An Epichloë festucae homologue of MOB3, a component of the STRIPAK complex, is required for the establishment of a mutualistic symbiotic interaction with Lolium perenne

Summary In both Sordaria macrospora and Neurospora crassa, components of the conserved STRIPAK (striatin‐interacting phosphatase and kinase) complex regulate cell–cell fusion, hyphal network development and fruiting body formation. Interestingly, a number of Epichloë festucae genes that are required for hyphal cell–cell fusion, such as noxA, noxR, proA, mpkA and mkkA, are also required for the establishment of a mutualistic symbiotic interaction with Lolium perenne. To determine whether MobC, a homologue of the STRIPAK complex component MOB3 in S. macrospora and N. crassa, is required for E. festucae hyphal fusion and symbiosis, a mobC deletion strain was generated. The ΔmobC mutant showed reduced rates of hyphal cell–cell fusion, formed intrahyphal hyphae and exhibited enhanced conidiation. Plants infected with ΔmobC were severely stunted. Hyphae of ΔmobC showed a proliferative pattern of growth within the leaves of Lolium perenne with increased colonization of the intercellular spaces and vascular bundles. Although hyphae were still able to form expressoria, structures allowing the colonization of the leaf surface, the frequency of formation was significantly reduced. Collectively, these results show that the STRIPAK component MobC is required for the establishment of a mutualistic symbiotic association between E. festucae and L. perenne, and plays an accessory role in the regulation of hyphal cell–cell fusion and expressorium development in E. festucae.


INTRODUCTION
Filamentous fungi exhibit complex life cycles. Numerous temporal and spatial cellular changes are required for progressive spore germination, septation, cell-cell fusion and the development of sexual and pathogenic structures, such as fruiting bodies and appressoria. The characterization of several protoperithecia (pro) mutants in Sordaria macrospora, and pro homologues in other fungi, has revealed that components of the fungal striatininteracting phosphatase and kinase (STRIPAK) complex are required for multiple developmental processes and appear to coordinate cross-talk between different signalling pathways, indicating that they have evolved diverse regulatory functions (K€ uck et al., 2016).
In Sordaria, GPI1 localizes to the plasma membrane, PRO45 to the nuclear membrane, PRO45/GPI1 to mitochondria, and KIN3 and KIN24 to septal pores, whereas, in N. crassa, HAM-2 and HAM-3 localize to the nuclear envelope (Bloemendal et al., 2010;Dettmann et al., 2013;Frey et al., 2015a,b, Nordzieke et al., 2015. A correct localization pattern of HAM-2 and HAM-3 is required for MAK-1 [cell wall integrity (CWI) pathway mitogenactivated protein (MAP) kinase] nuclear accumulation in a MAK-2 [pheromone response (PR) pathway MAP kinase]-dependent manner, as MAK-2 is required for phosphorylation of the STRIPAK component MOB3 (Dettmann et al., 2013). These results demonstrate that the STRIPAK complex components exhibit multiple is i s a n o p en a c c e s s a rt ic le u nd er th e te r ms o f t he C re a t iv e Co m m o ns A t tr ib ut i on L ic e ns e , w hi c h pe r mi t s u s e, di st r ib ut io n a n d r ep r od uc ti on i n a n y m ed i um , p r ov id e d t he o ri g in a l wo r k i s pr o pe rl y c i t ed .
temporal and spatial localization patterns, and play integral roles in transmitting signals from both the CWI and PR MAP kinase pathways, which, in turn, regulate multiple downstream developmental signalling pathways. Epichlo€ e festucae forms mutualistic symbiotic relationships with species of cool-season grasses, such as Festuca and Lolium, by both systemic colonization of the intercellular spaces within the host leaves and by colonization of the leaf surface via epiphyllous growth (Christensen et al., 2008;Clay and Schardl, 2002;Schardl, 1996). Mutations in genes encoding components of the Nox complex (noxA or noxR), CWI MAP kinases and scaffold (mpkA, mkkA and so), and the transcription factor proA, abolish cell-cell fusion in culture and trigger proliferative pathogenic-like hyphal growth in planta (Becker et al., 2015;Charlton et al., 2012;Takemoto et al., 2006Takemoto et al., , 2011Tanaka et al., 2006Tanaka et al., , 2008Tanaka et al., , 2013. The co-association of these phenotypes leads to the hypothesis that cell-cell fusion is required for the maintenance of a mutualistic symbiotic interaction of this fungal endophyte with the host grass. Given that the STRIPAK complex is important for cell-cell fusion in N. crassa and S. macrospora (Bloemendal et al., 2012;Dettmann et al., 2013), but has not been investigated in E. festucae, we set out to first identify whether the STRIPAK complex members were conserved in E. festucae and to test whether the homologue of the STRIPAK complex component MOB3 has a role in E. festucae hyphal cell-cell fusion and maintenance of a mutualistic symbiotic interaction with the plant host Lolium perenne.

Epichlo€ e festucae contains homologues of the STRIPAK complex
To determine whether E. festucae contains components of the STRIPAK complex, a TBLASTN search of the genome sequence was carried out using S. macrospora PRO11, PRO22, PRO45 and MOB3 as the query sequences. The E. festucae homologues identified were aligned with the S. macrospora sequences, as well as the corresponding polypeptide sequences from Fusarium graminearum, Magnaporthe oryzae, N. crassa and Podospora anserina (Figs S1-S4, see Supporting Information). The E. festucae homologues share 70% and 69% identity to S. macrospora and N. crassa PRO11 and HAM-3, 61% and 62% identity to PRO22 and HAM-2, 57% and 51% identity to PRO45 and HAM-4, and 48% and 51% identity to MOB3 and MOB-3, respectively. In addition, they contain all of the conserved domains found in these proteins, suggesting that E. festucae contains a functional STRIPAK complex. In particular, MobC contains a MOB domain, serine and threonine phosphorylation domains, a Cys2-His2 Zn 21 -binding domain and an SH3-binding domain, suggesting that MobC participates in protein phosphorylation and protein-protein interactions.
In S. macrospora, re-introduction of the N-terminal MOB3 domain complements the mob3 mutant defects, whereas introduction of the C-terminal domain does not (Bernhards and P€ oggeler, 2011). These findings may explain the variability in the sequence of the C-terminus observed between MOB3 homologues.
To investigate the role of the STRIPAK complex in the regulation of E. festucae hyphal cell-cell fusion and the interaction of this endophyte with its host L. perenne, we deleted the homologue of mob3, which we have named mobC, by transforming protoplasts of E. festucae with a restriction enzyme-generated linear fragment of pKG4 (Fig. S5a, see Supporting Information). Polymerase chain reaction (PCR) screening of 50 geneticin-resistant (Gen R ) transformants identified six (DmobC#21, DmobC#28, DmobC#31, DmobC#37, DmobC#38 and DmobC#40) that produced banding patterns consistent with targeted replacement events (Fig. S5b). Southern analysis of genomic DNA digests from these transformants confirmed that just two (DmobC#21 and DmobC#37) were single-copy 'clean' replacements at the mobC locus (Fig. S5c). Based on these results, these two strains were selected for further experiments.

Culture phenotype of DmobC
The culture morphology and radial growth of DmobC mutants on potato dextrose (PD) were indistinguishable from those of the wild-type (WT) (Fig. 1a); however, light microscopy revealed that the DmobC mutants produced intrahyphal hyphae (IHH) (Fig. 1b), a phenotype that has never been observed in WT cultures. In addition, DmobC mutants showed significantly more conidia than WT (Fig. 2a,c), a phenotype complemented by the re-introduction of the WT gene (Fig. 2c). Surprisingly, whereas mob3 deletion mutants in S. macrospora and N. crassa are completely defective in cell-cell fusion (Bernhards and P€ oggeler, 2011;K€ uck et al., 2016;Maerz et al., 2009), the E. festucae DmobC strains are still capable of undergoing hyphal cell-cell fusion, albeit at a four-to five-fold lower frequency than that of WT (Fig. 2b,d). Introduction of a WT copy of mobC into DmobC restored the WT fusion phenotype, confirming complementation of this mutant (Figs 2d and S5). These results suggest that MobC plays an accessory role in negatively regulating conidiation and positively regulating hyphal fusion in E. festucae.
To test whether deletion of mobC affected basal level phosphorylation of the MpkA (CWI) and MpkB (PR) MAP kinases, western blotting was carried out (Fig. S6, see Supporting Information). Two bands were detected in WT, corresponding to phosphorylated MpkA at 47 kDa and phosphorylated MpkB, the Fus3 homologue, at 41 kDa. Although MpkA phosphorylation was not detected in the DmpkA and DmkkA CWI kinase mutants (Becker et al., 2015), both MpkA and MpkB phosphorylation still occurred in the DmobC#21 and DmobC#37 mutants, with bands of similar intensity to that of WT.

Symbiotic interaction phenotype of DmobC
Given that IHH formation and hyperconidiation are phenotypes shared among many E. festucae symbiosis mutants (Becker et al., 2015;Tanaka et al., 2013), we next tested whether mobC is required for the symbiotic interaction of E. festucae with L. perenne. Plants infected with DmobC were severely stunted compared with WT-infected plants (Fig. 3a). Although there were no significant differences in the number of tillers per plant between WT-and DmobC-infected plants, both tiller length and root length were significantly reduced (Fig. S7, see Supporting Information). Introduction of a WT copy of mobC into the DmobC background rescued the WT symbiotic interaction phenotype (Figs 3a and S7). To evaluate the cellular phenotype of plants infected with DmobC, we harvested pseudostem tissue samples and examined a range of phenotype parameters by transmission electron microscopy (TEM) (Fig. 4) and confocal laser scanning microscopy (CLSM) (Fig. 5). We observed extensive hyphal colonization in DmobCinfected plants, with up to six hyphae per intercellular space in the mesophyll tissue. In contrast, WT associations contained mostly one to two hyphae per intercellular space (Figs 3b and 4a). In addition, the DmobC hyphae were frequently vacuolated (Fig. 4c); a subset of hyphae contained IHH (Fig. 4d), with the outer cell walls appearing less electron dense (Fig. 4e). Hyphae of DmobC were also very abundant in the vascular bundle tissue (c) Quantification of single colonies recovered from 300 lL of wild-type (WT), DmobC and DmobC/mobC complemented conidia suspensions recovered from a total of 15 cultures grown on potato dextrose agar plates for 7 days. Bars represent mean 6 standard error (n 5 3). Asterisks indicate significant differences from WT as determined by Welch's t-test. (d) Average cell-cell fusions observed per 340 objective lens magnification field. Bars represent mean 6 standard error (n 5 10). An asterisk indicates significant differences from WT as determined by Welch's t-test.
( Fig. 4b), which is seldom, if ever, colonized by WT hyphae. Hyphae in these tissues were electron dense, presumably reflecting the abundant supply of nutrients in these tissues for growth.
The prolific growth of the DmobC mutant in leaf tissue compared with the more restrictive growth of WT was also evident from CLSM analysis of pseudostem tissue stained with aniline blue (orange/red pseudocolour) and WGA-AF488 (wheat germ agglutinin coupled to AlexaFluor488, blue pseudocolour), which stain b-glucan and chitin, respectively (Fig. 5a). Not only were the DmobC hyphae more abundant than WT hyphae, but they also formed aberrant convoluted hyphal structures comprising many cells, as evident from the many fluorescent septa (Fig. 5b), and exhibited patchy chitin staining (Fig. 5c). Despite these dramatic changes in growth within the plant, DmobC hyphae were still capable of forming cell-cell fusions in planta (Fig. 5d).
mobC plays an accessory role in the regulation of E. festucae expressorium formation Endophytic E. festucae hyphae exit the host cuticle layer via an expressorium to form an epiphyllous hyphal network on the surface of the host plant. The formation of these structures requires functional NADPH oxidase NoxA and NoxB complexes (Becker et al., 2016). The inability of nox mutants to develop expressoria results in the formation of extensive subcuticular growth of hyphae, which eventually breach the surface of the leaves. Given the role of NoxA and NoxB in the development of expressoria in E. festucae-L. perenne associations, leaf tissues infected with the DmobC mutant were examined by CLSM for the formation of these structures (Fig. 6). Although a few expressoria were identified (Fig. 6b,e), the more common phenotype was extensive subcuticular growth of DmobC (Fig. 6c,f), suggesting that these strains were not fully competent to develop expressoria (Fig. 6). The frequency of expressoria formation in leaf tissue infected with DmobC was significantly lower than in WT (Fig. S8a, see Supporting Information). Instead, mutant associations had a much greater frequency of subcuticular hyphae (Fig. S8b), which were found to eventually rupture the cuticle of the leaf (Fig. S8c). These results suggest that MobC has an accessory role in regulating the differentiation of E. festucae expressoria.
An additional observation was the occasional emergence of hyphae through stomata (Fig. 6i). However, once on the cell surface, the cell walls of mutant hyphae, like WT hyphae, were remodelled with both cell wall and septa binding WGA-AF488 (captured in blue pseudocolour), whereas only the septa of endophytic hyphae bind WGA-AF488 (Becker et al., 2016).

DISCUSSION
Epichlo€ e festucae NoxA, NoxR, So, ProA, MpkA and MkkA are required for cell-cell fusion in culture and hyphal network formation in planta. These findings gave rise to the hypothesis that cell-cell fusion regulates and restricts hyphal growth in planta and is required for the establishment of a mutualistic association (Becker et al., 2015;Charlton et al., 2012;Takemoto et al., 2006Takemoto et al., , 2011Tanaka et al., 2006Tanaka et al., , 2008Tanaka et al., , 2013. Here, we show that E. festucae MobC, the homologue of the S. macrospora STRIPAK complex component MOB3, is essential for the maintenance of a mutualistic symbiotic interaction between E. festucae and L. perenne, and plays an accessory role in regulating cell-cell fusion and expressoria formation. In culture, E. festucae hyphal strands adhere to one another to form cables, which extend outwards from the colony centre. Cell-cell fusion within these cables occurs by tip-to-side fusion   events that are easy to observe and quantify using light microscopy (Becker et al., 2015). To date, we have been unable to test whether cell-cell fusion mutants of E. festucae are also defective in fruiting body development, as the sexual cycle of E.
festucae is highly complex and only occurs on the plant host and requires a third symbiont, a Botanophila fly, to transfer spermatia from a stroma of one mating type to a stroma of the opposite mating type (Bultman & Leuchtmann, 2009). In contrast, S. macrospora and N. crassa readily form fruiting bodies in culture, enabling phenotype analysis of cell communication mutants in both the vegetative and reproductive stages of development Fleissner et al., 2009;K€ uck et al., 2009;Roca et al., 2005;Teichert et al., 2014). In E. festucae, deletion of mobC results in a reduction in cell-cell fusion, but does not completely abolish it, as is the case in S. macrospora and N. crassa (Bernhards and P€ oggeler, 2011;Maerz et al., 2009), suggesting that MobC may be less tightly regulated in E. festucae than in S. macrospora and N. crassa, which have different life cycles.
In culture, E. festucae conidiation is sparse, whereas DproA, DnoxA, DmpkA and DmkkA mutants show a hyperconidiation phenotype and a loss of cell-cell fusion, suggesting that the regulatory circuits controlling these two phenotypes are linked . In addition, DmpkA and DmkkA mutants form IHH, structures that have not been observed in the WT strain (Becker et al., 2015). Deletion of mobC in E. festucae also results in an increase in conidiation and IHH formation. In contrast, deletion of mek-1 and mak-1 and the STRIPAK complex genes ham-3 and mob3 in N. crassa, and ham-3 homologues strA in A. nidulans and str1 in C. graminicola, reduces both conidiation and cell-cell fusion, indicating that the regulatory circuits for these developmental processes are 'wired' differently in these fungi (Maerz et al., 2009;Park et al., 2008;Simonin et al., 2010;Wang et al., 2010Wang et al., , 2016. In addition, no change was observed in the basal phosphorylation levels of MpkA (CWI) and MpkB (PR) kinases in the DmobC mutant. Given the pleiotropic nature of the DmobC mutation, it is difficult to determine whether the DmobC hyperconidiation and IHH formation phenotypes observed are a direct effect of the mobC deletion or an indirect effect. Cell-cell fusion is important for colony nutrient transfer, and the reduction or loss of fusion may induce starvation and impair the transport of nutrients to isolated hyphae.
In the grass host, E. festucae hyphae absorb nutrients directly from the intercellular spaces of the leaves. Mutations that result in prolific growth in planta or disrupt the formation of the hyphal network are likely to trigger a starvation response. This is most probably caused by multiple hyphae per intercellular space requiring and absorbing an excess of nutrients from the apoplast, or from the loss of the hyphal network, impairing nutrient transfer between hyphae. Analysis of the genes up-and down-regulated in the DproA, DnoxA and DsakA associations is consistent with this hypothesis . The genes that are significantly up-regulated in all three mutant associations include those involved in primary metabolism, peptide and sugar transport, and host cell wall degradation. Key genes that are down-regulated in this core gene set include genes involved in secondary metabolism and some genes that encode small secreted proteins.
As reported previously for the DproA, DnoxA, DsakA and DsidN mutants (Eaton et al., 2010;Johnson et al., 2013;Tanaka et al., 2006Tanaka et al., , 2013, DmobC also forms a pathogenic-like interaction with L. perenne; IHH are frequently observed, as is vascular bundle colonization and breakdown of the hyphal network in the leaves. Although some cell-cell fusions were observed in planta, the presence of highly convoluted hyphal structures in the intercellular spaces of DmobC-infected plants suggests that cell-cell fusion may also be impaired within the host. Interestingly, DmobC hyphae within the vascular bundles are electron dense, suggesting that they are much healthier than the mutant hyphae growing in the intercellular spaces. This location-specific phenotype observed for the DmobC mutant is consistent with the hypothesis that prolific growth, together with impaired hyphal fusion, triggers a starvation response in hyphae located outside of the nutrient-rich vascular bundles (Becker et al., 2015;Eaton et al., 2015).
The homologue of PRO11, str1, in C. graminicola is required for full virulence in maize and, although Dstr1 strains produce functional appressoria, infection and further colonization are attenuated (Wang et al., 2016). Unlike pathogens, E. festucae is not known to colonize leaves by the formation of appressoria-like structures on the leaf surface. However, a recent study has shown that E. festucae does form appressoria-like structures, named expressoria, that allow endophytic hyphae to breach the cuticle from the inside of the leaf to form a net of epiphyllous hyphae on the surface of the leaf. Deletion of noxA and noxR abolishes expressoria formation in E. festucae and, instead, the mutant hyphae form convoluted branched structures under the surface of the leaf cuticle (Becker et al., 2016). Although some WT-like expressoria structures were observed on the adaxial surface of L. perenne leaves infected with the DmobC mutant, more commonly, subcuticular, highly convoluted structures were observed, suggesting that MobC is important, but not absolutely required, for expressoria development. In this regard, the DmobC mutant of E. festucae is very similar to the str1 mutant of C. graminicola (Wang et al., 2016).
In conclusion, E. festucae contains highly conserved PRO11, PRO22, PRO45 and MOB3 homologues, suggesting that this fungal endophyte has a functional STRIPAK complex. A phenotype analysis of the E. festucae DmobC mutant showed that MobC is essential for the maintenance of a mutualistic symbiotic interaction between E. festucae and L. perenne, and plays an accessory role in regulating cell-cell fusion and expressorium development. The DmobC mutant phenotypes observed are similar to those previously reported for the mutants DmpkA and DmkkA in the CWI MAP kinase signalling pathway (Becker et al., 2015), suggesting that MobC, MpkA and MkkA are involved in the same signalling pathway. Whether, E. festucae MobC is directly linked to the CWI MAP kinase pathway, as proposed for N. crassa (Dettmann et al., 2013), remains to be tested.

Strains and growth conditions
Cultures of Escherichia coli were grown overnight in LB (Luria-Bertani) broth or on 1.5% LB agar containing 100 lg/mL ampicillin, as described previously (Miller, 1972). Cultures of E. festucae were grown on either 2.4% (w/v) PD, 1.5% water agar plates or in PD broth, as described previously (Moon et al., 1999(Moon et al., , 2000.

Plant growth and endophyte inoculation conditions
Endophyte-free seedlings of perennial ryegrass (Lolium perenne cv. Samson) were inoculated with E. festucae (Latch and Christensen, 1985). Plants were grown in root trainers in an environmentally controlled growth room at 228C with a photoperiod of 16 h of light (100 lE/m 2 /s) and, at 8 weeks post-inoculation, were tested for the presence of the endophyte by immunoblotting (Tanaka et al., 2005).

DNA isolation, PCR and sequencing
Plasmid DNA from Escherichia coli cultures was extracted using the High Pure Plasmid Isolation Kit (Roche, Basel, Switzerland) according to the manufacturer's instructions. Fungal genomic DNA used for Southern digests was extracted from freeze-dried mycelium, as described previously (Byrd et al., 1990). Standard PCR amplification was performed with Taq DNA polymerase (Roche) according to the manufacturer's instructions in a volume of 50 lL. Where proofreading activity was required, Phusion High-Fidelity DNA Polymerase (Thermo Scientific, Waltham, MA, USA) was used according to the manufacturer's instructions in a volume of 50 lL. Sequencing reactions were performed using the dideoxynucleotide chain termination method with the Big-Dye TM Terminator Version 3.1 Ready Reaction Cycle Sequencing Kit (Applied BioSystems, Carlsbad, California, USA) and separated using an ABI3730 genetic analyser (Applied BioSystems). Sequence data were then assembled and analysed using MacVector sequence assembly software, version 12.0.5.

Preparation of deletion and complementation constructs
The lists of all plasmids and primer sequences used to prepare the constructs can be found in Tables S1 and S2 (see Supporting Information).
The mobC replacement construct pKG4 was prepared by Gibson Assembly (Gibson et al., 2009) using PCR-amplified pRS426 vector (primers pRS426F and R), 1.2-kb 5 0 (primers KG45 and 46) and 1.5-kb 3 0 (primers KG47 and 48) mobC flanking sequences amplified from E. festucae genomic Fl1 DNA, and a 1.7-kb geneticin resistance cassette (primers genF and R), amplified from pII99 plasmid DNA. The in vitro recombined DNA mixture was transformed into chemically competent Escherichia coli DH5a cells and ampicillin-resistant transformants were screened using Clonechecker for plasmids with restriction enzyme digest patterns predicted from in silico construction of pKG4. The order of the fragments within these clones was verified by DNA sequencing. The mobC replacement fragment contained within pKG4 was excised by XhoI digestion, gel purified and then transformed into E. festucae protoplasts, as described below, using Gen R selection.
The mobC complementation construct pKG8 was prepared by Gibson Assembly (Gibson et al., 2009) using PCR-amplified pRS426 vector (primers pRS426F and R) and a 2.7-kb fragment containing the mobC gene (primers KG66 and 67) amplified from E. festucae genomic Fl1 DNA. This DNA mixture was transformed into Escherichia coli DH5a and pKG8 identified and verified as described above. Plasmid pKG8 was co-transformed with pSF15.15 into E. festucae DmobC protoplasts as described below using hygromycin-resistant (Hyg R ) selection.

Fungal transformations
Epichlo€ e festucae protoplasts were prepared as described previously by Young et al. (2005). Protoplasts were transformed with 2-3 lg of linear DNA excised with restriction enzyme or circular plasmid DNA using the method described by Itoh et al. (1994). Transformants were selected on regeneration (RG) medium (PD supplemented with 0.8 M sucrose) containing either hygromycin (150 lg/mL) or geneticin (200 lg/mL), and nuclear purified by three rounds of subculture on selection medium.

DNA hybridization
Following restriction digestion, E. festucae genomic DNA was separated by agarose gel electrophoresis, transferred to positively charged nylon membranes (Roche) (Southern, 1975) and fixed by UV light cross-linking in a Cex-800 UV light cross-linker (Ultra-Lum, Claremont, California, USA) at 254 nm for 2 min. Labelling of DNA probes with digoxigenin-dUTP (DIG), hybridization and visualization with nitroblue tetrazolium chloride and 5-bromo-4-chloro-3-indolyl-phosphate (NBT/BCIP) were performed using the DIG High Prime DNA Labelling and Detection Starter Kit I (Roche) according to the manufacturer's instructions.

Microscopy
Cultures to be analysed by microscopy were inoculated at the edge of a thin layer of water agar (1.5%), layered on top of a glass microscope slide embedded in a layer of water agar (1.5%), and grown for 5-7 days. Square blocks were then extracted and placed onto new slides, covered with a cover slip, and analysed using an Olympus (Shinjuku, Tokyo, Japan) IX71 inverted fluorescence microscope employing filter sets for the capture of differential interference contrast (DIC) or calcofluor (Fluorescent brightener 28, Sigma; concentration, 3 mg/mL) staining. For the quantification of hyphal fusions, 10 fields were examined at 3400 magnification from three independent colonies. For the quantification of conidiation, three PD agar plates, each containing five colonies, were grown at 228C for 7 days. Conidia were then harvested by scrubbing colonies with 2 mL of sterile water, which was then filtered through glass wool-packed tips. Suspensions of 300 lL were then spread onto PD agar plates for imaging and quantification.
The growth and morphology of hyphae in planta were determined by staining leaves with aniline blue diammonium salt (Sigma) and WGA-AF488 (Molecular Probes/Invitrogen, Eugene, OR, USA) as follows. Infected pseudostem tissue was sequentially incubated at 48C in 95% (v/v) ethanol overnight, and then treated with 10% potassium hydroxide for 3 h. The tissue was washed three times in phosphatebuffered saline (PBS) (pH 7.4) and incubated in staining solution [0.02% aniline blue, 10 ng/mL WGA-AF488 and 0.02% Tween-20 in PBS (pH 7.4)] for 5 min, followed by a 30-min vacuum infiltration step. Images were captured by CLSM using a Leica (Wetzlar, Germany) SP5 DM6000B confocal microscope (488-nm argon and 561-nm diodepumped solid-state laser, 320 or 340 oil immersion objective, NA 5 1.3) (Leica Microsystems). Three photomultiplier tubes (PMTs) were used to capture the emission fluorescence from the dyes, as well as plant autofluorescence. Blue pseudocolour (PMT1, 498-551 nm) was assigned to emission fluorescence from WGA-AF488 excited with the 488-nm argon ion laser. Two pseudocolours were assigned to emission fluorescence from aniline blue and plant autofluorescence (PMT2, 571-633 nm, green; PMT3, 661-800 nm, red) as the result of excitation with the 561-nm DPSS laser (Becker et al., 2016). For TEM, pseudostem sections were fixed in 3% glutaraldehyde and 2% formaldehyde in 0.1 M phosphate buffer, pH 7.2, for 1 h, as described previously (Spiers and Hopcroft, 1993). A Philips (Amsterdam, The Netherlands) CM10 transmission electron microscope was used to examine the fixed samples and the images were acquired using a SIS Morada (M€ uenster Germany) digital camera.

Bioinformatics analysis
Epichlo€ e festucae mobC was identified by TBLASTN analysis of the E. festucae Fl1 (E894) genome (http://csbio-l.csr.uky.edu/ef894-2011) with homologous protein sequences obtained from either the National Center for Biotechnology Information (NCBI) GenBank database (http://www.ncbi. nlm.nih.gov/) or the Broad Institute (http://www.broad.mit.edu). Identity and similarity scores were calculated after CLUSTALW pairwise alignments of sequences (Thompson et al., 1994), using MacVector version 12.0.5, had been performed. The E. festucae genome sequence data, as curated by C. L. Schardl at the University of Kentucky, are available at http://www. endophyte.uky.edu (Schardl et al., 2013). Sequences for each of the genes analysed in this study are available from that site.