bipartite interactions, antibiotic production and biosynthetic potential of the Arabidopsis leaf microbiome

    Nature Microbiology 3:909–919.

    https://doi.org/10.1038/s41564-018-0200-0

    • PubMed
    • Google Scholar
20. 1. Hemetsberger C
    2. Mueller AN
    3. Matei A
    4. Herrberger C
    5. Hensel G
    6. Kumlehn J
    7. Mishra B
    8. Sharma R
    9. Thines M
    10. Hückelhoven R
    11. Doehlemann G

    (2015) The fungal core effector Pep1 is conserved across smuts of dicots and monocots

    New Phytologist 206:1116–1126.

    https://doi.org/10.1111/nph.13304

    • Google Scholar
21. 1. Horner NR
    2. Grenville-Briggs LJ
    3. van West P

    (2012) The oomycete Pythium oligandrum expresses putative effectors during mycoparasitism of Phytophthora infestans and is amenable to transformation

    Fungal Biology 116:24–41.

    https://doi.org/10.1016/j.funbio.2011.09.004

    • Google Scholar
22. 1. Hyde KD
    2. Xu J
    3. Rapior S
    4. Jeewon R
    5. Lumyong S
    6. Niego AGT
    7. Abeywickrama PD
    8. Aluthmuhandiram JVS
    9. Brahamanage RS
    10. Brooks S
    11. Chaiyasen A
    12. Chethana KWT
    13. Chomnunti P
    14. Chepkirui C
    15. Chuankid B
    16. de Silva NI
    17. Doilom M
    18. Faulds C
    19. Gentekaki E
    20. Gopalan V
    21. Kakumyan P
    22. Harishchandra D
    23. Hemachandran H
    24. Hongsanan S
    25. Karunarathna A
    26. Karunarathna SC
    27. Khan S
    28. Kumla J
    29. Jayawardena RS
    30. Liu J-K
    31. Liu N
    32. Luangharn T
    33. Macabeo APG
    34. Marasinghe DS
    35. Meeks D
    36. Mortimer PE
    37. Mueller P
    38. Nadir S
    39. Nataraja KN
    40. Nontachaiyapoom S
    41. O’Brien M
    42. Penkhrue W
    43. Phukhamsakda C
    44. Ramanan US
    45. Rathnayaka AR
    46. Sadaba RB
    47. Sandargo B
    48. Samarakoon BC
    49. Tennakoon DS
    50. Siva R
    51. Sriprom W
    52. Suryanarayanan TS
    53. Sujarit K
    54. Suwannarach N
    55. Suwunwong T
    56. Thongbai B
    57. Thongklang N
    58. Wei D
    59. Wijesinghe SN
    60. Winiski J
    61. Yan J
    62. Yasanthika E
    63. Stadler M

    (2019) The amazing potential of fungi: 50 ways we can exploit fungi industrially

    Fungal Diversity 97:1–136.

    https://doi.org/10.1007/s13225-019-00430-9

    • Google Scholar
23. 1. Kämper J

    (2004) A PCR-based system for highly efficient generation of gene replacement mutants in Ustilago maydis

    Molecular Genetics and Genomics 271:103–110.

    https://doi.org/10.1007/s00438-003-0962-8

    • PubMed
    • Google Scholar
24. 1. Kämper J
    2. Kahmann R
    3. Bölker M
    4. Ma LJ
    5. Brefort T
    6. Saville BJ
    7. Banuett F
    8. Kronstad JW
    9. Gold SE
    10. Müller O
    11. Perlin MH
    12. Wösten HA
    13. de Vries R
    14. Ruiz-Herrera J
    15. Reynaga-Peña CG
    16. Snetselaar K
    17. McCann M
    18. Pérez-Martín J
    19. Feldbrügge M
    20. Basse CW
    21. Steinberg G
    22. Ibeas JI
    23. Holloman W
    24. Guzman P
    25. Farman M
    26. Stajich JE
    27. Sentandreu R
    28. González-Prieto JM
    29. Kennell JC
    30. Molina L
    31. Schirawski J
    32. Mendoza-Mendoza A
    33. Greilinger D
    34. Münch K
    35. Rössel N
    36. Scherer M
    37. Vranes M
    38. Ladendorf O
    39. Vincon V
    40. Fuchs U
    41. Sandrock B
    42. Meng S
    43. Ho EC
    44. Cahill MJ
    45. Boyce KJ
    46. Klose J
    47. Klosterman SJ
    48. Deelstra HJ
    49. Ortiz-Castellanos L
    50. Li W
    51. Sanchez-Alonso P
    52. Schreier PH
    53. Häuser-Hahn I
    54. Vaupel M
    55. Koopmann E
    56. Friedrich G
    57. Voss H
    58. Schlüter T
    59. Margolis J
    60. Platt D
    61. Swimmer C
    62. Gnirke A
    63. Chen F
    64. Vysotskaia V
    65. Mannhaupt G
    66. Güldener U
    67. Münsterkötter M
    68. Haase D
    69. Oesterheld M
    70. Mewes HW
    71. Mauceli EW
    72. DeCaprio D
    73. Wade CM
    74. Butler J
    75. Young S
    76. Jaffe DB
    77. Calvo S
    78. Nusbaum C
    79. Galagan J
    80. Birren BW

    (2006) Insights from the genome of the biotrophic fungal plant pathogen Ustilago maydis

    Nature 444:97–101.

    https://doi.org/10.1038/nature05248

    • PubMed
    • Google Scholar
25. 1. Khrunyk Y
    2. Münch K
    3. Schipper K
    4. Lupas AN
    5. Kahmann R

    (2010) The use of FLP-mediated recombination for the functional analysis of an effector gene family in the biotrophic smut fungus Ustilago maydis

    New Phytologist 187:957–968.

    https://doi.org/10.1111/j.1469-8137.2010.03413.x

    • Google Scholar
26. 1. Kim D
    2. Pertea G
    3. Trapnell C
    4. Pimentel H
    5. Kelley R
    6. Salzberg SL

    (2013) TopHat2: accurate alignment of transcriptomes in the presence of insertions, deletions and gene fusions

    Genome Biology 14:R36.

    https://doi.org/10.1186/gb-2013-14-4-r36

    • PubMed
    • Google Scholar
27. 1. Kitamoto D
    2. Akiba S
    3. Hiok C
    4. Tabuchi T

    (1990) Extracellular accumulation of mannosylerythritol lipids by a strain of candida Antarctica

    Agricultural and Biological Chemistry 54:31–36.

    https://doi.org/10.1271/bbb1961.54.31

    • Google Scholar
28. 1. Korczynska JE
    2. Danielsen S
    3. Schagerlöf U
    4. Turkenburg JP
    5. Davies GJ
    6. Wilson KS
    7. Taylor EJ

    (2010) The structure of a family GH25 lysozyme from Aspergillus fumigatus

    Acta Crystallographica. Section F, Structural Biology and Crystallization Communications 66:973–977.

    https://doi.org/10.1107/S1744309110025601

    • PubMed
    • Google Scholar
29. 1. Krombach S
    2. Reissmann S
    3. Kreibich S
    4. Bochen F
    5. Kahmann R

    (2018) Virulence function of the Ustilago maydis sterol carrier protein 2

    New Phytologist 220:553–566.

    https://doi.org/10.1111/nph.15268

    • Google Scholar
30. 1. Kruse J
    2. Doehlemann G
    3. Kemen E
    4. Thines M

    (2017) Asexual and sexual morphs of Moesziomyces revisited

    IMA Fungus 8:117–129.

    https://doi.org/10.5598/imafungus.2017.08.01.09

    • PubMed
    • Google Scholar
31. 1. Lee G
    2. Lee SH
    3. Kim KM
    4. Ryu CM

    (2017) Foliar application of the leaf-colonizing yeast Pseudozyma churashimaensis elicits systemic defense of pepper against bacterial and viral pathogens

    Scientific Reports 7:39432.

    https://doi.org/10.1038/srep39432

    • PubMed
    • Google Scholar
32. 1. Lefebvre F
    2. Joly DL
    3. Labbé C
    4. Teichmann B
    5. Linning R
    6. Belzile F
    7. Bakkeren G
    8. Bélanger RR

    (2013) The transition from a phytopathogenic smut ancestor to an anamorphic biocontrol agent deciphered by comparative whole-genome analysis

    The Plant Cell 25:1946–1959.

    https://doi.org/10.1105/tpc.113.113969

    • PubMed
    • Google Scholar
33. 1. Ma LS
    2. Wang L
    3. Trippel C
    4. Mendoza-Mendoza A
    5. Ullmann S
    6. Moretti M
    7. Carsten A
    8. Kahnt J
    9. Reissmann S
    10. Zechmann B
    11. Bange G
    12. Kahmann R

    (2018) The Ustilago maydis repetitive effector Rsp3 blocks the antifungal activity of mannose-binding maize proteins

    Nature Communications 9:1711.

    https://doi.org/10.1038/s41467-018-04149-0

    • PubMed
    • Google Scholar
34. 1. Mendoza-Mendoza A
    2. Berndt P
    3. Djamei A
    4. Weise C
    5. Linne U
    6. Marahiel M
    7. Vranes M
    8. Kämper J
    9. Kahmann R

    (2009) Physical-chemical plant-derived signals induce differentiation in Ustilago maydis

    Molecular Microbiology 71:895–911.

    https://doi.org/10.1111/j.1365-2958.2008.06567.x

    • PubMed
    • Google Scholar
35. 1. Mikiciński A
    2. Sobiczewski P
    3. Puławska J
    4. Maciorowski R

    (2016) Control of fire blight (Erwinia amylovora) by a novel strain 49M of Pseudomonas graminis from the phyllosphere of apple (Malus spp.)

    European Journal of Plant Pathology 145:265–276.

    https://doi.org/10.1007/s10658-015-0837-y

    • Google Scholar
36. 1. Mimee B
    2. Labbé C
    3. Pelletier R
    4. Bélanger RR

    (2005) Antifungal activity of Flocculosin, a novel glycolipid isolated from Pseudozyma flocculosa

    Antimicrobial Agents and Chemotherapy 49:1597–1599.

    https://doi.org/10.1128/AAC.49.4.1597-1599.2005

    • Google Scholar
37. 1. Morita T
    2. Konishi M
    3. Fukuoka T
    4. Imura T
    5. Kitamoto D

    (2007) Microbial conversion of glycerol into glycolipid biosurfactants, mannosylerythritol lipids, by a basidiomycete yeast, Pseudozyma antarctica JCM 10317T

    Journal of Bioscience and Bioengineering 104:78–81.

    https://doi.org/10.1263/jbb.104.78

    • Google Scholar
38. 1. Nadal M
    2. Gold SE

    (2010) The autophagy genes ATG8 and ATG1 affect morphogenesis and pathogenicity in Ustilago maydis

    Molecular Plant Pathology 11:463–478.

    https://doi.org/10.1111/j.1364-3703.2010.00620.x

    • PubMed
    • Google Scholar
39. 1. Ökmen B
    2. Kemmerich B
    3. Hilbig D
    4. Wemhöner R
    5. Aschenbroich J
    6. Perrar A
    7. Huesgen PF
    8. Schipper K
    9. Doehlemann G

    (2018) Dual function of a secreted fungalysin metalloprotease in Ustilago maydis

    New Phytologist 220:249–261.

    https://doi.org/10.1111/nph.15265

    • Google Scholar
40. 1. Parratt SR
    2. Laine AL

    (2016) The role of hyperparasitism in microbial pathogen ecology and evolution

    The ISME Journal 10:1815–1822.

    https://doi.org/10.1038/ismej.2015.247

    • PubMed
    • Google Scholar
41. 1. Rabe F
    2. Bosch J
    3. Stirnberg A
    4. Guse T
    5. Bauer L
    6. Seitner D
    7. Rabanal FA
    8. Czedik-Eysenberg A
    9. Uhse S
    10. Bindics J
    11. Genenncher B
    12. Navarrete F
    13. Kellner R
    14. Ekker H
    15. Kumlehn J
    16. Vogel JP
    17. Gordon SP
    18. Marcel TC
    19. Münsterkötter M
    20. Walter MC
    21. Sieber CM
    22. Mannhaupt G
    23. Güldener U
    24. Kahmann R
    25. Djamei A

    (2016) A complete toolset for the study of Ustilago bromivora and Brachypodium sp. as a fungal-temperate grass pathosystem

    eLife 5:e20522.

    https://doi.org/10.7554/eLife.20522

    • PubMed
    • Google Scholar
42. 1. Rau A
    2. Hogg T
    3. Marquardt R
    4. Hilgenfeld R

    (2001) A new lysozyme fold. Crystal structure of the muramidase from Streptomyces coelicolor at 1.65 å resolution

    Journal of Biological Chemistry 276:31994–31999.

    https://doi.org/10.1074/jbc.M102591200

    • Google Scholar
43. 1. Redkar A
    2. Hoser R
    3. Schilling L
    4. Zechmann B
    5. Krzymowska M
    6. Walbot V
    7. Doehlemann G

    (2015) A secreted effector protein of Ustilago maydis guides maize leaf cells to form tumors

    The Plant Cell 27:1332–1351.

    https://doi.org/10.1105/tpc.114.131086

    • PubMed
    • Google Scholar
44. 1. Ritpitakphong U
    2. Falquet L
    3. Vimoltust A
    4. Berger A
    5. Métraux J-P
    6. L'Haridon F

    (2016) The microbiome of the leaf surface of Arabidopsis protects against a fungal pathogen

    New Phytologist 210:1033–1043.

    https://doi.org/10.1111/nph.13808

    • Google Scholar
45. 1. Ruhe J
    2. Agler MT
    3. Placzek A
    4. Kramer K
    5. Finkemeier I
    6. Kemen EM

    (2016) Obligate biotroph pathogens of the genus Albugo are better adapted to active host defense compared to niche competitors

    Frontiers in Plant Science 7:820.

    https://doi.org/10.3389/fpls.2016.00820

    • PubMed
    • Google Scholar
46. 1. Rutledge PJ
    2. Challis GL

    (2015) Discovery of microbial natural products by activation of silent biosynthetic gene clusters

    Nature Reviews Microbiology 13:509–523.

    https://doi.org/10.1038/nrmicro3496

    • PubMed
    • Google Scholar
47. 1. Schilling L
    2. Matei A
    3. Redkar A
    4. Walbot V
    5. Doehlemann G

    (2014) Virulence of the maize smut Ustilago maydis is shaped by organ-specific effectors

    Molecular Plant Pathology 15:780–789.

    https://doi.org/10.1111/mpp.12133

    • PubMed
    • Google Scholar
48. Thesis

    1. Schipper K

    (2009)

    Charakterisierung eines Ustilago maydis genclusters, 

    das f€ur drei neuartige sekretierte Effektoren kodiert PhDitle. PhD Thesis.

    • Google Scholar
49. 1. Schirawski J
    2. Mannhaupt G
    3. Münch K
    4. Brefort T
    5. Schipper K
    6. Doehlemann G
    7. Di Stasio M
    8. Rössel N
    9. Mendoza-Mendoza A
    10. Pester D
    11. Müller O
    12. Winterberg B
    13. Meyer E
    14. Ghareeb H
    15. Wollenberg T
    16. Münsterkötter M
    17. Wong P
    18. Walter M
    19. Stukenbrock E
    20. Güldener U
    21. Kahmann R

    (2010) Pathogenicity determinants in smut fungi revealed by genome comparison

    Science 330:1546–1548.

    https://doi.org/10.1126/science.1195330

    • PubMed
    • Google Scholar
50. 1. Schroeckh V
    2. Scherlach K
    3. Nützmann HW
    4. Shelest E
    5. Schmidt-Heck W
    6. Schuemann J
    7. Martin K
    8. Hertweck C
    9. Brakhage AA

    (2009) Intimate bacterial-fungal interaction triggers biosynthesis of archetypal polyketides in Aspergillus nidulans

    PNAS 106:14558–14563.

    https://doi.org/10.1073/pnas.0901870106

    • PubMed
    • Google Scholar
51. 1. Seitner D
    2. Uhse S
    3. Gallei M
    4. Djamei A

    (2018) The core effector Cce1 is required for early infection of maize by Ustilago maydis

    Molecular Plant Pathology 19:2277–2287.

    https://doi.org/10.1111/mpp.12698

    • PubMed
    • Google Scholar
52. 1. Sharma R
    2. Ökmen B
    3. Doehlemann G
    4. Thines M

    (2019) Saprotrophic yeasts formerly classified as Pseudozyma have retained a large effector arsenal, including functional Pep1 orthologs

    Mycological Progress 18:763–768.

    https://doi.org/10.1007/s11557-019-01486-2

    • Google Scholar
53. 1. Stanke M
    2. Steinkamp R
    3. Waack S
    4. Morgenstern B

    (2004) AUGUSTUS: a web server for gene finding in eukaryotes

    Nucleic Acids Research 32:W309–W312.

    https://doi.org/10.1093/nar/gkh379

    • PubMed
    • Google Scholar
54. Report

    1. Stone BWG
    2. Weingarten EA
    3. Jackson CR

    (2018)

    The Role of the Phyllosphere Microbiome in Plant Health and Function

    Annual Plant Reviews Online.

    • Google Scholar
55. 1. Stroe MC
    2. Netzker T
    3. Scherlach K
    4. Krüger T
    5. Hertweck C
    6. Valiante V
    7. Brakhage AA

    (2020) Targeted induction of a silent fungal gene cluster encoding the bacteria-specific germination inhibitor fumigermin

    eLife 9:e52541.

    https://doi.org/10.7554/eLife.52541

    • PubMed
    • Google Scholar
56. 1. Teichmann B
    2. Linne U
    3. Hewald S
    4. Marahiel MA
    5. Bölker M

    (2007) A biosynthetic gene cluster for a secreted cellobiose lipid with antifungal activity from Ustilago maydis

    Molecular Microbiology 66:525–533.

    https://doi.org/10.1111/j.1365-2958.2007.05941.x

    • PubMed
    • Google Scholar
57. 1. Vannier N
    2. Agler M
    3. Hacquard S

    (2019) Microbiota-mediated disease resistance in plants

    PLOS Pathogens 15:e1007740.

    https://doi.org/10.1371/journal.ppat.1007740

    • PubMed
    • Google Scholar
58. 1. Vorholt JA

    (2012) Microbial life in the phyllosphere

    Nature Reviews Microbiology 10:828–840.

    https://doi.org/10.1038/nrmicro2910

    • PubMed
    • Google Scholar
59. 1. Walker TS
    2. Bais HP
    3. Grotewold E
    4. Vivanco JM

    (2003) Root exudation and rhizosphere biology

    Plant Physiology 132:44–51.

    https://doi.org/10.1104/pp.102.019661

    • PubMed
    • Google Scholar
60. 1. Wang QM
    2. Begerow D
    3. Groenewald M
    4. Liu XZ
    5. Theelen B
    6. Bai FY
    7. Boekhout T

    (2015) Multigene phylogeny and taxonomic revision of yeasts and related fungi in the ustilaginomycotina

    Studies in Mycology 81:55–83.

    https://doi.org/10.1016/j.simyco.2015.10.004

    • PubMed
    • Google Scholar
61. 1. Zuo W
    2. Ökmen B
    3. Depotter JRL
    4. Ebert MK
    5. Redkar A
    6. Misas Villamil J
    7. Doehlemann G

    (2019) Molecular interactions between smut fungi and their host plants

    Annual Review of Phytopathology 57:411–430.

    https://doi.org/10.1146/annurev-phyto-082718-100139

    • PubMed
    • Google Scholar

Acceptance summary:

Understanding functions and mechanisms of microbe-microbe interactions in plant microbiota communities is of great fundamental interest, and promising for agricultural application for example in biocontrol. This study reports on a specific molecular player mediating antagonism by a leaf-surface dwelling yeast on an oblicate biotrophic oomycete pathogen. The yeast Moesziomyces bullatus ex Albugo (MbA) was isolated from the biotrophic pathogen oomycete Albugo laibachii, which acts as a hub organism in Arabidopsis phyllosphere microbiome. The study promotes the establishment of Mba as a leaf epiphytic model fungus, by providing a sequenced genome, the transcriptome in response to interaction with a phyllosphere bacterial SynCom and A. laibachii and by establishing a transformation system for generation of k. o. strains by homologous recombination. The authors report that MbA antagonizes the oomycete pathogen A. laibachii, via a putatively secreted GH25 hydrolase with lysozyme activity. This important finding provides novel insights into the yet elusive molecular mechanisms involved in shaping the microbiome community on leaf surfaces through microbe-microbe interactions.

Decision letter after peer review:

[Editors’ note: the authors submitted for reconsideration following the decision after peer review. What follows is the decision letter after the first round of review.]

Thank you for submitting your work entitled "A fungal member of the Arabidopsis thaliana phyllosphere antagonizes Albugo laibachii via a putative secreted hydrolase" for consideration by eLife. Your article has been reviewed by three peer reviewers, and the evaluation has been overseen by a Reviewing Editor and a Senior Editor, and three reviewers. The reviewers have opted to remain anonymous.

Our decision has been reached after consultation between the reviewers. Based on these discussions and the individual reviews below, we regret to inform you that your work will not be considered further for publication in eLife in its present form.

Although the reviewers and ourself found the study original and the establishment of Moesziomyces albugensis as a genetic model as well as its characterization interesting, the reviewers detected several flaws, which prevent publication in eLife:

– the phylogenetic placing of "Moesziomyces albugensis" appears incorrect

– its mode of action on Albugo is not described and

– the GH25 enzyme and its function are not sufficiently characterized

We would be happy to consider an improved version of the manuscript, in which the reviewer comments have been addressed comprehensively; and which provides a better understanding of Moesziomyces action on Albugo and GH25 function.

Reviewer #1:

The authors establish Moesziomyces albugensis as a new "model fungus" at the leaf surface, which may act as a parasite on the pathogenic oomycete of Arabidopsis Albugo laibachii. They provide a sequenced genome, the transcriptome in response to interaction with the plant, phyllosphere bacteria and the oomycete A. laibachii, a transformation system for generation of k. o. strains by homologous recombination and characterize the mating type as well as filament and appressoria forming behavior in vitro.

In addition, they also report some interesting biology and show that M. albugensis antagonizes the oomycete pathogen A. laibachii, probably via a secreted putative GH25 hydrolase. This finding and other findings, that may be expected from this research line in the future, may yield biocontrol fungi to combat agriculturally relevant oomycete infections of crop leaves.

While M.albugensis is an exciting new fungus and the identification of the effector required for parasitism of A. laibachii is very interesting, the manuscript still suffers from a few unclarities.

– From the text I did not understand whether M. albugensis lives only epiphytically or can also colonize the plant tissue (it forms appressoria on hydrophobic surfaces) or whether it is exclusively associated with A. laibachii. This should be clarified.

– It is unclear how the bacterial SynCom was constructed and which isolates it contains.

– The quantification of A. laibachii infection seems kind of rough and prone to subjectivity. An additional, more quantitative method should be used for quantification, for example by qPCR using an amplicon from A. laibachii DNA normalized to a plant amplicon.

– It is unclear whether GH25 hydrolase suppresses infection of plant leaves by A. laibachii or whether it helps M. albugensis to parasitize A. laibachii by other means. The authors should examine the effect of isolated GH25 hydrolase on A. laibachii infection.

– At which stage of infection is A. laibachii blocked by GH25 hydrolase or M. albugensis?

Reviewer #2:

The study by Eitzen et al., characterizes a yeast species isolated from Arabidopsis phyllosphere classified as Moesziomyces albugensis (Ma). The report provides phylogenetic classification, genome analysis and functional analysis of Ma inhibitory activity towards the white rust oomycete Albugo laibachii (Al). Understanding functional interaction in the plant microbiota is a field of intense research with a great fundamental interest, as well as promising applied perspectives regarding the identification of novel biocontrol agents such as potentially Ma. I found the manuscript very pleasant to read and well structured. The work appeared rigorous and conclusions are appropriately supported by the available data. A number of biocontrol yeasts have been isolated in the past, but in my opinion this study stands out by providing direct genetic evidence of the anti-rust activity of Ma, through the identification of one underlying gene (encoding a putative GH25 enzyme) and the characterization of the corresponding mutant phenotype. The Ma genome sequence and gene expression data also provide an interesting resource for further analysis of the biology of this phyllosphere fungus.

1) I would have liked to see the activity of the GH25 characterized further. I am lacking a clear hint towards GH25 molecular activity on Al and possibly other members of the phyllosphere. Does ectopic (eg in plants) expression of GH25 has inhibitory effect on Albugo? Is there any indication on which species would this enzyme act? Is it acting specifically on Albugo? What protects Ma from its own secreted hydrolase? How conserved in GH25 across fungi? Is there anything specific about its sequence (and predicted activity) as compared to other GH25s? to other glycosyde hydrolases? To what extent is GH25 activity beneficial to Ma? Could the growth of this fungus measured on Albugo-infected vs Albugo-free plants? in vitro with or without Albugo?

I believe some info and discussion on these questions is important as the identification of GH25 is in my view the major novelty of this work.

2) The description of statistical analyses needs attention. I particular, I could not find information on how gene expression and differentially expressed genes were determined (see statistical comments).

Reviewer #3:

The manuscript deals with the characterization of a phyllosphere yeast that is supposed to have an antagonistic effect on Albugo laibachii.

While the manuscript is well written, it suffers from numerous cases of overinterpretation of the results and methodological flaws, as outlined below.

Abstract. It is beyond reason that the authors call their yeast isolate "Moesziomyces albugensis". First, there is not even a remotely valid description of the "species", and second, it is quite obvious that after checking some of the recent literature on Moesziomyces (part of which two of the authors even co-authored) that the isolate they are investigating is clearly Moeziomycesbullatus. This needs to be changed throughout the manuscript.

Abstract. That the isolates "prevents" Albugo infections is insufficiently shown (but see later criticism on this point).

Abstract. The evidence for the claim that GH25 is a major effector of the antagonism is weak, at best.

Introduction. This statement is not correct. there are several studies, e.g. by the Macia-Vicente group that have shown that the local environment drives the diversity of root-associated microbiomes.

Introduction. The genus Pseudozyma is obsolete. This should be termed Moesziomyces. And the species is known as well, it is Moesziomyces bullatus.

Introduction. I was surprised about the statement and have the read Kurse et al., manuscript. In fact, the authors report the opposite, an infrequent sexual reproduction on grass hosts as the most probable explanation for the retainment of a sexual reproduction.

Introduction. I really wonder why the authors cite this less comprehensive phylogeny, while two of them have shown, based on a broader sampling that in fact the species mentioned are embedded within Moesziomyces and that Pseudozyma aphidis is conspecific with Moesziomyces bullatus.

Introduction. The transfer of these species into Pseudozyma had already been done, thus, "can be renamed" is not a correct statement.

Introduction. This should read Moesziomyces bullatus or "A strain formerly classified as Pseudozyma aphidis (now Moesziomyces bullatus)".

Introduction. There is no such species name as Moesziomyces albuginensis. And this also would not make phylogenetic or evolutionary sense, as the Moesziomyces bullatus strain isolated from Arabidopsis thaliana is firmly embedded within Moesziomyces bullatus and there is no evidence that is should be considered as a biologically distinct species. Also, the evidence of disease "prevention" are at most anecdotal (see later criticism).

Results. This statement alone refutes the previous statement in the Introduction. If the species was isolated from and found to be closely associated with the spores of Albugo, even over years of continuous cultivation cycles, how can the authors confidently claim that the species prevents infection?

Introduction. There are many strains of Moesziomyces bullatus (as Pseudozymaaphidis) that are marked as biocontrol agent. The new strain would just be an addition to that, there is little novelty in this.

Results. This argumentation is impossible to follow. Just because something is isolated from a special location does not make it a new species.

Results. I really cannot see the reason for the authors claiming again the identification was just "Pseudozyma sp." while they could show in a previous manuscript [Kruse et al., 2017] that the species is Meosziomyces bullatus.

Results. This interaction does not need to be specific. The test against Ustilago maydis is also difficult to justify, as that species is largely unrelated to Moesziomyces bullatus. Instead, other species of the genus Moesziomyces, especially additional strains of Moesziomyces bullatus should have been used.

Results. While in general, it is interesting that the yeast showed an antogonistic effect, dose dependency should have been checked. Several species of the Ustulagoinales are known to produce biosurfactants that alter the properties of the plant cuticle. Thus, the effect observed can probably already explained by this. This could be addressed by additional experiments with a variety of Ustilago and Moesziomyces strains and demonstrating that a lower dose of the Albugo-associated strain inhibits the infection.

Subsection “The genome of Moesziomyces albugensis”. This lacks a lot of the usual quality control, such as genome completeness statistics, read mapping specifics, testing of the gene predictions by RNA mapping.

Subsection “The genome of Moesziomyces albugensis”. Evidence should be presented that this is not the result of a mis-assembly.

Subsection “The genome of Moesziomyces albugensis”. Even though "tempting", this speculation of the authors seems very far-fetched, considering that there are many unrelated groups of the Ustilaginales that cause tumors or not.

Subsection “The genome of Moesziomyces albugensis”. This part is interesting, but needs to be backed up by RNASeq and metabolic profiling.

Subsection “The genome of Moesziomyces albugensis”. If the authors reanalyzed that genome, they could probably see that most of the "effector losses" are the result of bad gene calls.

Subsection “The genome of Moesziomyces albugensis”. I wonder why the authors do not cite or discuss their own findings on several Moesziomyces species previously characterized as non-pathogenic. Ökmen et al., had found years ago that these yeasts possess functional copies of PEP1.

Subsection “The genome of Moesziomyces albugensis”. The interpretation here cannot be upheld. Given current data, it seems to be a specialty of Ustilago madis that the clusters are expanded (see e.g. Sharma et al., 2014). In addition, the "loss" or presence of the clusters is likely a simple function of phylogenetic distance. Judging from the unfortunately not well resolved phylogeny in [Wang et al., 2015], Kalmanozyma could simply be most unrelated to Ustilago maydis.

Subsection “Genetic characterization of M. albugensis”. This is the first part of the manuscript that seems solid, even though I wonder why non infection trials on the natural host of Moesziomyces bullatus were done.

Subsection “Identification of microbe-microbe effector genes by RNA-Seq”. While this cell biology part is again rather solid, the conclusions seem much too strong. For stating that the secretion of GH25 is important, the authors would need to construct a non-secreted version under a native promotor. In addition, the metabolic profile would need to be analysed, both of the secreted metabolites and those retained in the cells. It is conceivable that the GH25 enzyme processes a non-effective precursor to generate the observed effect.

As many results obtained are already discussed in the Results section, the Discussion section is sometimes repetitive. I encourage the authors to keep the Results section free from interpretation and to do this in the Discussion section. As many incorrect claims are made again in the Discussion, I do not comment on them again.

Subsection “Bioinformatics and computational data analysis”. The information regarding the genome assembly and gene prediction is at best minimal. This needs to much extended, also including the necessary quality checks as indicated earlier.

Figure 1. This is hardly useful to present the "phylogenetic position" of the strain. There are much more elaborate phylogenies and the authors should simply refer to them. Also, the separation into "apathogenic" and "pathogenic" strains is pointless, as they confirm what was previously known. Likely there are no apathogenic members of the Ustilaginales and the yeasts have just by chance been isolated first from the environment before the pathogenic stage was discovered.

[Editors’ note: further revisions were suggested prior to acceptance, as described below.]

Thank you for submitting your article "A fungal member of the Arabidopsis thaliana phyllosphere antagonizes Albugo laibachii via a secreted lysozyme" for consideration by eLife. Your article has been reviewed by three peer reviewers, including Caroline Gutjahr as the Reviewing Editor and Reviewer #1, and the evaluation has been overseen by a Reviewing Editor and Christian Hardtke as the Senior Editor.

The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission.

We would like to draw your attention to changes in our revision policy that we have made in response to COVID-19 (https://elifesciences.org/articles/57162). Specifically, we are asking editors to accept without delay manuscripts, like yours, that they judge can stand as eLife papers without additional data, even if they feel that they would make the manuscript stronger. Thus the revisions requested below only address clarity and presentation, except if you prefer yourself to add data regarding request No. 3.

Summary:

Eitzen et al., investigate a specific mechanism of antagonism by an epiphytic yeast on an obligate biotrophic oomycete pathogen. They established the yeast Moesziomyces bullatus ex Albugo (MbA), which represents a hub organism in the Arabidopsis phyllosphere microbiome, as a leaf epiphytic model fungus, by providing a sequenced genome, the transcriptome in response to interaction with a phyllosphere bacterial SynCom and the oomycete A. laibachii and by establishing a transformation system for generation of k. o. strains by homologous recombination.

They report that M. bullatus ex Albugo antagonizes the oomycete pathogen A. laibachii, via a putatively secreted GH25 hydrolase with lysozyme activity. This finding, and other findings that may be expected from this research line in the future, do not only add to a molecular understanding of microbial interactions on the leaf surface but may in the future yield biocontrol fungi to combat agriculturally relevant oomycete infections of crop leaves.

Essential revisions:

Based on the major comments of all three reviewers and a discussion, we request to address the following points before the manuscript can be accepted by eLife:

1) The reviewers found the phylogenetic tree in Figure 1 misleading (see also comments about the tree in the previous version of the manuscript). Since a more comprehensive and correct tree is published in Kruse et al., Figure 1 should be removed from the manuscript.

2) The GH25 hydrolase exhibits lysozyme activity, which hydrolyzes β‐(1,4) linkages between the NAM and NAG saccharides. These glucans are usually found on bacterial cell walls. However, to the reviewer's knowledge, the cell wall of oomycetes is composed of β-1,3, and β-1,6 glucans. How does GH25 hydrolase inhibit the oomycete Albugo laibachii? This question or hypotheses should be raised and discussed in the Discussion section.

3) The Title states that GH25 is a secreted enzyme. Unless the authors include an experiment showing secretion, statements on secretion should be toned down in the title and across the manuscript where applicable. (In several instances the authors have already carefully written about "an enzyme with predicted secretion peptide").

[Editors’ note: the authors resubmitted a revised version of the paper for consideration. What follows is the authors’ response to the first round of review.]

Reviewer #1:

The authors establish Moesziomyces albugensis as a new "model fungus" at the leaf surface, which may act as a parasite on the pathogenic oomycete of Arabidopsis Albugo laibachii. They provide a sequenced genome, the transcriptome in response to interaction with the plant, phyllosphere bacteria and the oomycete A. laibachii, a transformation system for generation of k. o. strains by homologous recombination and characterize the mating type as well as filament and appressoria forming behavior in vitro.

In addition, they also report some interesting biology and show that M. albugensis antagonizes the oomycete pathogen A. laibachii, probably via a secreted putative GH25 hydrolase. This finding and other findings, that may be expected from this research line in the future, may yield biocontrol fungi to combat agriculturally relevant oomycete infections of crop leaves.

While M.albugensis is an exciting new fungus and the identification of the effector required for parasitism of A. laibachii is very interesting, the manuscript still suffers from a few unclarities.

– From the text I did not understand whether M. albugensis lives only epiphytically or can also colonize the plant tissue (it forms appressoria on hydrophobic surfaces) or whether it is exclusively associated with A. laibachii. This should be clarified.

We apologize for being unclear in this point. M. albugensis belongs to a group of basidiomycete yeasts that has been found to lead epiphytic existence on the leaf surface. However, since the recombinant, self-compatible strain (CB1) which we generated in this study was found to form appressoria on hydrophobic surface, this hints towards the idea that in nature, the anamorphic epiphytic yeasts might still own the potential to form filamentous structure in presence of a correct mating partner. However, this remains rather speculative because the general question of how the ustilaginean yeasts actually relate to pathogenic strains remains to be solved. We added an additional clarification in subsection “Genetic characterization of MbA”.

– It is unclear how the bacterial SynCom was constructed and which isolates it contains.

We apologize for not having included the Syn Com isolates list. We have added this to the supplementary now and have explained how the syncom was assembled: For this SynCom we had isolated bacteria from different sites close to Tuebingen in Germany and have sequenced partial 16S to identify taxonomy. Based on these isolates and how often we have found the bacteria on different sites (at least 85% of all sites), we assembled a community with maximum phylogenetic diversity. We have further isolated bacteria from spores of our unsterile A. laibachii Nc14 isolate and have added those to cover more phylogenetic diversity.

The detailed information is now available in the new Supplementary file 1.

– The quantification of A. laibachii infection seems kind of rough and prone to subjectivity. An additional, more quantitative method should be used for quantification, for example by qPCR using an amplicon from A. laibachii DNA normalized to a plant amplicon.

We thank this reviewer for pointing this out. Our quantification method is based on the percentage of infected leaves vs non-infected leaves in each individual A. thaliana seedling. This method of counting pustules on leaves has been intensely used for evaluating rust fungal infections. As leaves of sterile Arabidopsis plants are small, there are no more than one to two pustules per leaf. Using this range would not give enough resolution to see significant differences and therefore we decided to analyze significant more plants and leaves and focus on presence and absence. For statistical analysis, infection percentage of each seedling is subjected to analysis of variance (ANOVA) for pairwise comparison of the conditions, with Tukey’s HSD test to determine significant differences among them (P values <0.05). We performed the experiments in six independent biological replicates with 12 technical replicates and observed significant difference between the Albugo treated seedlings and Albugo + Moeziomyces treated ones. Hence, we are convinced that the data shown in our study is robust and statistically significant.

However, based on the suggestion of this reviewer, we performed qPCR-based relative quantification of the Albugo biomass in A. thaliana in our new experiments where samples were treated with GH25 hydrolase and Mutant_GH25, respectively (see new Figure 8D). These experiments provide direct evidence for the biological activity of the GH25 hydrolase: Albugo biomass was reduced to about 50% in GH25 treatments compared to Mutant GH25 or untreated Albugo-infections. This new data is now included in the Results section and the Materials and methods section.

– It is unclear whether GH25 hydrolase suppresses infection of plant leaves by A. laibachii or whether it helps M. albugensis to parasitize A. laibachii by other means. The authors should examine the effect of isolated GH25 hydrolase on A. laibachii infection.

This is a very good question, and we think it is answered by the new data which is already mentioned in the point above: the plant infection experiments with the recombinant GH25 hydrolase show that it directly interferes with Albugo infection (see new Figure 8D). Based on this finding, our next question for upcoming work will be to elaborate the molecular mechanism by which the enzyme inhibits Albugo.

– At which stage of infection is A. laibachii blocked by GH25 hydrolase or M. albugensis?

We observed a slight but not significant decrease in zoosporangial germination efficiency on treatment with the GH25 hydrolase (new Figure 7—figure supplement 5) but found that plant colonization of Albugo is significantly reduced. This is also in line with the microscopic of A. thaliana leaves where colonization of Albugo is visualized by Trypan blue staining (new Figure 1—figure supplement 2). This shows that in presence of MbA, there is little to no colonization of Albugo in the leaf surface and only short hyphae are formed. We have also provided more information in the Results.

Reviewer #2:

The study by Eitzen et al., characterizes a yeast species isolated from Arabidopsis phyllosphere classified as Moesziomyces albugensis (Ma). The report provides phylogenetic classification, genome analysis and functional analysis of Ma inhibitory activity towards the white rust oomycete Albugo laibachii (Al). Understanding functional interaction in the plant microbiota is a field of intense research with a great fundamental interest, as well as promising applied perspectives regarding the identification of novel biocontrol agents such as potentially Ma. I found the manuscript very pleasant to read and well structured. The work appeared rigorous and conclusions are appropriately supported by the available data. A number of biocontrol yeasts have been isolated in the past, but in my opinion this study stands out by providing direct genetic evidence of the anti-rust activity of Ma, through the identification of one underlying gene (encoding a putative GH25 enzyme) and the characterization of the corresponding mutant phenotype. The Ma genome sequence and gene expression data also provide an interesting resource for further analysis of the biology of this phyllosphere fungus.

1) I would have liked to see the activity of the GH25 characterized further. I am lacking a clear hint towards GH25 molecular activity on Al and possibly other members of the phyllosphere.

To address this point, we expressed the GH25 hydrolase (and an enzymatic inactive version) in Pichia pastoris, purified the proteins by Ni-NTA affinity chromatography and performed lysozyme activity assays using the recombinant proteins. This showed that the GH25 has a lysozyme activity, which is abolished by a single amino acid exchange in the predicted active site. This data is now included in the manuscript (New Figure 8C) and is described in the Results and the Materials and methods section.

Does ectopic (eg in plants) expression of GH25 has inhibitory effect on Albugo?

To address the question whether the GH25 lysozyme has a direct effect on Albugo, we used the recombinant protein. As mentioned above, we did not observe a significant effect on Albugo zoospore formation (New Figure 7—figure supplement 5). On the other hand, we found that treatment of Albugo-infected Arabidopsis with the active (but not with the inactive) GH25 significantly reduced Albugo growth (New Figure 8D). This provides direct evidence showing that the GH25 lysozyme is actually inhibiting infection of Arabidopsis by Albugo.

Is there any indication on which species would this enzyme act? Is it acting specifically on Albugo?

In this study we were focusing on the Moesziomyces – Albugo interaction on Arabidopsis. However, as also suggested by reviewer 1, we believe that Moesziomyces and the GH25 could have a potential used for biocontrol in economically more relevant systems. We will therefore test its impact on Phytophthora infestans in future experiments.

What protects Ma from its own secreted hydrolase?

The most likely explanation is that the fungal cell wall does not contain a substrate of the GH25 lysozyme. We are not aware of any other protection mechanism in Moesziomyces.

How conserved in GH25 across fungi? Is there anything specific about its sequence (and predicted activity) as compared to other GH25s? to other glycosyde hydrolases?

The GH25 hydrolase encoded by gene g2490 is predicted has a lysozyme activity, which is essentially an enzyme that cleaves peptidoglycan in bacterial cell walls by catalyzing the hydrolysis of β‐(1,4) linkages between the NAM and NAG saccharides The active site of the GH25 hydrolase is found to contain a DXE motif which is found to be conserved between a wide range of fungi, ranging from other basidiomycetes, ascomycetes and even Chytrids (see also new Figure 7—figure supplement 4). We have additionally attached the raw file for PRALINE multiple sequence analysis for conserved sites.

To what extent is GH25 activity beneficial to Ma?

This is an interesting question, which we cannot answer based on experimental evidence so far. What we show is that the GH25 is required and responsible for the antagonism towards Albugo. One now could hypothesize that this antagonism itself is beneficial for Moesziomyces, i.e. for interspecific competition in niche differentiation. A final answer for this question, however, will be subject of future research.

Could the growth of this fungus be measured on Albugo-infected vs Albugo-free plants? in vitro with or without Albugo?

This is an interesting thought, which will definitely be part of future analysis on the ecological role of the GH25 for Moesziomyces as a member of the leaf phyllosphere, particularly in presence of Albugo. Regarding “in-vitro” growth, this is experimentally difficult, since Albugo is an obligate biotroph which cannot be propagated in axenic culture.

2) The description of statistical analyses needs attention. I particular, I could not find information on how gene expression and differentially expressed genes were determined.

How was gene expression level determined? How were differentially expressed genes identified? Besides, some information on RNAseq QC and bias correction is needed.

We have added additional, missing. Each file was checked for quality control of RNA-Seq reads prior to analysis with FastQC. Reads were mapped to the Moesziomyces and Arabidopsis genome with Tophat2 and a count table was generated for Moesziomyces counts with HT-seq count. DE-genes were determined with using the "limma"-package in R on "voom"-transformed count data and sample-specific weighting to reduce variance between samples (see Figure 6—figure supplement 1). Counts lower than 10 in average have been extracted. Differentially expressed genes were counted with a threshold of False discovery rate of 0.05 and log2FC > 0. We have the added additional information, in Results section.

Reviewer #3:

The manuscript deals with the characterization of a phyllosphere yeast that is supposed to have an antagonistic effect on Albugo laibachii.

While the manuscript is well written, it suffers from numerous cases of overinterpretation of the results and methodological flaws, as outlined below.

Abstract. It is beyond reason that the authors call their yeast isolate "Moesziomyces albugensis". First, there is not even a remotely valid description of the "species", and second, it is quite obvious that after checking some of the recent literature on Moesziomyces (part of which two of the authors even co-authored) that the isolate they are investigating is clearly Moeziomyces bullatus. This needs to be changed throughout the manuscript.

We accept the criticism by the reviewer. We therefore describe the characterized isolate as Moesziomyces bullatus. Nevertheless, it needs to be specified that the characterized strain is not a pathogenic strain of Moesziomyces bullatus, which is a pathogen of millet. Despite multiple attempts, we were neither able to mate the strain with any isolate of an available collection of pathogenic Moesziomyces bullatus, nor have we been able to infect a potential host plant. We have been trying to infect millet plants also with the self-compatible CB1 strain, which is able to form appressoria. However, also this strain did not show invasive growth in any of our experiments (data not shown). Therefore, to specify the isolate that is characterized in this study, we describe it as Moesziomyces bullatus ex Albugo on Arabidopsis (MbA for simplicity) throughout the manuscript. This classifies the strain as a member of the species Moesziomyces bullatus and at the same time it reflects that it has actually been isolated from Albugo on Arabidopsis plants and not as a pathogen of millet.

Abstract. That the isolates "prevents" Albugo infections is insufficiently shown (but see later criticism on this point).

We realized that “prevent” is not an appropriate term here and rephrased it to “reduces infection”, which underlines that we are observing a quantitative effect and not a complete prevention of Albugo infection.

Abstract. The evidence for the claim that GH25 is a major effector of the antagonism is weak, at best.

We accept this criticism, which is also shared by the other reviewers and have performed a series of additional experiments which significantly strengthens this part of the study and provides direct evidence for GH25 being an effector in the observed antagonism. As described above, we produced recombinant protein for the GH25 hydrolase and showed its enzymatic activity. Most importantly, we show that enzymatically active GH25 significantly reduces infection of Arabidopsis by Albugo, which we quantified using qPCR. The additional evidence is shown in the new Figure 8 and in the new Figure 1—figure supplement 2 and Figure 7—figure supplement 2. Detailed descriptions are provided in Materials and methods section and Results of the revised manuscript.

Introduction. This statement is not correct. there are several studies, e.g. by the Macia-Vicente group that have shown that the local environment drives the diversity of root-associated microbiomes.

We agree and have modified the sentence accordingly.

Introduction. the genus Pseudozyma is obsolete. This should be termed Moesziomyces. And the species is known as well, it is Moesziomyces bullatus.

We have mentioned the term Pseudozyma to introduce the anamorphic yeasts in the manuscript and report biocontrol activity of those group yeasts which have been commonly termed “Pseudozyma” in a substantial body of literature. In accordance to the corrected phylogeny, we mentioned the transfer of the genus Pseudozyma to Moesziomyces soon after in the Introduction.

Introduction. I was surprised about the statement and have the read Kurse et al., manuscript. In fact, the authors report the opposite, an infrequent sexual reproduction on grass hosts as the most probable explanation for the retainment of a sexual reproduction.

We modified the sentence to avoid confusion (Introduction).

Introduction. I really wonder why the authors cite this less comprehensive phylogeny, while two of them have shown, based on a broader sampling that in fact the species mentioned are embedded within Moesziomyces and that Pseudozyma aphidis is conspecific with Moesziomyces bullatus.

The corrected phylogeny has now been incorporated.

Introduction. The transfer of these species into Pseudozyma had already been done, thus, "can be renamed" is not a correct statement.

We have modified the sentence.

Introduction. This should read Moesziomyces bullatus or "A strain formerly classified as Pseudozyma aphidis (now Moesziomyces bullatus)".

We have changed this as described above.

Introduction. There is no such species name as Moesziomyces albuginensis. And this also would not make phylogenetic or evolutionary sense, as the Moesziomyces bullatus strain isolated from Arabidopsis thaliana is firmly embedded within Moesziomyces bullatus and there is no evidence that is should be considered as a biologically distinct species. Also, the evidence of disease "prevention" are at most anecdotal (see later criticism).

We have addressed this point as described above.

Results. This statement alone refutes the previous statement in the Introduction. If the species was isolated from and found to be closely associated with the spores of Albugo, even over years of continuous cultivation cycles, how can the authors confidently claim that the species prevents infection?

The fungal isolate used in this study in particular has been found associated with Albugo-infected A. thaliana leaf surface. Previous study (Agler et al., 2016) identified this strain of Moesziomyces as being a hub microbe in the A. thaliana phyllosphere. In this study we provide direct, quantitative evidence that it antagonizes the infection of Arabidopsis by the biotrophic oomycete pathogen Albugo laibachii via the GH25 hydrolase. However, we do not want to claim that Moesziomyces sp. in general has such a function (to justify this claim, one basically would need to test every fungal isolate individually on the experimental level), and therefore we specified the findings to the individual isolate which we have studied in detail in this study.

Introduction. There are many strains of Moesziomyces bullatus (as Pseudozyma aphidis) that are marked as biocontrol agent. The new strain would just be an addition to that, there is little novelty in this.

We agree with the reviewer that there are several reports on biocontrol activity by Pseudozyma sp., which is also addressed in the introduction part of our manuscript. However, we are convinced that our study is not a mere description of an organism with biocontrol activity. An important aspect is on the multitrophic interactions in a phyllosphere microbial community, where MbA appears to be regulating the presence of several other microbes in the network. Another novelty of our approach lies in the fact that we haven’t externally introduced an organism to combat pathogenicity; since the yeast is found as a member of the microbial community in the natural environment. Most importantly, our RNAseq analysis revealed the transcriptional activation of secretory enzymes in MbA when being co-cultivated with Albugo on the plant leaf, and this approach identified the GH25 lysozyme as a key determinant for the observed microbial antagonism. To the best of our knowledge, this study shows for the first direct and functional evidence on the genetic and protein level in a fungal – oomycete antagonism on/in-planta. Thus, the novelty of this study is o on the (i) community aspect and (ii) the mechanistic evidence of the microbial antagonism, rather than in the identification of a new fungal isolate.

Results. This argumentation is impossible to follow. Just because something is isolated from a special location does not make it a new species.

This has been corrected as described above.

Results. I really cannot see the reason for the authors claiming again the identification was just "Pseudozyma sp." while they could show in a previous manuscript [Kruse et al., 2017] that the species is Meosziomyces bullatus.

This has been corrected as described above.

Results. This interaction does not need to be specific. The test against Ustilago maydis is also difficult to justify, as that species is largely unrelated to Moesziomyces bullatus. Instead, other species of the genus Moesziomyces, especially additional strains of Moesziomyces bullatus should have been used.

The intention of this experiment was to observe if there is any difference between the well-characterized pathogenic smut Ustilago maydis and the MbA strain, which acts as a microbial hub in the A. thaliana phyllosphere. As suggested by this reviewer, it would be interesting to test other strains of Moesziomyces bullatus against the bacterial SynCom and Albugo to test if there is specific adaptation towards microbial interactions in different isolates of the same species. These lines will surely be followed by our future research in this context, where we want to better explore the ecology Ustilaginales yeasts in different environments.

Results. While in general, it is interesting that the yeast showed an antogonistic effect, dose dependency should have been checked. Several species of the Ustulagoinales are known to produce biosurfactants that alter the properties of the plant cuticle. Thus, the effect observed can probably already explained by this. This could be addressed by additional experiments with a variety of Ustilago and Moesziomyces strains and demonstrating that a lower dose of the Albugo-associated strain inhibits the infection.

We agree that Ustilaginales are known to produce biosurfactants and we would not like to exclude that these secondary metabolites have significant impact on the microbial interactions on the leaf surface. The focus and aim of this study, however, was to investigate possible functions of secreted proteins as potential microbe-microbe effectors. Thus, we decided to investigate the function of the GH25 in more detail and performed the characterization of the recombinant protein, showing that the active GH25 lysozyme is significantly inhibiting infection of Arabidopsis by A. laibachii (new Figure 8C,D).

Subsection “The genome of Moesziomyces albugensis”. This lacks a lot of the usual quality control, such as genome completeness statistics, read mapping specifics, testing of the gene predictions by RNA mapping.

We thank the reviewer for pointing out this lack of information on the RNA seq experiment. Each file was checked for quality control of RNA-Seq reads prior to analysis with FastQC. Reads were mapped to the Moesziomyces and Arabidopsis genome with Tophat2 and a count table was generated for Moesziomyces counts with HT-seq count. DE-genes were determined with using the "limma"-package in R on "voom"-transformed count data and sample-specific weighting to reduce variance between samples (new Figure 6—figure supplement 1). We have added additional information, in the Results section.

Subsection “The genome of Moesziomyces albugensis”. Evidence should be presented that this is not the result of a mis-assembly.

We have added the new Figure 2—figure supplement 1, which shows genomic comparison of MbA and Moesziomyces antarctica T-34 and confirms proper assembly..

Subsection “The genome of Moesziomyces albugensis”. Even though "tempting", this speculation of the authors seems very far-fetched, considering that there are many unrelated groups of the Ustilaginales that cause tumors or not.

We agree that this statement is very speculative and therefore deleted the sentence.

Subsection “The genome of Moesziomyces albugensis”. This part is interesting, but needs to be backed up by RNASeq and metabolic profiling.

Our analysis provided no evidence on upregulation of the secondary metabolite clusters in the RNA seq experiments, hence metabolic profiling was not performed further. Thus, we only speculate on them to be involved in antibacterial activity based on interaction of MbA against the SynCom bacterial members on plate (Figure 1—figure supplement 1).

Subsection “The genome of Moesziomyces albugensis”. If the authors reanalyzed that genome, they could probably see that most of the "effector losses" are the result of bad gene calls.

We have not analyzed the P. flocculosa genome with respect to gene loss. To avoid misunderstanding, we modified the sentence to make it clearer that we only refer to the statement by Lefebvre et al., (2013), which is apparently not in line with our observation made for MbA.

Subsection “The genome of Moesziomyces albugensis”. I wonder why the authors do not cite or discuss their own findings on several Moesziomyces species previously characterized as non-pathogenic. Ökmen et al., had found years ago that these yeasts possess functional copies of PEP1.

We have added the following in (subsection “The genome of MbA”) “Moesziomyces sp., possess functional homologues of the pep1 gene, a core virulence effector of U. maydis (Sharma et al., 2019), suggesting that such anamorphic yeasts have the potential to form infectious filamentous structures by means of sexual reproduction (Kruse et al., 2017).”

Subsection “The genome of Moesziomyces albugensis”. The interpretation here cannot be upheld. Given current data, it seems to be a specialty of Ustilago madis that the clusters are expanded (see e.g. Sharma et al., 2014). In addition, the "loss" or presence of the clusters is likely a simple function of phylogenetic distance. Judging from the unfortunately not well resolved phylogeny in [Wang et al., 2015], Kalmanozyma could simply be most unrelated to Ustilago maydis.

We agree and have removed the example of Kalmanozyma brasiliensis.

Subsection “Genetic characterization of M. albugensis”. This is the first part of the manuscript that seems solid, even though I wonder why non infection trials on the natural host of Moesziomyces bullatus were done.

We appreciate this suggestion by this reviewer. Actually, we tested two different cultivars of millet, Setaria sp. and Echinochloa crus-galli for susceptibility towards transgenic self-compatible (CB1) strain of Moesziomyces sp. However, we could not observe any penetration event in the plant surface, even though the organism was seen to be forming filamentous structure. The same was true for the self-compatible CB1 strain. Thus, we cannot make any statement on a potential role of the MbA isolate as a plant pathogen in our paper. On the one hand we were not able to infect a potential host, but on the other hand we also cannot exclude that under certain conditions it could be infectious.

Subsection “Identification of microbe-microbe effector genes by RNA-Seq”. While this cell biology part is again rather solid, the conclusions seem much too strong. For stating that the secretion of GH25 is important, the authors would need to construct a non-secreted version under a native promotor. In addition, the metabolic profile would need to be analysed, both of the secreted metabolites and those retained in the cells. It is conceivable that the GH25 enzyme processes a non-effective precursor to generate the observed effect.

We agree that the statement was quite bolt in the previous version of the manuscript. However, since the experiments with the recombinant GH25 lysozyme (Figure 8) confirmed its role in the interaction, we decided to stay with this phrase, which is now justified in our opinion.

As many results obtained are already discussed in the Results section, the Discussion section is sometimes repetitive. I encourage the authors to keep the Results section free from interpretation and to do this in the Discussion section. As many incorrect claims are made again in the Discussion, I do not comment on them again.

We thank the reviewer for pointing this out. As suggested, we have omitted certain sentences in the Results section to avoid being repetitive.

Subsection “Bioinformatics and computational data analysis”. The information regarding the genome assembly and gene prediction is at best minimal. This needs to much extended, also including the necessary quality checks as indicated earlier.

Again, we agree with the reviewer and have extended this part in the Results section.

Figure 1. This is hardly useful to present the "phylogenetic position" of the strain. There are much more elaborate phylogenies and the authors should simply refer to them. Also, the separation into "apathogenic" and "pathogenic" strains is pointless, as they confirm what was previously known. Likely there are no apathogenic members of the Ustilaginales and the yeasts have just by chance been isolated first from the environment before the pathogenic stage was discovered.

We discussed this critically and, although we agree with this reviewer that there are more extensive phylogenies published, we still believe that the figure is helpful for the reader as an introduction. However, we agree with this reviewer that the term “apathogenic” is not helpful and therefore this has been changed in the revised figure legend.

[Editors’ note: what follows is the authors’ response to the second round of review.]

Essential revisions:

Based on the major comments of all three reviewers and a discussion, we request to address the following points before the manuscript can be accepted by eLife:

1) The reviewers found the phylogenetic tree in Figure 1 misleading (see also comments about the tree in the previous version of the manuscript). Since a more comprehensive and correct tree is published in Kruse et al., Figure 1 should be removed from the manuscript.

As suggested, we have removed Figure 1 from the manuscript. We felt it was helpful as an introduction to the topic, but we agree that the information is not necessarily required in this manuscript.

2) The GH25 hydrolase exhibits lysozyme activity, which hydrolyzes β‐(1,4) linkages between the NAM and NAG saccharides. These glucans are usually found on bacterial cell walls. However, to the reviewer's knowledge, the cell wall of oomycetes is composed of β-1,3, and β-1,6 glucans. How does GH25 hydrolase inhibit the oomycete Albugo laibachii? This question or hypotheses should be raised and discussed in the Discussion section.

We fully agree that the observed effect of the GH25 on A. laibachii infection is surprising (we were surprised by this finding as well). The GH25 enzymes are widely conserved in the fungal kingdom and have been associated with both fungal hyperparasitism, as well as oomycete-oomycete parasitism in Phytium (Horner et al., 2012).

With regard of a direct effect on the Albugo cell wall, one should mention that there is quite sparse knowledge on the actual cell wall composition and we would actually not exclude that it contains a substrate of the GH25. Although the MbA GH25 obviously has lysozyme activity, a comprehensive biochemical characterization of the fungal GH25 family is missing. Thus, it might very well be that these enzymes have unexpected/unknown substrates in the oomycete cell wall. Next, it is unknown which consequences an eventual hydrolysis of an Albugo cell wall compound might have (e.g. an effect on cell wall integrity, or a modification of the cell wall surface which may lead to a block in signal perception – or even an induction of plant defense). In the revised manuscript we now include a discussion on potential functions of GH25 to give the reader an idea on how it might act on the molecular level.

Lastly, one cannot rule out that the effect on A. laibachii might be an indirect one: since oomycetes can be found in association with bacteria (which is a largely unstudied/neglected topic) one could even speculate that the GH25 actually targets Albugo-associated bacteria, which might be necessary for virulence of the oomycete. While such a tempting scenario would be also in line with findings of endosymbiotic bacteria in mycorrhizal fungi (work by P. Bonfante and others), at present this is purely speculative and unless there no evidence pointing towards this direction, we’d prefer to not discuss this aspect in the present manuscript.

3) The Title states that GH25 is a secreted enzyme. Unless the authors include an experiment showing secretion, statements on secretion should be toned down in the title and across the manuscript where applicable. (In several instances the authors have already carefully written about "an enzyme with predicted secretion peptide").

We accept this criticism and removed the word “secreted” from the Title, which now reads “…via a GH25 lysozyme”.

Author details

1. Katharina Eitzen

   1. Institute for Plant Sciences and Cluster of Excellence on Plant Sciences (CEPLAS), University of Cologne, Center for Molecular Biosciences, Cologne, Germany
   2. Max Planck Institute for Plant Breeding Research, Cologne, Germany

   Contribution

   Conceptualization, Investigation, Visualization, Methodology, Writing - original draft

   Competing interests

   No competing interests declared

2. Priyamedha Sengupta

   Institute for Plant Sciences and Cluster of Excellence on Plant Sciences (CEPLAS), University of Cologne, Center for Molecular Biosciences, Cologne, Germany

   Contribution

   Investigation, Visualization, Methodology, Writing - review and editing

   Competing interests

   No competing interests declared

3. Samuel Kroll

   Max Planck Institute for Plant Breeding Research, Cologne, Germany

   Contribution

   Investigation, Methodology

   Competing interests

   No competing interests declared

4. Eric Kemen

   1. Max Planck Institute for Plant Breeding Research, Cologne, Germany
   2. Department of Microbial Interactions, IMIT/ZMBP, University of Tübingen, Tübingen, Germany

   Contribution

   Conceptualization, Formal analysis, Supervision, Funding acquisition, Writing - original draft, Project administration, Writing - review and editing

   For correspondence

   eric.kemen@uni-tuebingen.de

   Competing interests

   No competing interests declared

5. Gunther Doehlemann

   Institute for Plant Sciences and Cluster of Excellence on Plant Sciences (CEPLAS), University of Cologne, Center for Molecular Biosciences, Cologne, Germany

   Contribution

   Conceptualization, Formal analysis, Supervision, Funding acquisition, Writing - original draft, Project administration, Writing - review and editing

   For correspondence

   g.doehlemann@uni-koeln.de

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0002-7353-8456