The adenylate cyclase gene (acyA) was originally cloned and sequenced from E. festucae var. lolii, previously Neotyphodium lolii (Leuchmann et al., 2014), strain Lp19, isolated from L. perenne. Gene disruption experiments were performed on the closely related strain E. festucae Fl1, isolated from Festuca trachyphylla (Hack.) Krajina. The strains used in this study are presented in Table 1.

Fungi were cultured on potato dextrose agar (PDA, Difco, Le Pont, De Claix, France) at 22°C in an 8 h light, 16 h dark cycle. L. perenne cv. Nui or Samson plants infected with E. festucae Fl1 wild-type or mutant strains were maintained in glasshouse conditions under ambient light and temperature.

We previously cloned and sequenced a 300 bp acyA PCR product from E. festucae var. lolii strain Lp19 (Johnson et al., 2007). The DNA fragment was used as a probe to screen an Lp19 genomic DNA lambda library (Fleetwood et al., 2007) to recover a larger fragment of the gene for functional analysis. A single lambda clone (designated 6163) with homology to the adenylate cyclase gene fragment was recovered (Genbank accession KR815911). Sequencing of the 6047 bp insert indicated the presence of 5819 bp of the adenylate cyclase open reading frame (ORF) plus 228 bp of the acyA 3' region. The first 1419 bp of the open reading frame was missing. This sequence was used to design the gene disruption vector. Later, de novo sequencing of the E. festucae Fl1 genome by Schardl et al. (2013) enabled a complete genomic acyA sequence (gene model Fl1M3.048730, http://www.endophyte.uky.edu/) to be recovered. BLASTn was used to recover the acyA gene from other haploid Epichloë species (Table 2).

The acyA genes from allopolyploids AR3046 (E. baconii/amarillans × E. bromicola) and AR1006 (E. typhina × E. bromicola) were recovered by mapping their genome read pairs (unpublished) to 7 kb genomic scaffolds (containing the acyA gene) of extant strains of their parental species using the “aln” algorithm BWA version 0.7.9a-r786. For AR3046, 43,768,368 78 bp read pairs with an insert size of ~290 bp were individually mapped to E. baconii strain ATCC 200745 and E. bromicola strain AL0434 (Schardl et al., 2014). For AR1006, 21,293,733 100 bp read pairs with an insert size of ~160 bp were mapped to E. typhina strain ATCC 200736 and E. bromicola strain AL0434. The mapped reads were extracted using SAMtools (Li et al., 2009) version 0.1.19-44428cd and imported into Geneious V8.1 (Biomatters, http://www.geneious.com) (Kearse et al., 2012) and consensus sequences for each mapping file were generated based on a threshold of 95% identity.

The acyA disruption vector (pCRVacyhph) was constructed using the Multisite Gateway system (ThermoFisher Scientific, Walden, MA, USA) following the manufacturer's instructions. The entry vector pDONR221/hygromycin, containing the hph cassette from pAN7-1 (Punt et al., 1987), was constructed as previously described (Fleetwood et al., 2007). Two further Multisite Gateway entry vectors, pDONRP2R-P3/AC3′ and pDONRP4-P1R/AC5′ containing a 5′ (3002 bp) and 3′ (3064 bp) region respectively, flanking the integration site in acyA, were created. PCR products were amplified from E. festucae var. lolii Lp19 genomic DNA using primer pairs AC5′-attB4 and AC5′-attB1 (for the 5′ flank) and AC3′-attB2 and AC3′-attB3 (for the 3′ flank) using Platinum Pfx DNA polymerase (ThermoFisher Scientific) according to the manufacturer's instructions. Primer sequences are listed in Supplementary Table 1. The PCR products were purified using the QIAquick PCR Purification Kit (Qiagen, Hilden, Germany), and the products quantified by fluorometric quantitation using the Qubit system (ThermoFisher Scientific). The PCR fragments were then recombined into Gateway donor vectors pDONRP4-P1R and pDONRP2R-P3 using Gateway BP Clonase II. The resulting vectors, pDONRP4-P1R/AC5′and pDONRP2R-P3/AC3,′ along with pDONR221/hygromycin, were then recombined into the destination vector pDESTR4-R3 using Gateway LR Clonase II Plus, to create pCRVacyhph.

Gene disruption experiments were originally designed to disrupt acyA in strain E. festucae var. lolii Lp19, however as this strain proved to be relatively intractable to homologous recombination the closely-related strain E. festucae Fl1 was used instead. Homologous recombinants were obtained by PEG-mediated transformation of protoplasts. Protoplasts were prepared using the method of Young et al. (1998), except that 10 mg/ml of Glucanex (InterSpex Products, San Mateo, CA, USA) was used to digest cell walls for 3 h at 30°C with shaking (100 rpm). E. festucae Fl1 was transformed using 5 μg of each plasmid by the method of Vollmer and Yanofsky (1986) with modifications (Itoh et al., 1994). Protoplasts were co-transformed with plasmid pCRVacyhph (see above) or pTEFEGFP (EGFP fused to the tef2 promoter from Aureobasidium pullulans) (Vanden Wymelenberg et al., 1997) plus either pPN1688 (Young et al., 2005) or pII99 (Namiki et al., 2001) for resistance against hygromycin B or geneticin (both Gibco, ThermoFisher Scientific) respectively. Transgenic colonies were selected on PDA containing 150 μg/mL or 200 μg/mL of hygromycin B or geneticin respectively, and regenerating colonies were purified to homogeneity by sub-culturing hyphal tips onto fresh selective media three times. A number of colonies were analyzed by Southern-blot hybridization to confirm that each contained a single integration of the hygromycin resistance cassette at the desired locus. Genomic DNA was extracted from putative recombinant strains using the method of Byrd et al. (1990) and 2 μg of DNA was digested to completion with HindIII. The DNA was subjected to standard agarose gel (0.8% w/v) electrophoresis, transferred to nylon (Hybond N⁺, GE Healthcare, Buckinghamshire, UK) and the membrane hybridized independently against two dideoxygenin-labeled probes following the manufacturer's instructions (Roche, Basel, Switzerland). Probe 1 (482 bp) was amplified by PCR using the TripleMaster polymerase system (Eppendorf, Hamburg, Germany) using primers ACseqrev4120/ACseqfor3657 and probe 2 (644 bp) was synthesized using primers AC SeqIntRev2/ACM13ForRev (see Supplementary Table 1 for primer sequences).

To complement the mutation, the full length wild-type acyA was PCR-amplified from E. festucae Fl1genomic DNA using Platinum Pfx Polymerase (ThermoFisher Scientific) and primers AcyAF and AcyAR. The 8371 bp fragment comprising 859 bp upstream of the putative start site, 7170 bp of the open reading frame and 342 bp of the 3′ un-translated region was purified according to the manufacturer's instructions (DNA Clean and Concentrator -5, Zymo Research, Irvine, CA, USA) and protoplasts of ΔacyA34 and ΔacyA42 disruption mutants transformed as described above. Ectopic integration of wild-type acyA was confirmed by PCR using primers ACKOF/R (data not shown).

Mycelial sections (approximately 1 mm²) were inoculated onto PDA plates supplemented with 2.5, 5.0 or 7.5 mM cAMP sodium salt (Sigma-Aldrich, St. Louis, MO, USA). Three control strains (E. festucae Fl1 wild-type plus two strains with ectopic insertions, acyA19 and acyA49) and three independent disruption mutants (ΔacyA34, ΔacyA42, and ΔacyA47) were inoculated in triplicate onto each medium supplemented with cAMP (including a no-cAMP control) in a randomized design. The radial measurement (mm) of the colony was taken at approximately 24 h intervals over 165 h. The first measurement was taken 20 h after the colonies were inoculated. The growth rate of each strain was obtained by least squares regression and the slope of the line used to represent the rate of radial growth in mm/h. Analysis of variance was used to estimate the effects of strain and cAMP concentration, and their interaction. Means for each strain and cAMP combination were obtained, together with the least significant difference (LSD—calculated at the 5% significance level) between means. The residual plot from the ANOVA was checked and displayed no evidence of heterogeneity of variance, thus the pooled LSD was used for comparing means at the same level of cAMP.

Mycelial sections were inoculated into incisions created in the SAM of sterile 5 days old L. perenne seedlings growing on 1.5% (w/v) water agar, according to the method of Latch and Christensen (1985). Strains inoculated included the wild-type, mutants ΔacyA34, ΔacyA42 and ΔacyA47, plus two independent complementation strains, ΔacyA34/acyA and ΔacyA42/acyA. Inoculated plants were then maintained at 22°C in the dark for 7 days, followed by 22°C in the light for 7 days, before being transplanted into potting mix and maintained under glasshouse conditions. After 12 weeks, six tillers from each plant were tested for endophyte infection by tissue print immuno-assay using an Epichloë-specific polyclonal anti-serum (Simpson et al., 2012). Plants were inoculated as described on two independent occasions.

E. festucae Fl1 wild-type and acyA deletion mutants were grown in triplicate on potato dextrose broth (PDB, Difco, Le Pont, De Claix, France), either 1X or diluted 1:100 in water, containing 0.8% (w/v) agarose. The broths were inoculated and grown for 5 days at 22°C in continuous light. Alternatively, cultures were grown on the same medium on sterile microscope slides and incubated as described above. If grown in a culture dish, 0.5 cm² of agar was cut from the outer edge of the colony, mounted onto a microscope slide, and a drop of water and a cover glass placed directly onto the specimen. If grown on a microscope slide, a drop of water was added to the mycelium and a cover glass applied. Cultures were imaged by bright field microscopy using a BX50 fluorescent microscope (Olympus, Tokyo, Japan) and a 40X UPLANFLN objective with a 0.75 numerical aperture. Images were taken using an Olympus Colorview III camera with AnalySIS^B image processing software. Endophytes in the epidermal layer of the host leaf sheath were stained with 0.15% (w/v) aniline blue and examined by bright field microscopy as described previously (Christensen et al., 2008). Leaf sheaths from two tillers per plant, and at least three infected plants per strain in each inoculation experiment, were examined.

To examine hyphae in the host shoot apex, semi-thin (0.5–1 mm) longitudinal or transverse sections were taken through the shoot apex and true stem of at least six independent L. perenne tillers infected with each strain and mounted on a microscope slide in water. CLSM images were captured on an inverted FluoView FV10i Confocal Laser Scanning Microscope (Olympus). For visualizing EGFP the excitation wavelength was 457 nm and detection wavelength was between 465 and 565 nm. Two dimensional images were taken using the 10x phase contrast objective, numerical aperture 0.4 (equivalent to UPLSAPO 10x). To examine hyphal morphology in the expansion zone of the developing leaf, young leaves of approximately 5 cm in length were dissected from tillers and cut in half along the midrib. A 1 cm section from the base of a halved blade was mounted in water on a microscope slide for imaging as described above. For each image, the optimal depth to show the most hyphae possible in a sample was selected, and the laser sensitivity then adjusted to immediately below saturation levels.

Endophyte-infected pseudostem material from two plants each infected with wild-type, ΔacyA34, ΔacyA42, or ΔacyA34/acyA, was dissected from L. perenne plants 0.5 cm above the tiller crown, and 0.5 mm transverse sections fixed in 2% (w/v) formaldehyde and 3% (w/v) glutaraldehyde in 0.1 M sodium/potassium phosphate (pH 7.2) for 2 h at room temperature. The samples were washed three times in 0.1 M sodium/potassium phosphate (pH 7.2) and post-fixed in 1% (w/v) osmium tetroxide in the same buffer for 1 h. After washing again as described above, the samples were subjected to a graded acetone/water series with 10 min each in 25, 50, 75, and 95% (v/v) acetone, followed by concentrated acetone for 2 h. The samples were embedded in Procure 812 resin (ProSciTech, Kirwan, Qld. Australia):acetone (50:50) overnight under constant stirring and then in 100% resin overnight. The resin was refreshed for a further 8 h and then mounted in 100% resin at 60°C for 48 h. Ultra-thin sections were collected onto a copper grid and stained with saturated uranyl acetate in 50% (v/v) ethanol for 4 min and then in lead citrate for a further 4 min. Specimens were examined at 46 000X magnification on a Philips CM10 (Philips Electron Optics, Eindhoven, The Netherlands) transmission electron microscope. Micrographs of at least 12 hyphae in developing leaves and in the second mature leaf sheath were captured. The thickness of hyphal cell walls was measured at eight equidistant positions around the circumference of each hypha using open source ImageJ 1.45 s software (Wayne Rasband, National Institutes of Health, USA). The cell wall thickness measurements were analyzed using one way ANOVA to compare the strains. The least significant differences (P = 0.05 level) between means were calculated.

Superoxide radicals in E. festucae in axenic culture were stained using NBT (Sigma-Aldrich) as described by Tanaka et al. (2006). Three replicates for each strain were included in the analysis and the experiment was repeated twice. Strains were grown on PDA for 1 week at 22°C under 8 h light and 16 h dark, and mycelia stained for 5 h at 22°C in continuous light by incubation in 20 μL of 0.05% (w/v) NBT dissolved in 0.05 M sodium phosphate pH 7.5. The reaction was stopped by removing the stain and adding 40 μL of absolute ethanol. The ethanol was removed after 5 min, and mycelia were analyzed at 400X magnification under bright field illumination using an Olympus BX50 compound microscope with a 40X UPLANFLN objective and a 0.75 numerical aperture. Images were taken using an Olympus Colorview III camera with AnalySIS^B image processing software.

Genomic DNA was extracted from freeze-dried pseudostems (leaf and blade tissue between the crown and the first ligule of a tiller) dissected from three tillers of each plant using the DNAeasy Plant Kit (Qiagen). The plants were infected with either the wild-type or ΔacyA42 mutant strain. There were three replicate plants per strain. Hyphal biomass (expressed as endophyte concentration) was determined by quantitative PCR of the single copy E. festucae NRPS-1 gene (EFM3.005350, http://www.endophyte.uky.edu/) from 1 ng of genomic DNA using a MyiQ™ cycler (Bio-Rad Laboratories, Hercules, California, USA), with primers 1-1F, and 1-1R (Supplementary Table 1) as described (Rasmussen et al., 2007; Liu et al., 2011). Hyphal biomass between strains was compared using one way ANOVA. The least significant differences (P = 0.05 level) between means was calculated.

A partial acyA gene (1419–5816 bp) was initially recovered from E. festucae var. lolii (Lp19). Subsequent to functional characterization of this gene in E. festucae Fl1 (reported here), the Fl1 genome became available and the full length acyA gene recovered (gene model Fl1M3.048730, http://www.endophyte.uky.edu/). Analysis of the partial and full length conceptual AcyA proteins from the Lp19 and Fl1 strains respectively, confirmed the presence of the key motifs consistent with fungal Class III adenylate cyclases, including the domains for adenylate cyclase G-alpha binding (IPR013716), ras association (IPR000159), leucine rich repeats (IPR001611, IPR003591, IPR025875), protein phosphatase 2C-like (IPR001932), and the adenylate cyclase (IPR001054) catalytic core (Figure 1A). The E. festucae var. lolii acyA sequence contains a 66 bp indel that is not present in the E. festucae Fl1 acyA. Consistent with other fungal species, Southern-blot hybridization confirmed that E. festucae Fl1 contains a single acyA gene (Figure 1C) and BLASTn analysis of the genomes of other Epichloë species (Schardl et al., 2013) also suggests the presence of a single acyA gene in the haploid strains examined (Table 2).

Since genome hybridization has been a relatively common occurrence in the Epichloë genus, we investigated whether two recently-sequenced allopolyploid strains had retained both copies of the acyA gene from their progenitors, with the attendant prospects for neofunctionalisation (as is observed in mammals). Reads from the genomes of strains AR1006 (E. uncinata) and AR3046 (Epichloë sp.) were mapped independently to the haploid genomes of extant strains related to the original progenitors. E. uncinata (AR1006) is a hybrid between E. bromicola and E. typhina (Craven et al., 2001; Moon et al., 2004), and AR3046 is a hybrid between E. bromicola and the E. baconii/E. amarillans clade (unpublished). Two acyA genes were recovered from each allopolyploid genome examined (Table 2), consistent with a gene originating from each of the species that contributed to the allopolyploid. The acyA homeologs from AR3046 (KT732649 and KT732650) and AR1006 (KT732647 and KT732648) were then mapped back to the genomes of the predicted progenitor species, which confirmed that each acyA gene mapped with higher identity to one or the other of the species that contributed the genes to the allopolyploid (Supplementary Table 2A). The acyA genes from AR3046 encoded conceptual full-length proteins indicating that both genes had the potential to be functional, however in AR1006 one of the genes (KT732647) encoded a full length protein while the other, KT732648, encoded a conceptual protein truncated after amino acid position 686. This gene has also lost a triplet 232–234 bp from the start codon and sustained a number of other deletions relative to the presumed parent E. typhina (Supplementary Table 2B). Alignment of acyA genes within each allopolyploid strain indicated they had no more identity between each other than was found in the comparisons between genes from other species (Supplementary Table 2C) confirming that the genes came together through a genome hybridization event and was not the result of gene duplication.

An acyA gene disruption vector (pCRVacyhph) was constructed using flanking regions designed from the partial E. festucae var. lolii Lp19 sequence, and gene disruption experiments were performed in strain Fl1 as the frequency of homologous recombination in Lp19 is extremely low. Insertion of the hygromycin cassette introduced a stop codon into the acyA open reading frame upstream of the catalytic core domain (Figures 1A,B). Three colonies (ΔacyA34, ΔacyA42, and ΔacyA47), each with a single gene disruption event and no ectopic insertions, were identified by PCR and confirmed by Southern-blot hybridization (Figure 1C). The recombination breakpoints of the mutant strains differed with respect to the presence or absence of the 66 bp indel (Figure 1C), presumably due to differences in the recombination site during homologous recombination in the disruption mutants. E. festucae ΔacyA34 contained the insertion, while ΔacyA42 and ΔacyA47 did not. The integrity of each gene immediately flanking the disrupted acyA was checked by PCR to ensure that the disruption event was confined to the acyA gene. Two primer pairs spanning the intergenic region and the flanking gene 5′ and 3′ of the acyA were used. No evidence of untargeted rearrangements were detected (Supplementary Figure 1). Two further colonies (acyA19 and acyA49) retained the intact acyA gene plus an ectopic insertion of the gene replacement vector and were used as transformation controls.

The three independent E. festucae mutants, ΔacyA42, ΔacyA34, and ΔacyA47, grew more slowly in axenic culture compared with the wild-type, or control strains acyA19 and acyA49 (Figure 2A). The mutant colonies were also highly compact compared to the controls. The radial growth rates of ΔacyA42, ΔacyA34, and ΔacyA47 increased in a dose-dependent manner in response to supplementation of the media with exogenous cAMP (Figure 2B). The growth rates of E. festucae ΔacyA42 and ΔacyA47 were statistically indistinguishable from wild-type when the medium was supplemented with 7.5 mM cAMP. E. festucae ΔacyA34 grew more slowly than the other strains under all conditions; however its growth rate on PDA containing 7.5 mM cAMP was almost 2 fold higher than when growing on PDA alone. The growth rates of the control strains (wild-type, acyA19 and acyA49) were not altered by cAMP supplementation (Figure 2B) indicating that endogenous cAMP does not limit growth of strains with functional AC enzymes.

E. festucae ΔacyA47 produced a faster-growing sector (named ΔacyA47var) on PDA supplemented with 150 μg/mL hygromycin. Southern-blot hybridization confirmed that the sector was identical to ΔacyA47 at the disruption locus (data not shown). Strains ΔacyA34 and ΔacyA42 did not produce overt spontaneous growth revertants, however if repeatedly sub-cultured onto fresh media, gradually grew faster until their growth rates were similar to wild-type (data not shown). PCR (using primers ACKO F and ACKO R) was used to check the fidelity of the integration locus during this period (data not shown). We saw no evidence of a loss of the integrated DNA over time in these colonies suggesting that the changes in phenotype were due to mutations or epigenetic changes at other loci.

We next examined the hyphal morphology of acyA disruption mutants in culture by bright field microscopy. In wild-type cultures, hyphae were long and straight, relatively sparsely branched and produced lateral branches at the proximal end of compartments immediately adjacent to septa several compartments behind the tip (Figure 3). Conversely, the hyphae of E. festucae ΔacyA34, ΔacyA42, and ΔacyA47 were highly convoluted, sometimes forming thick aggregates or cables, and produced multiple lateral branches. Excessive lateral branches accounted for the compact nature of the mutant colonies. The morphology of strains complemented with the wild-type acyA gene were similar to the wild-type strain (Figure 3) confirming the role of cAMP in suppressing lateral branches. This phenotype was reproducible, however similarly to the observations on effects of AC disruption on colony growth rate, hyphal morphology in mutant colonies was not stable over time and reverted to the wild-type form if the cultures were maintained continuously in axenic culture.

Synthesis of ROS is essential for mutualism in the E. festucae Fl1/L. perenne interaction, and disruption results in hyper-colonization of host tissues, stunting and premature leaf senescence (Tanaka et al., 2006, 2008). In order to determine whether the morphology of the E. festucae ΔacyA disruption mutants was linked to changes in ROS production, strains were grown in culture on microscope slides and stained with NBT. NBT forms a blue precipitate on exposure to superoxide ions, and blue deposits were typically observed in the hyphal apices of the E. festucae Fl1 wild-type (Figure 4) and occasionally in some compartments behind the tip (data not shown). NBT staining of E. festucae acyA disruption mutants revealed a substantial reduction in superoxide radicals relative to the wild-type strain (Figure 4). The localization of superoxide radicals was similar between the wild-type and mutants, however superoxide ions were not detectable in the majority of mutant hyphae (Figure 4). Strains complemented with the wild-type acyA gene were able to produce ROS at similar or higher levels than the wild-type confirming that cAMP signaling directly or indirectly regulates accumulation of superoxide ions in E. festucae.

To determine the role of fungal cAMP signaling in E. festucae during mutualistic interactions with L. perenne, mycelia of strains carrying the disrupted gene were inoculated though a small incision into young seedlings. In each inoculation experiment, between 25 and 50% of inoculated plants (n = 25) became infected with wild-type, ΔacyA34, ΔacyA42, and the complementation strains, however despite repeated attempts (n > 75 plants), no plants infected with the original E. festucae ΔacyA47 strain were obtained. It is unclear why this strain was apparently incapable of host colonization since the acyA locus was identical to ΔacyA34 and ΔacyA42. Inoculation of plants with E. festucae ΔacyA47var produced infected plants with similar frequency to the other strains. This strain was not included in further experiments due to its unstable phenotype.

We next investigated whether a functional fungal cAMP signaling pathway is required for a mutualistic interaction between the symbionts and the host grass. Three to five independent plants for each strain were examined. Consistent differences in the growth and phenotype of plants infected with wild-type, ΔacyA, or Δacy/acyA complemented strains were not observed, although plants infected with mutant strains did express a variable marginally-stunted phenotype (Supplementary Figure 2). After 3 months, hyphae within the mature leaf sheaths of infected plants were stained with aniline blue and analyzed by bright field microscopy. Wild-type hyphae in this tissue were long and straight, seldom branched, and oriented in the direction of leaf growth. In contrast, hyphae of the two independent acyA mutants were highly branched and convoluted (Figure 5). The phenotypes of ΔacyA34 and 42 in all the plants examined (at least three plants per strain per experiment) were highly consistent. The complementation strains ΔacyA34/acyA and ΔacyA42/acyA both resembled the wild type (Figure 5). To further investigate the role of cAMP signaling in endophyte colonization of the host at different developmental stages, we transformed ΔacyA42 with vector pTEFEGFP (for constitutive EGFP expression in E. festucae) and inoculated EGFP-expressing strains into L. perenne seedlings for examination by confocal microscopy. Wild-type E. festucae Fl1 transformed with the same plasmid was used as a control. An EGFP-expressing acyA-complemented strain was not included in this experiment as it was not technically feasible to conduct a third transformation on this mutant strain. We first examined the hyphae in longitudinal sections taken through the shoot apex at the base of the tiller. In host tissues immediately below the meristem, and in the youngest developing leaves, the mycelial density of ΔacyA42/EGFP was similar to wild-type controls (Figures 6I,J), however in young leaf sheaths above the shoot meristem (lower leaf sheath), hyphae of the mutant strain appeared more numerous than wild-type (Figures 6E–H). A key feature of E. festucae-L. perenne mutualism is that once host tissues are mature and have stopped expanding, colonizing endophytes also cease growing (Tan et al., 2001). To determine whether E. festucae ΔacyA42/EGFP was still capable of responding to host developmental cues, and to stop growing, epidermal peels from mature (upper) leaf sheaths were examined. Contrary to the uniform appearance of wild-type hyphae in this tissue, disruption of the cAMP signaling pathway resulted in a dense and heavily-branched mycelium (Figures 6A–D). To confirm the microscopy observations, the biomass of the wild-type and ΔacyA42 strains in the pseudostem of these plants was determined by quantitative PCR of a single copy E. festucae gene. Hyphal biomass (expressed as endophyte concentration) was an average of 2081 (SE 222.7) gene copies per ng of plant and endophyte genomic DNA for wild-type, and 3949 (SE 226.1) for the ΔacyA42 mutant, 1.9 fold higher than wild-type (Figure 7). The means of the technical replicates for each biological replicate per strain were compared using the Student's T-test. The T-test indicated that the mean hyphal biomass of the mutant strain was significantly greater than the wild-type (P = 0.004).

The impact of acyA disruption on hyphal ultrastructure in planta was investigated by TEM of infected L. perenne tillers. Tillers are comprised of bundles of leaves ranging in development from the most immature (in the middle of the tiller) to the fully mature outer leaf sheath. Transverse sections through the base of tillers were fixed and embedded. Ultra-thin sections of tillers infected with wild-type, ΔacyA34, ΔacyA42, and the ΔacyA34/acyA complementation strains were stained with osmium tetroxide and examined by TEM. The thickness of the cell wall was measured at eight equidistant positions around 12 hyphae from the immature leaf blade (Figure 8A) and the second fully mature leaf sheath (Figure 8B) of each tiller. In immature leaves, where both the plant and endophyte grow by intercalary growth, mean cell wall thickness of ΔacyA34 and ΔacyA42 was 70.3 and 62.8 nm respectively, significantly thinner (P ≤ 0.05) than the wild-type which had an average thickness of 111.3 nm (Figure 8A). The phenotype was nearly fully restored in the ΔacyA34/acyA complemented strain which had a mean cell wall thickness of 101.04 nm, indicating that cAMP signaling positively regulates cell wall biogenesis during intercalary growth. Conversely, in the fully developed leaf sheath where plant and fungal cells are no longer growing, there were no significant differences in cell wall thickness between any of the strains (Figure 8B). Cell walls of wild-type E. festucae hyphae in the leaf sheath were thicker than those in the immature leaf (139 nm vs. 111.3 nm respectively, P ≤ 0.001) confirming previous reports of thinner cell walls in E. festucae Fl1 hyphae in developing vs. mature host tissues (Christensen et al., 2008).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.