FGB1 and WSC3 are in planta-induced b-glucan-binding fungal lectins with different functions

In the root endophyte Serendipita indica, several lectin-like members of the expanded multigene family of WSC proteins are transcriptionally induced in planta and are potentially involved in b-glucan remodeling at the fungal cell wall. Using biochemical and cytological approaches we show that one of these lectins, SiWSC3 with three WSC domains, is an integral fungal cell wall component that binds to long-chain b1-3-glucan but has no affinity for shorter b1-3or b1-6-linked glucose oligomers. Comparative analysis with the previously identified b-glucan-binding lectin SiFGB1 demonstrated that whereas SiWSC3 does not require b1-6-linked glucose for efficient binding to branched b1-3glucan, SiFGB1 does. In contrast to SiFGB1, the multivalent SiWSC3 lectin can efficiently agglutinate fungal cells and is additionally induced during fungus–fungus confrontation, suggesting different functions for these two b-glucan-binding lectins. Our results highlight the importance of the b-glucan cell wall component in plant–fungus interactions and the potential of b-glucan-binding lectins as specific detection tools for fungi in vivo.


Introduction
Plant root-associated fungi thrive in challenging and rapidly changing environments. Their ability to colonize their hosts depends, among others, on their capacity to remodel the cell surface to withstand biotic and abiotic stresses and to limit plant immune recognition. The fungal cell wall (CW) is the first cellular structure that is exposed to the plant host and to other microbes. CW-derived polysaccharides, such as chitin and bglucans, are potent elicitors of plant immune responses and thus their detection needs to be prevented while maintaining CW integrity for successful colonization (Rovenich et al., 2016;Geoghegan et al., 2017;Latge et al., 2017;Hopke et al., 2018). Additionally, CW integrity plays a role in the response to other microbes in the soil. To avoid recognition of CW-derived polysaccharides and to limit stimulation of plant defense responses, fungi have evolved different strategies such as CW remodeling, masking, shielding and manipulation of glycaninduced host defense signaling (El Gueddari et al., 2002;van den Burg et al., 2006;de Jonge et al., 2010;Marshall et al., 2011;Fujikawa et al., 2012;Mentlak et al., 2012;Oliveira-Garcia & Deising, 2013Sanchez-Vallet et al., 2013;Fesel & Zuccaro, 2016a;Takahara et al., 2016;Wawra et al., 2016;Melida et al., 2018). Whereas several mechanisms of chitin masking/ shielding and avoidance of plant host immune perception are known, fungal mechanisms dedicated to the modulation of bglucan recognition have only been described recently (Emsley & Cowtan, 2004;Oliveira-Garcia & Deising, 2013Wawra et al., 2016). In the maize (Zea mays) pathogen Colletotrichum graminicola, synthesis of b-glucan is rigorously downregulated during biotrophic development in the plant host possibly leading to a depletion of this polymer at the CW of biotrophic hyphae (Oliveira-Garcia & Deising, 2016). A further strategy for evading b-glucan-triggered immunity is given by the fungal specific bglucan-binding lectin FGB1 of the root endophyte Serendipita indica (Si, syn. Piriformospora indica). SiFGB1 was shown to bind b-glucan with high specificity and to be capable of suppressing b-glucan-triggered immunity in different plants (Wawra et al., 2016).
Sequencing of several genomes from root-associated fungi uncovered an expansion in the sebacinoid genomes of genes encoding proteins with carbohydrate-binding properties (Zuccaro et al., 2011;Lahrmann & Zuccaro, 2012;Kohler et al., 2015). The physiological relevance of this expansion is unclear but a large set of these genes are transcriptionally induced during root colonization of different hosts suggesting that they contribute to the endophytic lifestyle of these fungi (Zuccaro et al., 2011;Lahrmann & Zuccaro, 2012;Kohler et al., 2015;Lahrmann et al., 2015;Fesel & Zuccaro, 2016b). Specifically, genes encoding lectin-like proteins such as the chitin-binding LysM (Lysin Motif) proteins which are known as suppressors of host immunity, the cellulose-binding CBM1 (Carbohydrate-Binding Module Family 1) proteins, which are potentially involved in loosening of the plant CW, and proteins with WSC domain(s) (cell wall integrity and stress response components) are remarkably enriched (Gaulin et al., 2002;Saloheimo et al., 2002;de Jonge & Thomma, 2009;Lahrmann & Zuccaro, 2012;Kohler et al., 2015). The S. indica genome encodes for 36 WSC proteins with 23 of those being significantly differentially expressed during colonization of plant roots (Zuccaro et al., 2011;Lahrmann & Zuccaro, 2012). Twenty-eight of the WSC domain-containing proteins are predicted to be lectins devoid of known enzymatic domains (Goldstein et al., 1980;Gabius et al., 2002;Zuccaro et al., 2011). Even though their functions in plant-microbe interaction have not been analyzed so far, at least one lectin with WSC domains can be found among plant responsive genes in several root-associated fungi (Dore et al., 2015;Kohler et al., 2015). This implies a role for WSC lectins in plant colonization.
Proteins with a WSC domain were first described as cell surface sensors involved in detecting and transmitting CW status to the cell wall integrity (CWI) signaling pathway in S. cerevisiae (Verna et al., 1997;Lodder et al., 1999). These S. cerevisiae WSC proteins possess small C-terminal cytoplasmic domains, a single transmembrane domain, a WSC domain at the N-terminus and a periplasmic ectodomain rich in Ser/Thr residues proposed to function as mechanosensors of the extracellular matrix (Rajavel et al., 1999;Philip & Levin, 2001). Proteins with WSC domain (s) and a transmembrane anchor are also involved in the activation of the CWI pathway in the yeast Kluyveromyces lactis, the filamentous fungi Aspergillus nidulans, Neurospora crassa, Beauveria bassiana and in the brown algae Fucus serratus (Rodicio et al., 2008;Futagami et al., 2011;Maddi et al., 2012;Herve et al., 2016;Tong et al., 2016). WSC domains also are present in the fungal b1-3-exoglucanase of Trichoderma harzianum (Cohen-Kupiec et al., 1999), suggesting that this domain mediates binding to fungal CW-associated carbohydrates. Yet, no biochemical information about their glycan interactors is available. Interestingly, also in the nematode-trapping fungus Monacrosporium haptotylum an expansion of genes encoding proteins with WSC domains was found. The authors proposed an involvement of these proteins in adhesion to fungal cells (Andersson et al., 2013). A WSC domain also was found in human polycystin 1 (PKD1), a plasma membrane protein that is defective in autosomal dominant polycystic kidney disease (Ponting et al., 1999), indicating that this domain is conserved from yeasts to mammalian cells but it is not present in higher plants.
Here we report on the biochemical proprieties, carbohydratebinding affinities and localization of two plant responsive lectins from S. indica, the~39 kDa integral CW component WSC3 with three WSC domains (PIIN_05825) and the previously identified 6.2 kDa plant immune suppressor SiFGB1 (PIIN_03211; Wawra et al., 2016). These two lectins are induced in planta, bind b-glucans with diverse carbohydrate-binding motifs that share no sequence homologies to each other and have different functions during plant colonization.

Materials and Methods
Detailed description of the identification and phylogenetic analyses of lectin-like proteins in fungal genomes, fungal colonization assays, assessment of gene expression, adhesion of Serendipita indica spores to barley roots in the presence of WSC3, confocal microscopy, cloning of SiWSC3 and transformation of S. indica, SDS-PAGE and Western blotting, cloning and expression of SiWSC3-His, expression and purification of SiFGB1-His, FITC488 labeling of SiWSC3-His and native SiFGB1, circular dichroism spectroscopy of SiWSC3-His, SiWSC3-His pull-down assay with cell wall (CW) preparations, enzymatic extraction of CW proteins and CW stress assay in Pichia pastoris and primer list can be found in Supporting Information Methods S1.

Fungal strains, plant material and growth conditions
The dikaryotic S. indica wild-type (WT) strain DSM11827 (Leibniz Institute DSMZ-German Collection of Microorganisms and Cell Cultures, Braunschweig, Germany) and the dikaryotic S. indica transformants carrying a hygromycin resistance gene were cultivated at 28°C on solid or liquid complex medium (CM) with shaking at 120 rpm with or without hygromycin (80 lg ml À1 final concentration; Carl Roth, Karlsruhe, Germany) as described in Hilbert et al. (2012). Additionally, a homokaryotic S. indica transformant carrying a geneticin resistance gene was used as reference for experiments including the S. indica homokaryotic transformant m5 expressing FGB1:GFP (Wawra et al., 2016; GFP, green fluorescent protein). The Colletotrichum tofieldiae strain Ct61  was kindly provided by Paul Schulze-Lefert from the Max Planck Institute for Plant Breeding Research, Cologne and Soledad Sacrist an from the Universidad Polit ecnica de Madrid, and propagated on solid CM supplemented with 1.5% agar in darkness at 25°C. The haploid solopathogenic Ustilago maydis strain SG200 was grown in liquid YEPS light (0.4% (w/v) yeast extract, 0.4% (w/v) peptone, 2% (w/v) sucrose) as described in Kamper et al. (2006).
For the fungal confrontation experiments S. indica and Bipolaris sorokiniana strain ND90Pr (Leibniz Institute DSMZ-German Collection of Microorganisms and Cell Cultures) were grown in liquid MYP medium (0.7% (w/v) malt extract, 0.1% (w/v) peptone, 0.05% (w/v) yeast extract) at 28°C and 120 rpm of shaking for 4 or 3 d, respectively. The mycelium of both fungi was filtered through a Miracloth filter (Merck, Darmstadt, Germany) and washed three times with sterile ddH 2 O before disrupting the mycelial aggregates with a blender (Kinematica, Lucerne, Switzerland). The crushed mycelium was regenerated for 1 d at 28°C at 120 rpm of shaking in fresh MYP medium. For the confrontation assay the mycelium was again filtered through a Miracloth filter and 0.5 g each of S. indica and B. sorokiniana mycelium were mixed together, re-suspended in

Research
New Phytologist 5 ml sterile ddH 2 O and added to 30 g of autoclaved Cologne land soil (CAS10). As a control, 1 g of filtered S. indica mycelium was re-suspended in 5 ml sterile ddH 2 O and mixed with 30 g of soil. After 42 h at 28°C the fungal mycelium was harvested from the surface of the soil, flash frozen in liquid nitrogen and stored at À80°C until RNA was extracted.
For the fungal colonization experiments Arabidopsis thaliana Col-0 and Hordeum vulgare Golden Promise were cultivated and inoculated with fungal spores as described previously (Wawra et al., 2016). Arabidopsis roots from three square petri plates each containing 20 plants were pooled per biological replicate before DNA or RNA extraction. Roots from four barley plants grown in a single jar were pooled per biological replicate.

Transmission electron microscopy and quantification of glycan labeling
The dikaryotic S. indica WT strain, the homokaryotic S. indica reference transformant, the dikaryotic S. indica WSC3-GFP transformant T 3 and the homokaryotic S. indica FGB1:GFP transformant m5 were grown on solid CM plates for 3 wk. For TEM analysis, samples were washed and subsequently fixed in 2.5% glutaraldehyde + 2% paraformaldehyde in 0.05 M sodium cacodylate buffer, pH 6.9, for 2 h at room temperature followed by an overnight incubation at 4°C. After thorough rinsing with 0.05 M sodium cacodylate buffer, the samples were post-fixed for 1 h on ice with 0.5% osmium tetroxide in 0.05 M sodium cacodylate buffer, pH 6.9, supplemented with 0.15% potassium ferricyanide. Thereafter, the samples were again thoroughly rinsed with 0.05 M sodium cacodylate buffer and dehydrated in an ethanol series from 10% to 100%, then in different ethanol : acetone mixtures and finally in 100% acetone. Subsequently, samples were infiltrated with 25% Araldite 502/EmBed 812 (EMS, Hatfield, PA, USA) in acetone. Further resin infiltration and final embedding was performed with the help of the EMS poly III, an evaporation-controlled automated embedding and polymerization device (EMS). Ultrathin sections (70-90 nm) were prepared as described previously (Micali et al., 2011;Kleemann et al., 2012). For the detection of b1-3-glucan, the samples were immunogold-labeled as described in Micali et al. (2011) using a 1 : 100 dilution of the mouse monoclonal anti-b1-3-glucan antibody (Cat. no. 400-2; Biosupplies Australia Pty, Parkville, Vic., Australia). Detection of chitin was performed using undiluted WGA conjugated with 10 nm gold particles (EY Laboratories, San Mateo, CA, USA) for 3 h at room temperature, followed by thorough rinsing with TRIS buffer and water. Sections were stained with 0.1% (w/v) potassium permanganate in 0.1N sulfuric acid for 1 min (Sawaguchi et al., 2001). Transmission electron microscopy (TEM) was performed using a Hitachi H-7650 TEM (Hitachi, Krefeld, Germany) operating at 100 kV. The acquired pictures were further analyzed using the FIJI software (Schindelin et al., 2012). To quantify chitin and b1-3-glucan within the fungal CW the corresponding gold particles were counted from~50 TEM images per fungal strain per treatment. After counting the gold particles, the length of the CW was quantified and the number of gold particles per lm of CW was calculated for each individual image.

Quantitative assay for lectin-induced cytoagglutination
The ability of WSC3, FGB1 and WGA to aggregate fungal cells/ spores was investigated for B. sorokiniana, S. indica, and U. maydis. Ustilago maydis was grown over night in 5 ml YEPS light and the OD 600 of the culture was measured subsequently. The culture was diluted with fresh YEPS light to an OD 600 of 0.4. B. sorokiniana spores were diluted to 500 spores ml À1 in MYP medium and S. indica spores were diluted to 50 000 spores ml À1 in CM. One hundred microliters of the cell/spore solutions were transferred to individual wells of a 96-well plate. The recombinant SiWSC3-His was sterile-filtrated using a 0.22-lm filter and was added to the individual wells with a final concentration of 10 lM. As controls, 10 lM WGA-AF594 (Invitrogen), 10 lM native FGB1 or 10 lM WSC3-FITC488 were added to the individual wells. As a mock control, sterile ddH 2 O was used. Sterile ddH 2 O was added to a final volume of 150 ll to each well.
Phenotypic assessment of U. maydis sporidias was done microscopically after 4 h of incubation at 28°C with 250 rpm of shaking. The agglutination effect was quantified by counting the aggregated U. maydis cells relative to the total number of cells. An aggregate was defined as a structure with two or more cells being in direct contact with each other but not connected, as for example in dividing cells.
Fluorescence labeling of WSC3-His and nFGB1 and Microscale Thermophoresis (MST) Fluorescent labeling of WSC3 for MST measurements was done using the Biotinum CF594 succinimidyl ester protein labeling kit (#92216) according to the manufacturer's protocol. Native FGB1 was labeled using the Lightning-Link ® (R-PE) Kit (703-0015; Innova Biosciences, Expedeon, San Diego, CA, USA) according to the supplier's instructions. Data were recorded on a Monolith NT.115 instrument in standard coated capillaries using a fluorescence excitation of 510-550 nm and emission detection at 560-590 nm. WSC3-His CF594 was used at a concentration of 60 nm (for pre-tests) or 500 nM (for binding affinity measurements) with the MST power set to high at a laser intensity of 60%. Buffers were either 37.5 mM MES + 75 mM NaCl + 0.05% Tween 20 pH 5.0 or phosphate buffered saline containing 0.05% Tween 20 at pH 7.4. For nFGB1 R-PE a protein concentration of 20 nM was used in 25 mM MES buffer pH 5.0 containing 1.25 mg ml À1 BSA, 0.25% glycerol, 0.5% Tween 20 and 50 mM NaCl. Binding affinities were measured through a series of 16 successive 29 dilutions of the ligand stock solutions dissolved in the respective assay buffers.

Isothermal titration calorimetry (ITC)
Isothermal titration experiments were performed using a VP-ITC instrument at 20°C. The instrument was heat-pulse-calibrated and the protein samples were extensively dialyzed against water www.newphytologist.com before use. Titrant stock solutions were prepared with the same batch of water as used for dialysis. All solutions used were degassed before filling the sample cell and syringe. The ITC stirring speed was set to 300 rpm; the feedback gain mode was set to medium. Because the initial injection generally delivers inaccurate data, the first step was omitted from the analysis. The collected data were analyzed using the program ORIGIN (MicroCal, Malvern Panalytical, Malvern, UK) and binding isotherms were fitted using the binding model provided by the supplier. Errors correspond to the SD of the nonlinear least-squares fit of the data points of the titration curve. For the titration of WSC3-His to laminarin a 18.5 lM protein bait solution was titrated with a 1 mM laminarin solution. The first injection (1 ll) was followed by 29 titrations with 6 ll each and 150 s of spacing. The titration curve was baseline-corrected and subtracted with the data from the control titration of the laminarin stock into water. For testing of chitohexaose binding a WSC3-His bait concentration of 16 lM was used with a chitohexaose stock solution of 1.5 mM. Laminarihexaose and gentiobiose (1 mM stock each) were titrated against 20 lM WSC3-His bait solution.

Enzymatic preparation of debranched beta-glucan from laminarin
In order to generate linear b1-3-glucan without b1-6-glucose side chains we used a biocatalytic strategy as described previously (Becker et al., 2017). In brief, 100 mg of laminarin from Laminaria digitata (Sigma) was hydrolyzed overnight at 37°C with 100 nM purified enzyme (~5 lg ml À1 ) of a b1-6-exo glucosidase of glycoside hydrolase family 30 (GH30) that specifically removes the b1-6-linked glucose side chains from laminarin. This reaction leads to glucose and the linear b1-3-glucan. Completeness of the conversion was confirmed by testing the activity of a b1-3endo glucanase of family GH17 that only cleaves undecorated, linear b1-3-glucan (Becker et al., 2017;Unfried et al., 2018) and shows little activity on the native yet high activity on the debranched laminarin. The reaction was stopped by boiling the sample for 5 min at 100°C. Precipitated protein was removed by filtration through 0.2 lm Costar Spin-X Filters (Corning, Kaiserslautern, Germany). The glucose was separated from the b1-3-glucan by size exclusion chromatography using a HiTrap Desalting column (GE Healthcare, Solingen, Germany) according to the manufacturer's instructions. The column was equilibrated and eluted with Milli-Q water. The solution was dried in vacuum overnight at 45°C to obtain a white powder. The enzymatic digestion with both enzymes and absence of glucose in the final product was recorded by high performance anionic exchange chromatography with pulsed amperometric detection (HPAEC-PAD) on a Dionex system (Unfried et al., 2018).

SiWSC3 and SiFGB1 alter CW properties
Comparative genomics and phylogenetic analysis identified a strong taxa-specific expansion for the lectin-like WSC proteins in some fungal genomes, symptomatic of a rapid evolution (Figs 1, S1, S2; Table S1). This was especially evident in the sebacinoid fungi and in the saprotrophic white rot fungus Auricularia subglabra (Figs 1, S1, S2). From transcriptional analyses we identified a plant-responsive clade of WSC lectin-like members in the sebacinoid fungi (Lahrmann et al., 2015) (Fig. 1b,c, S1). Among this subset SiWSC3 (PIIN_05825) was highly transcriptionally induced and was selected for further functional characterization. Expression analysis by quantitative reverse transcription polymerase chain reaction (qRT-PCR) confirmed the expression pattern of SiWSC3 during plant root colonization in Arabidopsis and barley with stronger expression in the latter. Additionally, induction of SiWSC3 was detected during contact of S. indica with the plant pathogenic fungus B. sorokiniana in a soil confrontation assay (Fig. 2a). Because WSC proteins were proposed to bind to the CW of fungi and SiWSC3 harbors a predicted secretion signal peptide, we hypothesized that SiWSC3 may localize to the CW of S. indica. To study its subcellular localization the SiWSC3 gene was expressed as fusion with a C-terminal GFP tag under the control of the S. indica FGB1 promoter which is highly active in planta and in complex medium (CM) but less active in other axenic media (Wawra et al., 2016). Mycelium and culture filtrate of five independent S. indica transformants grown in CM and MYP were tested for secretion of the SiWSC3:GFP fusion protein (Fig. 2b). The highest amount of full-length SiWSC3:GFP fusion protein was detected in the mycelial sample of the transformants T 1 and T 3 grown in CM. The full-length fusion protein was not detectable by Western blot in the culture filtrates but the presence of free GFP indicated that secretion occurred. The full-length and cleaved SiWSC3 protein versions most likely remained bound to carbohydrates present at the surface of the fungal cells in the mycelium fraction (Fig. 2b). To verify this hypothesis cytological analysis were performed with the S. indica transformant T 3 which produced a higher amount of the SiWSC3:GFP fusion protein and the transformant T 2 for which no band was detected in the Western blot. Additionally, the fungal transformant m5, harboring a SiFGB1:GFP fusion construct under the control of the S. indica FGB1 promoter, was used as control. SiFGB1:GFP was previously shown to localize to the fungal CW and to additionally be one of the most abundant proteins secreted by S. indica in the culture filtrate (Wawra et al., 2016). The SiWSC3:GFP fluorescence signal co-localized with the signal of the chitin-binding lectin WGA-AF594 at the CW for transformant T 3 (Fig. 2c), whereas no specific fluorescence could be detected for the transformant T 2 using identical confocal laser scanning microscope settings (Fig. 2d). SiFGB1: GFP fusion showed in addition to the CW localization, a strong signal at the fungal septa and in structures that resembled the endoplasmic reticulum (Fig. 2e) (Rico-Ramirez et al., 2018). The localization of the SiWSC3:GFP and SiFGB1:GFP fluorescence signal at the fungal CW suggests that modification of both lectins at the C-terminus does not severely impact ligand-protein binding.
The effects of overexpression of these lectins on the CW polysaccharide composition was assessed by transmission electron microscopy (TEM) using chitin-and b1-3-glucan-specific immunogold-labeling (Mayhew, 2011

Research
New Phytologist corresponding controls labeled either with gold-conjugated WGA or with the b1-3-glucan-specific antibody. Whereas goldconjugated WGA labels for chitin was constant in all fungal samples (Figs 2f, S3), the amount of gold labels for b1-3-glucan was significantly increased in the SiFGB1:GFP overexpressing transformant m5 compared to the control (Figs 2g, S3). The data obtained for the transformant m5 corroborate previous results obtained by NMR where the ratio between chitin and glucan was found to be altered in the CW of this fungal transformant (Wawra et al., 2016). The amount of gold labels for b1-3-glucan was significantly reduced in the CWs of the SiWSC3:GFP overexpressing transformant T 3 which could result from alteration in glucan composition or in availability of b1-3-glucan to the antibodies. In both cases deregulation of SiWSC3 leads to alteration of CW properties which are different from those observed for SiFGB1. As anticipated by the confocal microscopy analysis, TEM analyses indicate that chitin is found prevalently at the CW of S. indica hyphae (Figs 2f, S3), whereas b1-3-glucan also is abundantly present/exposed at the septa and in the septal pore swellings of the dolipore (Figs 2g, S3).

SiWSC3 and SiFGB1 bind to b-glucan polysaccharides in a different manner
In order to biochemically characterize the SiWSC3 protein, a His-tag fusion (SiWSC3-His) was heterologously produced in the yeast Pichia pastoris under the control of the AOX1 (a) Number of predicted lectin-like proteins in the genome of sebacinoid fungi compared to the average values of these proteins in different fungal genomes that are grouped based on their predominant lifestyle and colonized tissue into symbionts (root, shoot, lichens), plant pathogens (shoot, root + shoot), animal pathogens, saprotrophs, and Serendipita indica, S. vermifera and S. herbamans. Functional protein domains involved in carbohydrate binding were predicted from 79 fungal genomes (listed in Supporting Information Table S1) using the Pfam database (Finn et al., 2010). Lectin-like proteins were defined as proteins that only contain one or a combination of the shown nonenzymatic domains and were identified from PfamScan output with custom Java applications. The number of proteins shown is the average over all fungal genomes belonging to one of the groups (S. indica, S. vermifera and S. herbamans were not included in the average). Bar charts were created using GNUPLOT promoter. Western blot analysis and enzymatic deglycosylation revealed that this protein is abundantly secreted and is glycosylated in this fungus with an apparent retention on SDS-PAGE corresponding to a molecular weight of~55 kDa before and~3 9 kDa after deglycosylation (Fig. S4a,b). Additionally, in P. pastoris expression of SiWSC3-His leads to its incorporation into the CW and to increased resistance to the CW stressors Calcofluor White and Congo Red (Fig. S4c,d).

New Phytologist
In order to determine whether SiWSC3 specifically binds to fungal CW polysaccharides a pull-down experiment was performed using insoluble, protein-free polysaccharide preparations of H. vulgare and S. indica CWs. In this experiment, SiWSC3-His co-precipitated with the S. indica but not with the H. vulgare CW polysaccharide preparations (Fig. S4e). This suggests a specificity of SiWSC3 for an oligosaccharide or polysaccharide of fungal origin as it was shown previously for SiFGB1 (Wawra et al., 2016). Therefore, the heterologously produced SiWSC3-His was used for ITC analysis to determine its affinity to different oligoand polysaccharide ligands possibly found in fungal CWs. Accordingly, a soluble b1-3-glucan with b1-6-linkages consisting of~30 glucose units (laminarin), a linear b1-3-glucan hexamer (laminarihexaose), a b1-6-linked glucose dimer (gentiobiose) and a chitin octamer (chitooctaose) were included in the survey. Upon titration of the soluble laminarin to SiWSC3-His in water an exothermal binding reaction was observed with a K d value of 12.5 lM AE 8.8 lM. No significant binding was detected for the other polysaccharides tested (Fig. 3a). Circular dichroism (CD) spectroscopy showed that neither laminarin binding nor pH affected the secondary structure of SiWSC3 (Fig. S5), suggesting that SiWSC3 has a preformed carbohydrate-binding site that can accommodate the b-glucan ligand. The ITC measurement also revealed an apparent stoichiometry for the reaction of 1 : 3 meaning that one SiWSC3 binds three b1-3/1-6-glucan molecules. This suggests that each of the three WSC-domains may bind one b1-3/1-6-glucan molecule. Alternatively, it is possible that the three WSC domains act together to increase the binding affinity to higher order b-glucan structures such as the triple-helical structure found for laminarin in solution (Sletmoen & Stokke, 2008;Kanagawa et al., 2011). Because SiWSC3 did not bind to any of the shorter glucose oligomers, which cannot assume a triple-helical conformation in solution, it is unclear whether the fungal specific b1-6-glycosidic bonds are important for the binding of SiWSC3-His to laminarin. To clarify if the b1-6-side chains of laminarin are required for binding to WSC3 and FGB1, we performed carbohydrate-binding studies in aqueous solution with microscale thermophoresis (MST). We tested laminarin and size exclusion chromatography purified debranched laminarin after treatment with the FbGH30 enzyme derived from the bacterium Formosa sp. strain B. This enzyme was previously shown to specifically hydrolyze the b1-6-linked glucose side chains of laminarin (K M : 3.1 AE 0.2 mM and K cat /K M : 21124 M À1 s À1 ) producing linear b1-3-glucan (Becker et al., 2017;Unfried et al., 2018). The MST analysis at pH 5 showed that whereas FGB1 binds to native laminarin this lectin does not bind to debranched laminarin. WSC3 on the contrary can bind to native and debranched laminarin with similar affinities (Figs 3b, S5, S6). These data suggest that FGB1 but not WSC3 requires the b1-6 side chains of laminarin for efficient binding. This is in agreement with the fact that FGB1 but not WSC3 can bind to gentiobiose, a disaccharide composed of two units of D-glucose joined with a b1-6 linkage ( Fig. 3a and Wawra et al., 2016).
A number of additional potential ligands for WSC3 were explored using MST analyses at two different pH levels with the following carbohydrates: linear b1-4 linked (Glc)4 backbone carrying 3 glucose (Glc) units attached to this backbone by b1-6 glycosidic bonds (Xyloglucan heptasaccharide, Megazyme O-X3G4); the linear b1-4 linked Glc pentamer (Cellopentaose, Megazyme O-CPE); Glcb1-3Glcb1-4Glcb1-3Glc tetramer (Cellotriosyl-glucose, Megazyme O-BGTETB); a mixture of Glcb1-4Glcb1-3Glcb1-4Glc and Glcb1-3Glcb1-4Glcb1-4Glc tetramers (Megazyme O-BGTETC, Cellobiosyl-cellobiose + Glucosyl-cellotriose;); as well as Gala1-4Galb1-4Glc trimer (Globotriose, IsoSep AB, 35/03). In this screening WSC3 did not bind to any of the above-mentioned sugars with a micromolar affinity apart from the control laminarin at pH 5.5 but not at pH 7.4 (Figs S5, S6). Fig. 2 SiWSC3 is transcriptionally induced during root colonization and during contact with a root pathogen and localizes to the Serendipita indica cell wall. (a) Expression of SiWSC3 quantified by quantitative reverse transcription polymerase chain reaction (qRT-PCR) during root colonization of Hordeum vulgare (red bars) and Arabidopsis thaliana (blue bars) or during contact with the root pathogenic fungus Bipolaris sorokiniana (green bar) in soil at the indicated times. The expression of SiWSC3 was calculated using the 2 ÀDDCt method relative to the expression of SiTEF. Fold changes in SiWSC3 expression during colonization of H. vulgare and A. thaliana and in confrontation with B. sorokiniana were calculated using the SiWSC3 expression levels after 5 d growth on 1/10 PNM medium, 7 d growth on ½ MS or 2 d growth in soil, respectively. Error bars indicate AE SE of the mean calculated from four biological replicates. dpi, d post-inoculation; hpi, h post-inoculation. (b) Western blot detection of SiWSC3-GFP with an anti-GFP antibody in mycelia (left) and culture filtrates (right blot) of five S. indica transformants and wild-type (WT) strain grown in complex medium (CM) or MYP medium. The band at c. 70 kDa corresponds to the SiWSC3:GFP fusion protein (highlighted in the blot; GFP, green fluorescent protein) and the band at c. 30 kDa corresponds to free GFP after cleavage of the fusion protein. (c) Subcellular localization of SiWSC3:GFP produced in S. indica transformant T 3 under the control of the SiFGB1 promoter grown in CM using confocal microscopy. A specific fluorescence signal (green) is visible at the septa and cell walls. Red shows the chitin stain WGA-AF594. (d) Confocal microscopy of the negative control transformant T 2 originating from the same transformation event as the transformant T 3 . (e) Subcellular localization of SiFGB1:GFP produced in S. indica transformant m5 under the control of the FGB1 promoter grown in CM using confocal microscopy. A strong specific fluorescence signal (green) is visible at the septal rings (white arrowheads) and in the endoplasmic reticulum (red arrowheads). Bars, 10 lm. (f, g) Relative quantification of chitin (f) and b-glucan (g) in the cell wall of S. indica by immunogold-transmission electron microscopy (TEM) and representative TEM images. Chitin was visualized by gold conjugated wheat germ agglutinin (WGA) and b1-3-glucan was visualized by a primary monoclonal mouse antibody with the help of a gold conjugated secondary anti-mouse antibody. Number of gold particles at the cell wall were counted from at least 48 TEM images for each fungal strain. Immuno-gold particles in the septa and dolipore were omitted from the analysis. CW, cell wall; asterisk, dolipore; black arrowheads, septa; blue arrowheads, show exemplarily gold particles; light grey bar, homokaryotic reference transformant; blue bar, homokaryotic SiFGB1:GFP transformant m5; dark gray bar, dikaryotic WT reference strain; green bar, dikaryotic SiWSC3:GFP transformant T 3 . Significances were calculated by Student's t-test: ***, P < 0.005. Bars, 250 nm (see Supporting Information Fig. S3 for more details). Schematic representations of FGB1 and WSC3: red, predicted signal peptide; blue, FGB1 carbohydrate-binding domain as predicted from alignment analysis (Wawra et al., 2016); green, WSC domain as predicted by SMART (http://smart.embl-heidelberg.de). Box-whisker plots: horizontal lines, median; circles, outliers; whiskers, minimum and maximum of 1 st and 4 th quartile. Taken together the affinity measurements with ITC and MST lead to the conclusion that the natural substrate for SiWSC3 might be a long-chain b1-3-glucan with a higher order structure. This also suggests that in CW preparations of barley roots longchain b1-3-glucans are not present or not accessible to SiWSC3. Indeed synthesis of b1-3-glucans have been reported in barley only during the transient production of callose (Chowdhury et al., 2014).

The multivalent SiWSC3 but not SiFGB1 agglutinates fungal cells
Hyphae interact with soil particles, roots and soil microbes forming a filamentous network that promotes foraging for soil nutrients. Thus, adhesion is an important hyphal feature during interaction with roots and other fungi. To test if the addition of SiWSC3-His or SiFGB1would have an effect on fungal cell adhesion we incubated these two lectins with the filamentous basidiomycete S. indica, the yeast basidiomycete U. maydis or the filamentous ascomycete B. sorokiniana, spanning a considerable degree of CW diversity in their compositions and molecular architectures. The chitin-binding antifungal lectin WGA served as control protein (Mirelman et al., 1975;Wawra et al., 2016). The growth phenotypes of the fungi were assessed microscopically after overnight growth for S. indica and B. sorokiniana and after 4 h of growth for U. maydis. SiWSC3-His displayed a strong agglutination effect on all tested fungi whereas SiFGB1 did not. WGA also led to the formation of fungal cell aggregates but less dense compared to those produced in the presence of SiWSC3-His (Fig. 4a). Because the filamentous growth of S. indica and B. sorokiniana complicates the quantification of the lectininduced agglutination of SiWSC3, a statistical analysis was performed by calculating the percentage of U. maydis sporidia included in aggregates relative to the total number of sporidia (Fig. 4b). Whereas SiFGB1 did not significantly increased agglutination compared to the mock treatment, SiWSC3-His and WGA increased agglutination of fungal cells remarkably. The ability of SiWSC3 to agglutinate cells is likely due to its multivalent nature compared to SiFGB1 with just one functional carbohydrate-binding domain. These results argue in favor of longchain b1-3-glucans being present in the CWs of these fungi.
In order to assess if these lectins mediate fungal adhesion to roots, S. indica chlamydospores were incubated with barley roots in a solution containing either native SiFGB1 or SiWSC3-His dissolved in water or in buffer. Addition of the lectins had no effect on spore adhesion to the roots (Fig. 4c) and the incubation with SiWSC3-His did not negatively affect S. indica growth (Fig. S7a). To explore the effect of SiWSC3 on colonization, barley roots were inoculated with S. indica spores in both the presence and absence of 10 lM SiWSC3-His and grown for 3 d

Research
New Phytologist under sterile conditions. As positive control barley roots were inoculated with S. indica spores in presence and absence of 10 lM disulfide bonded SiFGB1-His heterologously produced in E. coli. Subsequently, gDNA was extracted and the colonization rate was measured by quantification of the relative amount of fungal DNA to plant DNA by qRT-PCR. This pharmacological experiment resulted in a significant increase in fungal colonization in the presence of SiFGB1-His as shown previously for the native SiFGB1 (Wawra et al., 2016), whereas SiWSC3-His showed no effect on the S. indica colonization rate (Fig. 4d). Similarly, the use of the S. indica WSC3 overexpression transformants did not result in enhanced colonization (Fig. S7b). This suggest that WSC3 is not capable of suppressing plant immunity as observed for FGB1 and it is possibly involved in fungal CW reinforcement and cohesive adhesion between hyphal cells.
b-glucan-binding lectins as a nondestructive molecular probe for fungi in complex samples The most abundant building block of fungal CWs is glucan, which often makes up to 50-60% DW (Fesel & Zuccaro, 2016a). Although b1-3-glucose chains also can be found in the CW of plants in the form of callose, polymers containing b1-6-glycosidic bonds have only been found in the CW of fungi and members of the phylum Stramenopiles, such as in some genera of oomycetes. It is proposed that the b1-6-glycosidic bonds are responsible for connecting glucan chains with each other and thus for conferring rigidity to the CW (Bowman & Free, 2006;Latge, 2007). The multi-branched b-glucans can be firmly bound to the CW or loosely bound and accumulate around the fungus as gelatinous material. The characterization of these two novel high affinity glucan binding lectins prompted us to test their ability to specifically detect fungal-derived glycans in complex samples. FITC488 conjugates of native SiFGB1 and SiWSC3-His were generated and used as probes to detect the fungal CW in different root-associated fungi such as the endophytes S. indica and C. tofieldiae and the plant pathogen B. sorokiniana during root colonization. Cytological analysis showed that SiWSC3-FITC488 did not localize to the fungal or plant CWs. Subsequent, agglutination experiments with the conjugated SiWSC3 protein showed that the labeling (with a labeling efficiency calculated at~55%) had a significant effect on the ability of this protein to agglutinate fungal cells and thus likely on the interaction of SiWSC3 to b-glucan (Fig. 4b), explaining the absence of binding at the fungal CW in confocal microscopy. A reason could be that the covalent carboxamide bonds formed with primary amines of the protein during the reaction with 5-FAM-X [6-(Fluorescein-5-carboxamido) hexanoic acid, succinimidyl ester] belonged either to an amine group important for the interaction with the ligand and/or to a structurally important amino acid side chain (Holmes & Lantz, 2001). WSC3 labeling with CF594, as used in MST analyses, did not severely impair binding to laminarin but produced artefacts during confocal imaging in complex biological samples and was not further used in cytological analyses. By contrast, SiFGB1-FITC488 bound efficiently to the CWs of all three fungi (Fig. 5). Chitin was hardly stainable with WGA-AF594 in B. sorokiniana and C. tofieldiae, suggesting that these two fungi do not expose chitin in planta. Yet their CWs were stainable with SiFGB1-FITC488, indicating that b-glucans are a good target for molecular probes. The remarkably small size of SiFGB1 makes it more permeable compared to glucan antibodies or larger lectins such as WGA, staining more efficiently the hyphae inside living colonized host cells during fluorescence live cell imaging. In our study, the use of SiFGB1-FITC488 revealed the presence of a thick and diffuse polysaccharide matrix around the hyphae of these fungi. The matrices were visible around the hyphae outside of the host cells but also were observed frequently around intracellular hyphae (Figs 5,. During host cell penetration the fungal matrix diffuses around the penetration zone (clearly visible in Fig. S8a), suggesting that it is not tightly bound. After washing of the hyphae with water the matrix was no longer visible, showing that it is water soluble and loosely attached to the fungal surface (Fig. S8d). Due to its water solubility we could not detect this matrix regularly around fungal hyphae of S. indica with b-glucan antibody in the TEM analysis. Nevertheless, in some of the TEM sections the matrix was still visible and stained by the antibody, indicating that this matrix is made of b1-3/1-6-glucans (Fig. S9). These data show that the secretion of a fungal b-glucan matrix is a common feature of rootassociated fungi independent of their lifestyle and taxonomy.

Biochemical properties of FGB1 and WSC3
The number of biochemically characterized b-glucan binding lectins from fungi is very limited (Wawra et al., 2016). WSC domains are conserved from yeasts to mammalian cells but their sugar ligand/s are unknown. The aim of this study was to characterize the binding ability of a plant responsive WSC domain containing lectin from Serendipita indica and to compare its properties with those of the recently described b-glucan binding lectin SiFGB1 which possesses a structurally unrelated carbohydrate-binding domain. The performed isothermal titration calorimetry (ITC), circular dichroism (CD) spectroscopy and microscale thermophoresis (MST) measurements represent the first experimental proof that long chain b1-3-glucans are the preferred polysaccharide bound by proteins with WSC domain(s). SiWSC3 binds b-glucan with a K d value of 12.5 lM AE 8.8 lM and a molar ratio of protein to substrate of 1 : 3. Although the affinity of WSC3 to b-glucan is lower compared to the K d value of~100 nM for the native SiFGB1 (Wawra et al., 2016) this is still a strong binding affinity as many lectins bind in the millimolar range (Navarra et al., 2017). Our data show that the higher affinity of SiWSC3 for longer carbohydrate polymers most likely requires a complex 3D polysaccharide structure. This is in contrast to SiFGB1 which requires the presence of b1-6 side chains and can also bind to the glucose dimer gentiobiose (Wawra et al., 2016). In common with other lectins, SiWSC3 seems to have a preformed carbohydrate-binding site, which can accommodate the b-glucan ligand without undergoing a strong conformational change in its structure. This feature seems to minimize the energetic penalty paid upon binding to carbohydrate ligands  (Kanagawa et al., 2011). Remarkably, binding of the ligand to SiFGB1 leads to dramatic secondary structural changes which are thought to be necessary for the immunosuppressive function of this protein in planta (Wawra et al., 2016). Mutational analyses of FGB1 will help proving this hypothesis in future studies.

FGB1 and WSC3 functions
Several lectins from phytopathogenic fungi have been characterized including beta-trefoil lectins from R. solani and S. sclerotiorum, actinoporin-like lectins from S. rolfsii and SiFGB1 had no effect on fungal cell aggregation, whereas WSC3-His and WGA-AF594 treatment significantly increased the number of aggregated cells. Letters indicate independent groups according to an unpaired Student's t-test (P < 0.05; n > 100). (c) Adhesion of S. indica spores to barley roots were calculated by counting the number of attached spores to the root surface from images acquired by confocal laser scanning microscopy. Serendipita indica spore solution was mixed either with an equal volume of water or Tris buffer pH 8 (mock, red bars), 10 lM nFGB1 (blue bars) or 10 lM SiWSC3-His (green bars) diluted in water or in Tris buffer pH 8. Error bars represent AE SE of the mean of three biological replicates. No significant differences between the treatments were observed using an unpaired Student's t-test (P < 0.05, indicated by the letter 'a'). (d) Fungal colonization of barley roots. Serendipita indica spore solution was mixed either with water (mock, red bar), 10 lM recombinant His tagged FGB1 diluted in water (blue bar) or 10 lM SiWSC3-His diluted in water (green bar) and the colonization rate was quantified by quantitative reverse transcription polymerase chain reaction (qRT-PCR) measuring the relative amount of fungal gDNA (SiTEF) compared to plant gDNA (HvUBI) at 3 d post-inoculation (dpi). The colonization rate was normalized to the respective mock-treatment which was set to 1. Error bars represent AE SE of the mean of at least six biological replicates. No significant difference between the mock and SiWSC3-His treatment was observed whereas recombinant FGB1-His significantly increased colonization. Letters indicate independent groups according to an unpaired Student's t-test (P < 0.05).

New Phytologist
X. chrysenteron and cyanovirin-like lectin from G. zeae. Yet no relationship of these proteins to the infection process could be established (Birck et al., 2004;Leonidas et al., 2007;Koharudin et al., 2011;Matei et al., 2011;Skamnaki et al., 2013;Varrot et al., 2013). So far only lectins with LysM domains and SiFGB1 could be linked to plant colonization. A dual role was suggested for the chitin-binding LysM effectors ChELP1 and ChELP2 of the hemibiotrophic fungus Colletotrichum higginsianum and for SiFGB1 thus multiple functions might represent a common strategy of some fungal lectins (Takahara et al., 2016). In our study the deregulation of SiFGB1 and SiWSC3 led to an alteration of fungal cell wall (CW) properties that is most clearly noticeable by the alteration of the response to CW stressors and by the binding of the b1-3-glucan antibody to the fungal CW (Figs 2g, S4c and www.newphytologist.com Wawra et al., 2016). This indicates that both lectins are capable of interacting with the fungal CW. It was shown that during S. indica growth a large proportion of SiFGB1 is secreted into the culture supernatant (Wawra et al., 2016) but no obvious accumulation of free SiWSC3 could be detected in this study. The absence of SiWSC3 in S. indica culture supernatant and the ability to agglutinate fungal cells hints to a potential role in strengthening the S. indica CW against external stresses and in adhesion between hyphal cells. The absence of an effect on colonization further supports the idea that this lectin does not have host immunosuppressive functions. Thus, we propose that SiWSC3 acts as a proteinaceous glue that connects neighboring b-glucan fibrils noncovalently in the CW of S. indica. Similar to the fungal Agglutinin-Like Sequence (Als) family, which includes sexual agglutinins, virulence factors and flocculins (Hoyer & Cota, 2016), the WSC lectin-like family could mediate cell-cell and cell-environment interactions.

Detection of b-glucans by lectins
Due to its remarkable strong binding affinity, specificity and small size, SiFGB1 represents a valuable tool to study CW development and composition in fungi and possibly also in oomycetes. The potential of SiFGB1 as a protein probe for bglucans is exemplified by the labeling of the extracellular polysaccharide matrix with b1-6-linked glucoses surrounding fungal hyphae during root colonization. Such an extracellular matrix is known from bacteria and some fungal animal pathogens as polysaccharide cement crucial for the formation of biofilms and protection from the host enzymatic activity and recognition (Flemming & Wingender, 2010;Priegnitz et al., 2012). The thick and diffuse appearance of the matrix in this study also is in line with results from a recent work where Kang and coworkers analyzed the CW architecture of the human pathogen Aspergillus fumigatus by solid-state nuclear magnetic resonance. There the authors found well-hydrated and relatively mobile matrix formed by b1-3, b1-4 and b1-6 linked glucans (Kang et al., 2018).

Conclusions
Here we demonstrated that SiFGB1 requires b1-6-glucan linkages for efficient binding, whereas SiWSC3 binds indistinctly b1-6 branched and debranched long-chain b1-3-glucans. Direct comparison with the chemically labeled chitin-binding lectin wheat germ agglutinin (WGA) conjugated with the fluorescence dye Alexa Fluor 594 shows that SiFGB1 can be used to detect fungi not stainable with WGA-AF594. The existence of the two lectins SiWSC3 and SiFGB1 in S. indica, which both exhibit affinity to b-glucans but fulfill distinct biological tasks, illustrates the importance of b-glucan as an essential component of the fungal CW that needs to be fostered to prevent recognition while maintaining their integrity. Thus, the expansion for lectin WSC proteins in S. indica could enable this endophyte to cope with extremely challenging environments for the CW such as those found in planta and during confrontation with other fungi.

Supporting Information
Additional Supporting Information may be found online in the Supporting Information section at the end of the article.         Methods S1 Supplemental materials and methods and primer list.