The functionally conserved effector Sta1 is a fungal cell wall protein required for virulence in Ustilago maydis.

(1)The biotrophic fungus Ustilago maydis causes the smut disease of maize. The interaction with its host and induction of characteristic tumors are governed largely by secreted effectors whose function is mostly unknown. To identify effectors with a prominent role in virulence, we used RNA-seq and found that the gene sta1 is upregulated during early stages of infection. (2)We characterized Sta1 by comparative genomics, reverse genetics, protein localization, stress assays and microscopy. (3)sta1 mutants show a dramatic reduction of virulence and show altered colonization of tissue neighboring the vascular bundles. Functional orthologs of Sta1 are found in related smut pathogens infecting monocot and dicot plants. Sta1 is secreted by budding cells but is attached to the cell wall of filamentous hyphae. Upon constitutive expression of Sta1 fungal filaments become susceptible to Congo red, β-glucanase and chitinase, suggesting that Sta1 alters the structure of the fungal cell wall. Constitutive or delayed expression of sta1 during plant colonization negatively impacts on virulence. (4)Our results suggest that Sta1 is a novel kind of effector, which needs to modify the hyphal cell wall to allow hyphae to be accommodated in tissue next to the vascular bundles.


Introduction
Filamentous plant pathogens secrete effector proteins during host infection to suppress plant immunity and to modulate plant signaling and metabolism (Lo Presti et al., 2015;Tanaka et al., 2015b;Toruno et al., 2016). Secreted effectors can either be delivered into the host cytoplasm to directly manipulate processes inside plant cells (cytoplasmic effectors), or can stay in the apoplastic space to protect fungal cells from plant defense components (apoplastic effectors). Apoplastic effectors required for virulence can be inhibitors of plant enzymes that can be detrimental to the pathogen like chitinases, peroxidase and proteases (Lange et al., 1996;Tian et al., 2004;Rooney et al., 2005;Tian et al., 2005;Tian et al., 2007;van Esse et al., 2008;Song et al., 2009;Naumann et al., 2011;Hemetsberger et al., 2012;Mueller et al., 2013;Okmen et al., 2018). Apoplastic effectors can also attach to the fungal cell wall and the most prominent examples of these are LysM domain proteins, which bind chitin in the fungal cell wall and protect fungi from adverse effects of plant chitinases (de Jonge et al., 2010;Marshall et al., 2011;Mentlak et al., 2012;Sanchez-Vallet et al., 2013;Takahara et al., 2016). Fungal cell wall bound effectors have also been shown to protect from antifungal host proteins (Ma et al., 2018). These examples document that apoplastic effectors often play important roles in plant-pathogen interactions (Doehlemann & Hemetsberger, 2013;Wang & Wang, 2018).
Ustilago maydis is a biotrophic fungal pathogen causing smut disease in maize (Zea mays) (Banuett, 1995;Vollmeister et al., 2012;Zuo et al., 2019). Characteristic disease symptom elicited by U. maydis are tumors, which can develop on all above ground organs of the maize plant. U. maydis encodes several hundred putative secreted effector proteins and many of these contribute to virulence (Kamper et al., 2006). A recent time-resolved RNA-seq analysis of U. maydis genes during host colonization revealed that the genes encoding putative secreted effector proteins are specifically upregulated and are expressed in waves during the course of fungal development on and inside the plant tissue (Lanver et al., 2018).
Comparative genomics of smut pathogens uncovered that U. maydis and related smut fungi including the maize head smut pathogen Sporisorium reilianum, the sugar cane pathogen Sporisorium scitamineum, the barley pathogen Ustilago hordei and Melanopsichium pennsylvanicum infecting Persicaria species share a small number of core effectors (Schuster et al., 2018). One of the

Accepted Article
This article is protected by copyright. All rights reserved characterized core effector is Pep1, a protein of 178 amino acids containing four cysteine residues (Doehlemann et al., 2009), which inhibits the activity of the apoplastic maize peroxidase POX12 (Hemetsberger et al., 2012). The essential virulence-promoting function of Pep1 is conserved in other smuts including U. hordei and M. pennsylvanicum (Hemetsberger et al., 2015). Another core effector crucial for virulence is Cce1 (Seitner et al., 2018). Cce1 possesses similar features as Pep1, e.g. is a protein of only 129 amino acids and possesses eight cysteine residues. Its virulence-promoting function can be complemented by the ortholog from U. bromivora (Seitner et al., 2018). Although the function of Cce1 is still unclear, the authors speculate that Cce1 may also inhibit early plant defense responses in the apoplast. A third apoplastic core effector is Rsp3 and its virulence-promoting function is conserved in the ortholog from S. reilianum (Ma et al., 2018). Rsp3 localizes at the surface of fungal cells and protects them from the maize antifungal proteins AFP1 and AFP2 (Ma et al., 2018). These three examples illustrate that core effectors of U. maydis have important virulence functions.
In this study, we identified the sta1 as a novel core effector gene in U. maydis that is transiently upregulated during the early infection stages and is required for virulence. To carry out its virulence function, sta1 expression needs to occur during a specific time window after plant colonization. Sta1 is a cell wall protein specifically attached to hyphae and likely needed to re-organize the fungal cell wall structure at a specific stage during host colonization.
For virulence assay, U. maydis strains were grown in YEPSL (0.4% yeast extract, 0.4% peptone, 2% sucrose) on a rotary shaker (200 r.p.m.) at 28 °C overnight. Cells were pelleted and resuspended in H 2 O (OD 600 = 1.0) and then injected into the stem of 7 day-old maize seedlings with a syringe as

Accepted Article
This article is protected by copyright. All rights reserved previously described (Kamper et al., 2006). Disease symptoms were scored at 12 days post infection using a previously developed scoring scheme (Kamper et al., 2006). Disease symptoms were quantified based on three biological replicates, and are presented as stacked histograms. The raw data of all infection assays as well as the statistical analysis can be found in Table S1. Teliospore germination was performed as previously described (Flor-Parra et al., 2007).

Plasmid construction and mutant generation
Either the Gibson Assembly kit (New England Biolabs, Frankfurt am Main, Germany) or standard molecular cloning techniques were used for plasmid construction. DNA assembly using the Gibson Assembly kit was performed according to the manufacturer's protocol. All plasmids generated or used in this study are listed in Table S2. All primers used for DNA amplification are listed in Table S3.
To generate deletion mutants of sta1 (UMAG_12226) in U. maydis, the deletion construct pKO_HygR_sta1 was generated. The left and right border regions (1000 bp each) of sta1 were amplified by polymerase chain reaction (PCR) with primers Um12226_LB_Gib-F and Um12226_LB_Gib-R, and Um12226_RB_Gib-F and Um12226_RB_Gib-R, respectively. Amplified fragments were mixed with a hygromycin resistance marker cassette amplified with primers Um12226_RB_Gib-F and Um12226_RB_Gib-R from plasmid pBS-Hyg (Molina & Kahmann, 2007) and pBlueScript linearized by EcoRI-BamHI, and were assembled using the Gibson Assembly kit.
From the resulting plasmid, a 4 kb SspI fragment containing the hygromycin resistance marker cassette flanked by the left and right border regions of sta1 was used for transformation of U. maydis SG200, FB1 and FB2 strains. Gene replacement mutants were identified by Southern blot analysis.
To generate a complementation strain of SG200Δsta1, the construct p123_Sta1 was generated.
Genomic DNA from SG200 containing promoter and open reading frame of sta1 were amplified by PCR with primers um12226pro-F and um12226-R. The amplified fragment was introduced into the integrative plasmid p123 (Aichinger et al., 2003) after digestion by NdeI and AscI. To complement the virulence defects of SG200Δsta1 by Sta1-HA, the construct p123_Sta1_HA was generated.
Genomic DNA from SG200 containing promoter and open reading frame of sta1 were amplified by PCR with primers um12226pro-F and um12226-R2. The amplified fragment was introduced into p123 as described above. To generate a strain where Sta1 fused with mCherry-HA is expressed from

Accepted Article
This article is protected by copyright. All rights reserved its own promoter, the plasmid p123_Sta1_mCherry_HA was generated as follows. From genomic DNA of SG200, sta1 including promoter and open reading frame was amplified with primers um12226pro_Gib-F and um12226_Gib-R. mCherry was amplified by PCR with primers mCherry_HA_Gib-F and mCherry_HA_Gib-R from plasmid p35S-Tin2 26-207 -mCherry-HA-3xNLS (Tanaka et al., 2015a). Amplified fragments were assembled using the Gibson Assembly kit into p123 linearized by KpnI-NotI. Prior to transformation, plasmids were linearized with SspI. Transformants were screened by southern analysis and strains were chosen in which a single copy of the plasmid was inserted in the ip locus of SG200Δsta1 (Loubradou et al, 2001) To generate complementation strains of SG200Δsta1 by sta1 orthologs from other smut pathogens, the 323 bp promoter region of U. maydis sta1 and open reading frame of sta1 orthologs from other smut pathogens were amplified by PCR with the primers indicated in Table S2 using genomic DNA from U. hordei strain 4875-4, S. reilianum SRZ5-1, S. scitamineum SscI8, and M. pennsylvanicum MP4, respectively. Amplified fragments were assembled using the Gibson Assembly kit into KpnI-NotI linearized p123. To generate a complementation construct for the sta1 ortholog from T. cyperi, a synthetic gene codon-optimized for U. maydis was generated (Eurofins Genomics, Ebersberg, Germany) and fused with the 323 bp promoter region of U. maydis sta1 in p123 as described above.
Transformants containing a single copy of SspI linearized complementation constructs inserted in the ip locus of SG200Δsta1 were identified as previously described (Loubradou et al., 2001).

Accepted Article
This article is protected by copyright. All rights reserved genes were amplified by PCR with primers um12226ORF-F and um12226ORF-R, and tin2ox-F2 and tin2entry-R, respectively. Amplified fragments were assembled using the Gibson Assembly kit and introduced into NcoI-AscI sites downstream of the actin promoter. p123_P actin _Sta1_3xHA and p123_P actin _Tin2_3xHA were linearized with AgeI and inserted in single copy into the ip locus of SG200Δsta1 as described (Loubradou et al., 2001).
To generate a strain where mCherry-HA carrying a signal peptide is expressed from the sta1 promoter, the plasmid p123_P sta1 _SP_mCherry_HA was generated as follows. From genomic DNA of SG200, the region containing the sta1 promoter and signal peptide was amplified with primers um12226proGib-F4 and um12226proGib-R6. mCherry-HA was amplified by PCR with primers mCherry_HA_Gib-F3 and mCherry_HA_Gib-R from plasmid p35S-Tin2 26-207 -mCherry-HA-3xNLS (Tanaka et al., 2015a). Amplified fragments were assembled using the Gibson Assembly kit into KpnI-NotI linearized p123. p123_P sta1 _SP_mCherry_HA was linearized with SspI and inserted in single copy into the ip locus of SG200Δsta1 as described (Loubradou et al., 2001).
To generate complementation strains of SG200Δsta1 by sta1 under the UMAG_04033 promoter, the 685 bp promoter region of UMAG_04033 and open reading frame of sta1 were amplified from SG200 genomic DNA by PCR with the primers um04033pro_Gib-F and um04033pro_Gib-R, and um12226ORF_Gib-F and um12226ORF_Gib-R, respectively. Amplified fragments were assembled using the Gibson Assembly kit into KpnI-NotI linearized p123 to yield p123_P UMAG_04033 _Sta1. p123_P UMAG_04033 _Sta1 was linearized with SspI and inserted in single copy into the ip locus of SG200Δsta1 as described (Loubradou et al., 2001). All strains used in this study are listed in Table S4.

Confocal microscopy
The proliferation of U. maydis in infected maize leaf tissue was visualized by confocal microscopy as previously described (Tanaka et al., 2014). One cm of leaf area located 2 cm below the injection hole was excised at 2 days post infection. Leaf samples were destained by ethanol and treated with 10% (w/v) KOH at 85 °C for 4 h. Fungal hyphae were stained with Wheat Germ Agglutinin-Alexa Fluor 488 (WGA-AF488; Invitrogen, Karlsruhe, Germany). Plant cell walls were stained with propidium iodide (Sigma-Aldrich, Schnelldorf, Germany) by incubating decolorized samples in staining solution (1 µg ml -1 propidium iodide, 10 µg ml -1 WGA-AF488) and observed with a TCS-

Accepted Article
This article is protected by copyright. All rights reserved SP8 confocal laser-scanning microscope (Leica Microsystems, Wetzlar, Germany) under the following conditions; WGA-AF488: excitation at 488 nm and detection at 500-540 nm; propidium iodide: excitation at 561 nm and detection at 580-660 nm. To visualize mCherry fusion proteins in the infected maize tissue, plant tissues were directly observed with a TCS-SP8 (Leica Microsystems, Wetzlar, Germany) as previously described (Tanaka et al., 2015a); excitation at 561 nm and detection at 580-630 nm. Plasmolysis was performed as previously described (Tanaka et al., 2014). Fungal plasma membrane was visualized after treatment with 10 µM FM4-64 (Thermo Fisher Scientific, Dreieich, Germany) for 10 min followed by excitation at 514 nm and detection at 650-700 nm. To prepare leaf cross sections, 1 cm x 1 cm regions of infected leaf area were embedded in 5% (w/v) low gelling temperature agarose (Sigma-Aldrich, Schnelldorf, Germany; A9414). Embedded samples were cut at 80 µm thickness by a Leica VT1000S Vibrating Blade microtome (Leica Microsystems, Wetzlar, Germany). Leaf cross sections were directly observed by a TCS-SP5 confocal laser-scanning microscope (Leica Microsystems, Wetzlar, Germany). Plant cell walls were visualized by autofluorescence (excitation at 405 nm and detection at 420-470 nm). Fungal hyphae were visualized by WGA-AF488 staining as described above.

qRT-PCR analysis
Total RNA was extracted with TRIzol reagent (Thermo Fisher Scientific, Dreieich, Germany) from infected maize leaves with SG200 or SG200Δsta1 at 2 dpi by excising 2-3 cm segments from below the injection holes. At least 15 leaf segments were pooled and ground into a fine powder using a mortar/pestle under liquid nitrogen. Quantitative real-time PCR reactions were performed as previously described (Tanaka et al., 2014). All reactions were performed with three biological replicates. Relative gene expression in infected leaf tissues was calculated in relation to the values obtained for GAPDH of Z. mays. Fungal biomass was determined as previously described (Brefort et al., 2014). Primers used for qRT-PCR are listed in Table S3.

Western blot analysis
Hemagglutinin (HA)-tagged proteins in infected leaf extracts were detected as previously described (Tanaka et al., 2014;Tanaka et al., 2019). For immunoprecipitation, 5 ml culture supernatant or

Accepted Article
This article is protected by copyright. All rights reserved protein extract from infected leaf tissue was incubated with 20 µl monoclonal anti-HA−Agarose antibody produced in mouse (Sigma-Aldrich, Schnelldorf, Germany) at 4°C for 2 hr on a rotary shaker. Bound proteins were eluted from the beads by heating to 99 °C for 10 min. Proteins were separated by 12% SDS-polyacrylamide gel electrophoresis. A rabbit polyclonal anti-haemagglutinin antibody (Sigma-Aldrich, Schnelldorf, Germany) was used as the primary antibody at 1:10,000 dilution. Anti-rabbit IgG horseradish peroxidase-linked antibody (Cell Signaling Technology, Leiden, the Netherlands) was used as a secondary antibody at 1:10,000 dilution. To detect signals, SuperSignal West Femto Maximum Sensitivity Substrate (Thermo Fisher Scientific, Dreieich, Germany) was used as substrate for horseradish peroxidase and the signal was visualized by exposure to X-ray film.

Accession numbers
The genes and encoding protein sequences from U. maydis and other smut pathogens are available at NCBI under the following accession numbers: U. maydis sta1 (UMAG_12226), XP_011389862.1;

Identification of Sta1 as a novel virulence-promoting effector in U. maydis
To identify novel virulence effector genes in U. maydis, we focused on gene conservation in several smut species and the expression pattern during plant colonization (Lanver et al., 2018). UMAG_12226 encodes a putative novel secreted protein and RNA-seq data indicate that the gene is specifically

Accepted Article
This article is protected by copyright. All rights reserved 1a). UMAG_12226 encodes a protein of 169 amino acids (18.9 kDa) carrying a signal peptide at the N-terminus predicted by SignalP5.0 (Almagro Armenteros et al., 2019) (Fig. 1b). The protein contains eight cysteine residues without any known motifs or domains (Fig. 1b). To analyze whether UMAG_12226 contributes to virulence, we generated deletion mutants of UMAG_12226 in the solopathogenic strain SG200 of U. maydis and performed virulence assays. The deletion mutants of UMAG_12226 showed a severe reduction of virulence (Fig. 1c), although anthocyanin induction, which is indicative of biotrophic growth, was not impaired (Fig. 1c). The ability to induce large and normal size tumors was most significantly impaired in the mutants and tumors formed were generally small (< 1 mm in diameter) or developed only in ligula tissue (Fig. 1d). Therefore, we designated UMAG_12226 as sta1 (small tumor-associated 1). The introduction of the sta1 gene in single copy in the ip locus of SG200Δsta1 fully complemented all virulence defects (Fig. 1c). Complementation of the deletion mutant was also observed when Sta1 was fused with either a hemagglutinin (HA) epitope tag or mCherry at the C-terminus (Fig. 1d), illustrating functionality of these fusion proteins. In addition, we also generated sta1 mutants in compatible haploid U. maydis strains FB1 and FB2, which infect via dikaryotic hyphae. Similar to SG200Δsta1, the cross of FB1Δsta1 x FB2Δsta1 showed a dramatic reduction of tumor formation in leaf tissue (Fig. S1a,b). However, we could observe tumor development in the stem of infected maize seedlings (categorized as "Heavy tumors"; Fig. S1a,b) and such tumors contained teliospores, which were able to germinate (Fig. S1c,d). This indicates that the sta1 deletion mutants can complete the sexual life cycle.

Host colonization and tumor development are altered when sta1 is deleted
To detect differences in host colonization between SG200 and SG200Δsta1, we visualized fungal hyphae by staining with WGA-AF488 at 2 dpi, when the expression of sta1 reaches its maximum. At this time point, hyphae of SG200 had colonized leaf epidermal cells, passed through mesophyll tissue and reached the bundle sheath cells in vascular structures. However, at 2 and 4 dpi we could neither observe significant differences of fungal colonization between SG200 and SG200Δsta1 by confocal microscopy (Fig. S2a) nor in biomass (Fig. S2b). When fungal structures in infected leaf tissue were visualized by three-dimensional confocal microscopy, it was apparent that SG200 hyphae in discrete vascular bundles were connected (Fig. 2a). In SG200Δsta1, the fungal hyphae had also colonized leaf

Accepted Article
This article is protected by copyright. All rights reserved epidermal cells and reached the bundle sheath cells (Fig. 2a). However, hyphae of SG200Δsta1 in vascular bundles were rarely connected to hyphae in neighboring vascular bundles (Fig. 2a). This suggests that Sta1 is likely to be needed for efficient colonization of tissue adjacent to the vasculature.
Next, we prepared leaf cross sections of plants infected with SG200, SG200Δsta1 and H 2 O as mock control at 6 dpi, when tumor formation is apparent. In tumor tissue colonized by SG200, mesophyll cells were enlarged and bundle sheath cells had largely disappeared due to resumed cell division and endoreduplication (Fig. 2b) as described before (Matei et al., 2018). In contrast, in SG200Δsta1infected leaf area lacking tumors, the structure and size of mesophyll cells were similar to the mock control and bundle sheath cells were unaltered (Fig. 2b). In addition, we found that the cell wall of mesophyll cells infected by SG200Δsta1 showed stronger autofluorescence compared to those infected by SG200 (Fig. 2b). We also compared autofluorescence in SG200-infected leaf areas lacking tumors to corresponding areas infected by SG200Δsta1 (Fig. S3a). Here we could observe stronger autofluorescence in SG200Δsta1-infected tissue than in SG200-infected tissue (Fig. S3b). As this could suggest plant cell wall fortifications, we next analyzed by qRT-PCR the expression of diagnostic genes from the phenylpropanoid pathway in plants infected with either SG200 or SG200Δsta1 at 2 dpi. This analysis revealed that the 4-coumarate CoA ligase (4CL) genes as well as cinnamyl alcohol dehydrogenase (CAD) gene were significantly upregulated after infection with SG200Δsta1 (Fig. S4). The upregulation of these genes could indicate enhanced lignification.

Accepted Article
This article is protected by copyright. All rights reserved Mannhaupt and R. Kahmann, unpublished). In all orthologs, six of the eight cysteine residues are present and their spacing is conserved (Fig. S5). A protein related to Sta1 was also found in Testicularia cyperi, which infects Rhynchospora spp. and belongs to the family Anthracoideaceae ( Fig. S5) (Kijpornyongpan et al., 2018). Although the sequence identity with U. maydis Sta1 is only 30.17 % (Table S5), five of the eight cysteine residues are conserved (Fig. S5). In Ustilaginaceae, the sta1 gene loci are highly syntenic, although in M. pennsylvanicum the distance between sta1 and the neighboring gene on the left is extended (Fig. 3a). An extension of the region between sta1 and the neighboring gene on the left is also seen in T. cyperi (Fig. 3a). In the Brassicaceae-infecting smut Thecaphora thlaspeos, belonging to the family Glomosporiaceae that is considered as a sister taxon of the order Urocystales (Vanky et al., 2008;Courville et al., 2019), an ortholog of sta1 could not be detected. In a phylogenetic analysis, T. thlaspeos is placed distantly from the other plant pathogenic smuts (Fig. 3b) and the region between orthologous neighboring genes to the left (UMAG_11014 in U. maydis) and right (UMAG_03371 in U. maydis) is extended to >100 kb (Fig. 3a).
To investigate whether orthologs are functionally conserved, the U. maydis sta1 deletion mutant was complemented by introducing orthologs from S. reilianum, S. scitamineum, U. hordei and M.
pennsylvanicum. All genes were expressed from the sta1 promoter of U. maydis in SG200Δsta1. All orthologs tested could fully complement the virulence phenotype of the sta1 mutant (Fig. 3c). To examine whether Sta1 function is also conserved in the more distantly related smuts, we complemented SG200Δsta1 by the gene from T. cyperi. In contrast to the orthologs from Ustilaginaceae, we detected only partial complementation in two independent complementation strains (Fig. 3d). This result indicates that the function of Sta1 proteins is fully conserved in Ustilaginaceae smut pathogens while the more distantly related Sta1 protein from T. cyperi shows partial functional divergence.

Secreted Sta1 protein is detected on hyphal cell walls
We studied the localization of Sta1-mCherry-HA protein expressed under the sta1 promoter during leaf infection by confocal microscopy. Here we could observe mCherry fluorescence around biotrophic hyphae, while the fluorescence from a strain expressing cytosolic mCherry-HA under the sta1 promoter accumulated inside fungal cells (Fig. 4a). To visualize whether Sta1 protein resides in

Accepted Article
This article is protected by copyright. All rights reserved the apoplast or is attached to hyphae, we also performed plasmolysis of leaf tissues infected with strains expressing secreted mCherry-HA, Sta1-mCherry-HA or cytosolic mCherry-HA. Plasmolysis should release the plant plasma membrane from the plant cell wall and from hyphae, which are encased by the plant plasma membrane (Doehlemann et al., 2009). Without plasmolysis, secreted mCherry-HA was located in the region surrounding fungal hyphae (Fig. S6), because biotrophic hyphae are encased by the host plasma membrane. However, due to diffusion of secreted mCherry-HA into apoplastic space the fluorescence disappeared after plasmolysis (Fig. S6). Cytosolic mCherry-HA was detected in hyphae also after plasmolysis (Fig. S6). In contrast, Sta1-mCherry-HA was detectable on the surface of fungal hyphae also after plasmolysis (Fig. S6), indicating binding of Sta1 to hyphae.
We generated the strain SG200Δsta1:P actin -Sta1-3xHA that constitutively expresses Sta1-3xHA under the actin promoter and immunoprecipitated Sta1-3xHA protein from culture supernatants. Note that the expression level of the actin promoter is approximately three times higher than the maximum expression level of the sta1 promoter (Lanver et al., 2018). A signal at the expected size of Sta1-3xHA protein (20.71 kDa without signal peptide) was detected (Fig. 4b), showing that Sta1 can be secreted. To investigate whether Sta1-HA produced by biotrophic hyphae of SG200Δsta1:Sta1-HA after colonization is detectable at the expected size, we attempted to immunoprecipitate Sta1-HA protein from leaf tissue infected with SG200Δsta1:Sta1-HA. However, we failed to detect Sta1-HA protein when total protein was extracted by a buffer lacking SDS (Fig. 4c). Using the same extraction buffer after infection with SG200Δsta1:P sta1 -SP-mCherry-HA, mCherry-HA protein could be detected in such extracts (Fig. 4c). However, when the extraction buffer contained 0.1 % SDS, full length Sta1-HA and Sta1-mCherry-HA as well as degradation products could be detected after immunoprecipitation (Fig. 4d). These results suggest that Sta1 protein is attached to biotrophic hyphae.
Next we attempted immunostaining of Sta1-HA in non-permeabilized budding cells or filamentous cells of SG200Δsta1:P actin -Sta1-3xHA induced on parafilm in the presence of hydroxy fatty acids (Mendoza-Mendoza et al., 2009). While we could not detect any fluorescence in budding cells from axenic culture (Fig. 5a), which were previously shown to secrete Sta1-3xHA (Fig. 4b), a strong signal was detected on the surface of filamentous cells on parafilm (Fig. 5b). In contrast, SG200Δtin2:P actin -

Accepted Article
This article is protected by copyright. All rights reserved Tin2-3xHA did not show a signal (Fig. S7). Tin2 is a translocated effector (Tanaka et al., 2014), which consequently should not bind to the fungal cell wall. Budding cells, which failed to form filaments on parafilm, did not show any signal (Fig. S8). As negative control, filamentation of SG200 was induced with hydroxy fatty acids on parafilm, but in this case no signal could be detected (Fig.   5b). We also showed that budding cells taken from a colony of SG200Δsta1:P actin -Sta1-3xHA grown on a PD plate did not show any fluorescence signal (Fig. S9). Taken together, these results indicate that Sta1 protein specifically attaches to U. maydis filaments but fails to bind to budding cells.
Furthermore, we simultaneously visualized Sta1-3xHA protein by immunostaining and the fungal plasma membrane by FM4-64 in non-permeabilized filaments of SG200Δsta1:P actin -Sta1-3xHA. We could observe the Sta1-3xHA signal as well as the FM4-64 signal at the cell periphery but the two signals were not overlapping (Fig. 6a). However, after plasmolysis, the two signals separated and the Sta1-3xHA signal remained at the cell periphery, while the FM4-64 signal was now internalized (Fig.   6b). The latter is characteristic for vesicular uptake, which is a fast reaction taking only minutes (Wedlich-Soldner et al., 2000). These results indicate that Sta1 resides in the hyphal cell wall.

Overexpression of Sta1 protein alters cell wall integrity in filamentous cells
While hydrophobicity of SG200, SG200Δsta1:P actin -Sta1-3xHA and SG200Δtin2:P actin -Tin2-3xHA was indistinguishable (Fig. S10), filaments extending from colonies formed by SG200Δsta1:P actin -Sta1-3xHA were shorter compared to SG200 and SG200Δtin2:P actin -Tin2-3xHA (Fig. 7a), suggesting that filamentous cells of SG200Δsta1:P actin -Sta1-3xHA might have an altered cell wall. To examine whether this results in an altered sensitivity to abiotic stresses, we spotted serial dilutions of SG200, SG200Δsta1:P actin -Sta1-3xHA and SG200Δtin2:P actin -Tin2-3xHA on PD-charcoal plates containing stressors including Congo red, Calcofluor white, H 2 O 2 and SDS. While we could not detect significant differences among the three strains in the presence of Calcofluor white, H 2 O 2 and SDS ( Fig. 7b), SG200Δsta1:P actin -Sta1-3xHA showed a severe reduction of filamentation in the presence of Congo red (Fig. 7b). However, growth of budding cells of these strains was not affected by Congo red on a PD-Congo red plate (Fig. S11).

Accepted Article
This article is protected by copyright. All rights reserved SG200Δsta1:P actin -Sta1-3xHA were efficiently lysed by chitinase and β-glucanase, while SG200 and SG200Δtin2:P actin -Tin2-3xHA were not (Fig. 7c). Upon chitinase treatment, the filamentous cells of SG200Δsta1:P actin -Sta1-3xHA showed chains of rounded structures (Fig. 7d) while filaments of SG200 were mostly unaffected (Fig. 7d). In contrast, we did not observe these structures after βglucanase treatment (Fig. S12). Overall, these results indicate that the constitutive expression of Sta1 protein in filaments results in an altered cell wall structure. When yeast-like colonies of the same three strains were treated with chitinase and β-glucanase in the same concentrations as used for filamentous cells, none of the three strains was lysed even when higher concentrations of chitinase and β-glucanase were used (Fig. S13). These results show that constitutive expression of Sta1 influences susceptibility to β-glucanase and chitinase specifically in filamentous cells.
To investigate whether cell wall alterations can also be visualized when comparing SG200 and SG200Δsta1, we first performed RT-PCR on RNA from SG200 filaments grown on a PD-charcoal plate. The sta1 expression was detectable, while tin2 effector gene expression was not (Fig. S14a).
Filamentous colonies and filaments were not distinguishable between SG200 and SG200Δsta1 (Fig.   S14b). Furthermore, we could observe significant differences neither in growth between SG200 and SG200Δsta1 on PD-charcoal plates containing different stressors (Fig. S14c) nor in sensitivity to chitinase and β-glucanase (Fig. S14d). In particular, no differences in sensitivity to Congo red were apparent between SG200 and SG200Δsta1. We consider it likely that SG200 produces only small amounts of Sta1 under these conditions, which does not allow to detect phenotypic differences to SG200Δsta1.

Altered timing of sta1 expression interferes with the function of Sta1
Since the expression of Sta1 peaks at an early stage of plant infection (Fig. 1a), we hypothesized that Sta1 might be necessary at this specific infection stage. To test this hypothesis, we introduced the sta1 gene in strain SG200Δsta1 under the UMAG_04033 promoter that is induced at a later infection stage (with maximum expression after 8 days post infection) but shows an expression level similar to sta1 (Fig. 8a). Two independent SG200Δsta1 derivatives expressing sta1 under this late promoter did not show complementation of virulence (Fig. 8b). To examine whether constitutive expression of Sta1 influences virulence, we also performed virulence assay of SG200Δsta1:P actin -Sta1-3xHA. In this

Accepted Article
This article is protected by copyright. All rights reserved strain, we observed only weak complementation of virulence (Fig. S15). These results suggest that Sta1 function is needed specifically in a certain time window during infection.

Discussion
In this study, we have characterized the virulence-promoting Sta1 effector as a protein that needs to be expressed in a short time window during plant colonization and specifically attaches to fungal hyphae. We found that Sta1 orthologs exist in all Ustilaginaceae but also in T. cyperi, a fungus belonging to the distantly related Anthracoideaceae family in the order Ustilaginales (Kijpornyongpan et al., 2018). However, while four orthologs from Ustilaginaceae fully complemented the virulence phenotype of U. maydis sta1 mutants, the T. cyperi ortholog could only partially complement, suggesting functional divergence. Our finding of partial complementation by the T. cyperi ortholog of sta1 is the first example for functional effector complementation from such a distantly related species, suggesting an ancient function of this effector.
In contrast to secreted proteins that bind to cell wall polysaccharides in other plant pathogenic fungi (van den Burg et al., 2006;de Jonge et al., 2010;Marshall et al., 2011;Mentlak et al., 2012;Takahara et al., 2016), Sta1 does not contain any characterized domains or motifs that implicate carbohydrate binding. At this point, we cannot exclude the possibility that Sta1 might bind to another protein whose expression occurs in the filamentous form only. Currently only two proteins of this class are known.
One is the repellent protein Rep1, which provides for surface hydrophobicity but is not needed for virulence (Wosten et al., 1996) and is therefore unlikely to be a binding partner for Sta1. A second protein is Rsp3, a virulence-promoting effector that does not contain a carbohydrate-binding domain protein to the cell wall has altered the cell wall properties of hyphae. In line with this, we observed that filaments that constitutively express Sta1 also show increased susceptibility to chitinase and βglucanase, which are enzymes degrading chitin and β-glucan respectively. This makes it likely that Sta1 has altered the cell wall structure or has increased accessibility of the chitin and glucan layer to attack by these enzymes. However, we cannot formally rule out the possibility that the increased susceptibility to these enzymes is due to the about 3-fold higher than native expression of Sta1 conferred by placing sta1 under the actin promoter. In contrast to chitinase, which induced changes to intracellular structure, β-glucanase treatment did not induce such a structural phenotype in filaments constitutively expressing Sta1. We speculate that disruption of the inner chitin layer triggers a much more severe stress response than the response triggered by removal of β-glucan in the outer layer of the fungal cell wall as previously reported in filaments of another basidiomycete fungus (Mogilnaya et al., 2017).

Accepted Article
This article is protected by copyright. All rights reserved The peak of sta1 expression is at 2 dpi, a time point prior to tumor development in leaf tissue, and subsequently sta1 expression levels decrease gradually as fungal infection progresses. When sta1 was expressed from a late promoter (expression peak at 8 dpi), or when sta1 was constitutively expressed, the virulence-promoting function was compromised, suggesting that correct timing of sta1 expression is necessary. This could reflect that attachment of Sta1 protein to hyphae already during penetration might increase vulnerability of fungal hyphae to plant-derived cell wall degrading enzymes. Maize chitinases and β-glucanases are upregulated during early colonization stages but downregulated at 24 hpi and later time points (Doehlemann et al., 2008), which is likely achieved by secreted effectors. In this respect it would be interesting to analyze whether the strongly reduced virulence of the strain constitutively expressing sta1 is caused by developmental defects of hyphae in epidermal tissue.
The proposed benefit of cell wall modification by attachment of Sta1 protein to hyphae of U. maydis when expressed from the native promoter is not clear yet. With respect to growth, the sta1 mutant was comparable to wild type at 2 dpi, suggesting that the mutant does not have critical morphological defects and appears fit for survival in leaf tissue. However, detailed confocal microscopy found that biotrophic hyphae of the sta1 mutant had reached the vascular bundles but were less efficient in spreading to regions between vascular bundles. This could suggest that mutant hyphae, which lack

Accepted Article
This article is protected by copyright. All rights reserved the phenotype observed, we cannot exclude the possibility that the mutant effects seen are of secondary nature.
These studies have allowed us to classify Sta1 as a novel type of effector that can bind specifically to U. maydis hyphae, and which likely needs to be bound to hyphae at a specific developmental stage during plant colonization. We speculate that this may go along with alterations in fungal cell wall properties that prevent the elicitation of plant defense responses in a defined tissue type and time window after infection.
The following Supporting Information is available for this article: Fig. S1. Virulence of deletion mutants of sta1 in haploid strains.              Table S1. Raw data for infection assays. Table S2. Plasmids used in this study. Table S3. Primers used in this study.

Accepted Article
This article is protected by copyright. All rights reserved