The β‐1,3‐glucanosyltransferases (Gels) affect the structure of the rice blast fungal cell wall during appressorium‐mediated plant infection

Abstract The fungal wall is pivotal for cell shape and function, and in interfacial protection during host infection and environmental challenge. Here, we provide the first description of the carbohydrate composition and structure of the cell wall of the rice blast fungus Magnaporthe oryzae. We focus on the family of glucan elongation proteins (Gels) and characterize five putative β‐1,3‐glucan glucanosyltransferases that each carry the Glycoside Hydrolase 72 signature. We generated targeted deletion mutants of all Gel isoforms, that is, the GH72+, which carry a putative carbohydrate‐binding module, and the GH72− Gels, without this motif. We reveal that M. oryzae GH72 + GELs are expressed in spores and during both infective and vegetative growth, but each individual Gel enzymes are dispensable for pathogenicity. Further, we demonstrated that a Δgel1Δgel3Δgel4 null mutant has a modified cell wall in which 1,3‐glucans have a higher degree of polymerization and are less branched than the wild‐type strain. The mutant showed significant differences in global patterns of gene expression, a hyper‐branching phenotype and no sporulation, and thus was unable to cause rice blast lesions (except via wounded tissues). We conclude that Gel proteins play significant roles in structural modification of the fungal cell wall during appressorium‐mediated plant infection.

. Evidence for this comes from S. cerevisiae Δgas1, which shows a decrease in β-1,3-glucan content in the mutant wall, compared with the wild-type strain, coupled with a rise in β-1,3-glucan in the growth medium (Ram et al. 1998). Such data implies that Gas proteins are involved in the incorporation of β-1,3-glucan into the wall, but that they are not involved in glucan synthesis (Ram et al. 1998). An analysis of products resulting from in vitro incubation of recombinant Gas proteins with a reduced laminarioligosaccharide suggests a two-step transglycosylating mechanism for these enzymes. Here, Gas proteins cleave a β-1,3 glycosidic linkage in the glucan chain and subsequently reform a β-1-3 linkage between the reducing end of one released chain and the nonreducing end of side branches in existent β-glucans (Hurtado-Guerrero et al. 2009). Thus, the transglycosylating activity of Gas proteins leads to the integration of nascent β-1,3-glucan chains into the existing ß-glucan network. However, a role for Gas proteins in incorporating β-1,3-glucan into the wall has not been demonstrated in vivo. Thus far, the phenotype of GAS deletion mutants has been taken as proxy evidence in support of this model, being, specifically, loss of β-glucan to the medium, reduction in alkali-insoluble wall glucan, and induction of the cell wall integrity (CWI) pathway (Ram et al. 1998;Fonzi 1999;Carotti et al. 2004;Mouyna et al. 2005;Gastebois, Fontaine, Latge, & Mouyna 2010b).
The filamentous fungus Magnaporthe oryzae is the causal agent of rice blast disease (Couch & Kohn 2002). Under blast-favorable conditions, up to 30% of the annual rice crop can be lost to infection; controlling disease would constitute a major contribution to ensuring global food security (Talbot 2003). Disease is initiated when a threecelled conidium detaches from conidiophore-laden host lesions and attaches to the plant surface, by release of apical spore tip mucilage (Hamer, Howard, Chumley, & Valent 1988). Germination leads to formation of a short germ tube, which matures at its tip into an appressorium. This infection structure forms in response to host cues, such as the hard, hydrophobic leaf surface and plant cutin, as well as absence of nutrients (Skamnioti & Gurr 2007;Wilson & Talbot 2009).
Autophagy then occurs in the conidium whose content is recycled into the appressorium (Veneault-Fourrey, Barooah, Egan, Wakley, & Talbot 2006), which is lined with melanin on the inner edge of the fungal wall.
Turgor pressure rises within this newly sealed chamber (De Jong, McCormack, Smirnoff, & Talbot 1997), leading to the emergence of a narrow penetration peg, which pushes through the cuticle and cell wall, expands to form a primary hypha, and then differentiates into bulbous invasive hyphae. The fungus spreads rapidly through a susceptible host (Kankanala, Czymmek, & Valent 2007;Khang et al. 2010), culminating in lesions on aerial tissues, which discharge prolific numbers of conidia, thereby promoting epidemic disease spread (Skamnioti & Gurr 2009).
The fungus is capable of causing disease on approximately fifty grass and sedge species. Blast disease is thus of concern with regard to its changing demographics and ability to move to new hosts (Yoshida et al. 2016), with its movement fuelled by global climate change (Bebber, Ramotowski, & Gurr 2013).
Our understanding of the mechanisms which underpin pathogenesis remain far from complete, and thus has not yet fuelled the hunt for target-specific antifungals (Skamnioti & Gurr 2009). Attractive amongst prospective targets is the fungal cell wall. However, little is known about the organization of the M. oryzae wall or about wall variation between cell types during plant infection. Previously, research has considered the architecture of the spore surface, revealing a multi-layered rodlet surface structure, composed of the hydrophobin Mpg1, which is important in appressorium attachment and morphogenesis (Talbot, Ebbole, & Hamer 1993;Talbot et al. 1996;Kershaw, Thornton, Wakley, & Talbot 2005). Electron micrographs by Howard and Valent (1996) and Mares et al. (2006) also showed, respectively, the layered structures of the conidium and hyphal cell, purportedly comprising β-1,3-glucans and chitin.
At present, the polysaccharide composition of the M. oryzae wall remains unknown. Recently, however, Fujikawa et al. (2009Fujikawa et al. ( , 2012 revealed that it carries α-1,3-glucan moieties and that these surfacelying polymers play a role in camouflaging the fungus from recognition by the host immune system during formation of infectious hyphae. In this report, we provide the first detailed profile of the M. oryzae wall carbohydrate composition and structure. We consider the roles of the Gel family of β-1,3-glucanosyltransferases in infective and vegeta-  (Sillo et al. 2013), and a putative GPI anchor ( Figure 1a).
To unmask likely evolutionary relationships of M. oryzae GEL genes, we used maximum likelihood (ML) analysis (Sillo et al. 2013) to compare 237 proteins belonging to 24 Pezizomycotina (e.g., M.
oryzae, A. fumigatus), 25 Saccharomycotina (e.g., S. cerevisiae, C. albicans), and 2 Schizosaccharomyces (S. pombe, S. japonicas). Three Basidiomycota sequences were used as outgroup taxa (Figure 1b). We assayed the effect of various cell wall perturbation chemistries (CR, CFW), applied cell wall and plasma membrane stresses (SDS, alkaline pH, sorbitol, and glycerol), and oxidative stress (hydrogen peroxide). Surprisingly, we observed growth reduction of Δgel4 and Δgel3Δgel4 mutants on minimal medium (MM; by approximately 25%), and in CM supplemented with CR (30%) or SDS (25% for Δgel3Δgel4; Figure S2). Interestingly, the emergent germ tubes of  Sillo et al. (2013) respective protein products localise to the cell periphery of the threecelled spores and emergent germ tubes up to 4 hours post-inoculation (hpi; Figure 2a and b). Appressoria were, however, not labeled by the fusions, indeed, by 8 hpi GEL4 expression is reduced and then (c) Guy11 transformed with both GEL3:mCherry and GEL4:eGFP fusions at 0 hpi shown in split red and green, as well as merged channels. The arrow points to a spore that is expressing the GEL4:eGFP fusion only, therefore appearing invisible in the red channel. The arrow head points to differential circumferential localization of GEL3:mCherry, while GEL4: eGFP persists along the edges of the spore cell-cell boundaries. Projections of Z-stacks following expression of (d) Δgel3/GEL3:mCherry and (e) Δgel4/GEL4:eGFP during development of penetration pegs (arrows) and infection hyphae on onion peels, at 24 hpi and rice, at 24 and 48 hpi. GEL4 is not visible at these stages; the transmitted-light micrograph insert shows that melanized appressoria with invasion hyphae are present. GEL4 is strongly expressed in vegetative mycelia of 10-day-old cultures; GEL3 is not. The confocal images were collected for both red and green channels to indicate the autofluorescence for the opposite fluorophore. The scale bars are 5 (a, b, c) or 10 (d, e) μm disappears completely. When Δgel3/GEL3:mCherry and Δgel4/GEL4: eGFP fusions were expressed simultaneously in Guy11, some differential labeling was observed; in extreme cases, only GEL4 was visible but not GEL3 (Figure 2c, arrow). GEL3 was more highly expressed and could be tracked during germling development on onion epidermis. This "surface" supports development of penetration pegs and invasive hyphae (Chida & Sisler 1987) ( GEL4 is not expressed during plant infection but it is expressed in vegetative hyphae ( Figure 2). Thus, GEL3 and GEL4 are expressed in conidia, but show differential localization during vegetative and invasive hyphal growth, with GEL3 most strongly associated with host invasion.
2.5 | GH72 + Gels are not essential for spore and appressorium development and infection As GEL3 and GEL4 are both expressed during conidial development and GEL3 is expressed during infection, we investigated the role of GH72 + in pathogenicity. We followed germling and appressorium development on hydrophobic glass slides and compared the number of melanized appressoria at 8 hpi between the strains. There was no significant difference between the Guy11, single Δgel3 and Δgel4, and double Δgel3Δgel4 mutants, or the complemented strains ( Figure S2e). Furthermore, we observed no difference in the development of penetration pegs and invasion hyphae on onion epidermis at 24 hpi ( Figure S2f). Indeed, the mutants were fully pathogenic on barley ( Figure S2g and h).

| Monosaccharide composition of M. oryzae cell wall polysaccharides
There has been no detailed analysis of the monosaccharide composition and specific glycosidic linkages of the WT strain Guy11 wall hitherto. We therefore investigated wall monosaccharide composition in Guy11 and compared it with Δgel3Δgel4, grown in CM. Total wall polysaccharides were extracted, fully hydrolyzed to their constituent monosaccharides and analyzed by GC/EI-MS. Table 1 shows only minor differences in total mannose, galactose, glucose and N-acetylglucosamine content between three independent double Δgel3Δgel4 mutants and Guy11. We also observed that when Guy11 is grown in MM, the wall mannose content was reduced significantly, but was compensated by a significant increase in glucose. Growth conditions thus affect cell wall composition (Aguilar-  Figure S3a). The most upregulated gene was GEL2, which showed a threefold upregulation compared to nongerminated spores at 0 hpi, at 24 hpi, coincident with the time of invasive hypha development. qRT-PCR results also confirmed that GEL4 (and GEL2) are slightly upregulated in mycelium compared to spores while GEL3 (and GEL1) are downregulated, as seen by confocal microscopy.
GEL5 is weakly expressed in spores but strongly upregulated in mycelium ( Figure S3b and S3c). Pathogenicity assays of single, double, and triple mutants confirmed that all strains, with the exception of Δgel1Δgel3Δgel4 (which does not sporulate), produce melanized appressoria (Figure 4a), penetration pegs, and invasive hyphae and are all as pathogenic as Guy11  Figure 4b). Thus, GH72 + and GH72 − members are both dispensable for pathogenicity, but specific isoforms are essential for spore formation and host infection (see below).

| A Gel
2.9 | Δgel1Δgel3Δgel4 has a hyperbranching phenotype and does not produce conidia  mycelium also revealed a hyperbranching phenotype (Figure 5a, CFW). In addition, there are differences in general staining intensity, perhaps due to the less branched glucans allowing greater accessibility to CFW, and greater intensity at growing tips, where the newly synthesized glucans are unlikely to have branched or be highly cross-linked.
The mutant mycelial cells are short, often round, and branch frequently ( Figure 5b). Furthermore, when grown across a glass cover slip for 6 days, Δgel1Δgel3Δgel4 formed terminal rounded tip ends, which then continued to grow and form hyphae (Figure 5c, CR).
Sensitivity to exposure to the fungal wall-degrading enzyme Glucanex was used to compare the rates of release of protoplasts by Δgel1Δgel3Δgel4 with Guy11 from mycelial tissues (Δgel1Δgel3Δgel4 does not sporulate). This revealed that Δgel1Δgel3Δgel4 releases fewer protoplasts and at a slower rate than the Guy11 strain-approximately 5-10-fold fewer protoplasts than Guy11, some 180 minutes postexposure to wall-degrading enzymes (Figure 5d). This data suggests that the altered mutant wall is more resistant to Glucanex degradation than WT-a result that attests to the unknown enzyme specificity of these members of the Gel family. Δgel1Δgel3Δgel4 protoplasts were We compared the mycelial walls of Δgel1Δgel3Δgel4 and Guy11 by TEM (Figure 5e). This revealed no gross differences in wall thickness between the strains, with Δgel1Δgel3Δgel4 walls being 81.1 ± 40.6 nm thick and Guy11 walls at 73.8 ± 35.2 nm (P = 0.342, n = 50). We compared cryo-SEM images of Δgel1Δgel3Δgel4 and Guy11 mycelium near its growing edge, showing again the mutant's densely branching phenotype ( Figure S4). Finally, we collected SEM images of Δgel1Δgel3Δgel4 mutant and Guy11, revealing that the mutant surface appears stippled, whilst Guy11 is smooth but with ECM extruded from the wall-a feature absent from the triple mutant ( Figure 5f).

| Monosaccharide composition and linkage analysis of M. oryzae cell wall polysaccharides in the triple Δgel1Δgel3Δgel4 mutant
We determined the monosaccharide composition of alkali soluble and insoluble fractions (Table 2), and specific glycosidic linkages in the Δgel1Δgel3Δgel4 wall. Consistent with the double mutant Δgel3Δgel4, the triple mutant showed a greater abundance of linear 1,3-glucans (approximately 18% higher than WT). Indeed, with a decreased proportion in terminal-and 1,3,6-glycosidic linkages, the glucans are characterized by a higher degree of polymerization and a lower number of 1,6-branching points ( Figure 6). In essence, 1,3-Glcp in Δgel3Δgel4 (P = 0.042, n = 3) and Δgel1Δgel3Δgel4 (P = 0.002, n = 4), and t-Glcp in Δgel1Δgel3Δgel4 (P = 0.025, n = 4); all such values (of double and triple mutant variants) are thus statistically significant from Guy11.

| Transcriptional analysis of the triple Δgel1Δgel3Δgel4 mutant strains and Guy11
The triple mutant strain Δgel1Δgel3Δgel4 shows a nonsporulating, hyper-branching phenotype. We asked whether this altered morphology correlated with specific changes in genes expression between the mutant and wild-type strains-we thus investigated which genes were differentially expressed as compared with Guy11. We identified global patterns of gene expression in two independent Δgel1Δgel3Δgel4 mutant strains, compared with Guy11, by RNA-Seq analysis. Three independent replicates were analyzed from each strain. Figure S5a shows the overall Euclidean distance (distance between two points in space as showing a measure of the differences between the wild type and mutant strains) between all samples. Individual replicates from each sample cluster together and expression data from the Linkage analysis of purified cell wall polysaccharides from Guy11, Δgel3Δgel4, and Δgel1Δgel3Δgel4 mutant strains (GC/EI-MS). Liquid complete medium was inoculated with spores or hyphal residues (as the triple mutant does not sporulate) and shaken at 150 rpm at 24°C for 4 or 7 days. Cell wall polysaccharides were purified and analyzed as described in Section 4. The percentage of monosaccharide derivatives identified from each of the three strains was determined from four technical replicates derived from each of the three independently grown biological replicates two individual mutants are far closer to each other than to Guy11.
Based on p-values (adjusted for multiple testing, using Benjamini-Hochberg method) <0.01 and at least two-fold difference in expression, the two mutants share 310 genes upregulated and 235 genes downregulated, compared to Guy11 (Table S6 and S7).
GO terms that are more highly represented in genes that showed differential upregulation in Δgel1Δgel3Δgel4 (as compared with the whole genome) are shown in Figure S5b. Of these, the most interesting are the glycoside hydrolases (GH) (GO:0016798). Nineteen GH encoding genes are upregulated, of which, 14 are predicted to be secreted (Supp Table S8). Fungal cell wall remodeling enzymes include the glucan 1,3-beta-glucosidase and chitinase, as well as the wall-building chitin synthase and polysaccharide-degrading enzymes, predicted to be extracellular, such as alpha amylase, xylanase, alpha-galactosidase, and beta-fructofuranosidase. Interestingly, GEL2 is upregulated strongly in the mutant, possibly to compensate for the absence GEL1,  Table S9.
GO terms that are more highly represented in downregulated genes found in Δgel1Δgel3Δgel4 are summarized in Figure S5c and listed in Table S7. Single GEL gene deletions of all family members did not reveal any phenotypic differences from Guy11, apart from reduced growth of  (Kamei et al. 2013). There is, however, no obvious difference in cell wall thickness, as evidenced by TEM. The mutant strain wall appeared rougher than the WT wall, and ECM was absent from Δgel1Δgel3Δgel4.
Cell wall analysis revealed only minor differences in the glucose between the Δgel1Δgel3Δgel4 mutant and WT, but galactose is significantly increased, while mannose reduced. Perhaps the most surprising is the 10-fold increase in arabinose and xylose in the triple mutant.
Monosaccharide linkages (mol%) were obtained from four technical replicates of each of three biological replicates.

| Transmission electron microscopy
Mycelial squares (app 5 × 5 mm) were cut from the growing edge of 10-day CM plates, fixed and viewed as described in Samalova et al. (2014).  ), using the TopHat2 splice site-aware aligner (Kim et al. 2013).

| Scanning electron microscopy
Counts of reads mapping to each gene in the genome were generated using the HTSeq-count function of the HTSeq package (Anders, Pyl, & Huber 2015). Relative gene expression was quantified and differentially expressed genes identified using DESeq (Anders & Huber 2010). Gene ontology (GO) annotation of the M. oryzae genome and analysis of GO categories were performed using BLAST2GO (Conesa & Götz 2008).