Epichloë scottii sp. nov., a new endophyte isolated from Melica uniflora is the missing ancestor of Epichloë disjuncta

Here we describe a new, haploid and stroma forming species within the genus Epichloë, as Epichloë scottii sp. nov. The fungus was isolated from Melica uniflora growing in Bad Harzburg, Germany. Phylogenetic reconstruction using a combined dataset of the tubB and tefA genes strongly support that E. scottii is a distinct species and the so far unknown ancestor species of the hybrid E. disjuncta. A distribution analysis showed a high infection rate in close vicinity of the initial sampling site and only two more spots with low infection rates. Genetic variations in key genes required for alkaloid production suggested that E. scottii sp. nov. might not be capable of producing any of the major alkaloids including ergot alkaloid, loline, indole-diterpene and peramine. All isolates and individuals found in the distribution analysis were identified as mating-type B explaining the lack of mature stromata during this study. We further release a telomere-to-telomere de novo assembly of all seven chromosomes and the mitogenome of E. scottii sp. nov.


INTRODUCTION
Epichloë species (Clavicipitaceae, Hypocreales) live in a pleiotropic, constitutive symbiosis with grasses of the subfamily Pooideae. During the vegetative state, they colonize the intercellular space of the aerial plant tissue without causing any visible pathogenic symptoms (Scott et al. 2012). Most species are capable of vertical transmission by infection of ovules of developing seeds . For species reproducing asexualy, distribution via clonal growth by tillering of the host and seed transmission are the only ways of reproduction. For sexual morphs of Epichloë, horizontal transmission starts with a switch from restricted endobiotic to proliferative epibiotic growth. This results in the forming of stromata, which enclose the developing inflorescences causing the typical "choke disease" (Scott and Schardl 1993). The mating system of Epichloë is bipolar and heterothallic. Flies act as the vector and transfer spermatia to stromata of the opposite mating type. After karyogamy and meiosis, ascospores are ejected and infect new host plants.
Most Epichloë species known only as an a asexual morph arose from interspecific hybridizations of different haploid Epichloë ancestors (Schardl 2010). The genus Epichloë currently comprises 15 haploid and 24 hybrid species (Leuchtmann et al. 2014;Leuchtmann et al. 2019;Campbell et al. 2017;Shymanovich et al. 2017;Tian et al. 2020) with E. sinensis, a hybrid of haploid species from the E. poae and the E. sibirica clades, being the latest addition (Tian et al. 2020). With the exception of the hybrid E. disjuncta, and the contributors of the "Lolium-associated clade" in certain Open Access IMA Fungus hybrid species found in some Lolium spp., all ancestors of the described hybrid Epichloë spp. can be attributed to existing phylogenetic clades. For E. disjuncta, only one of the ancestors falls into a known clade related to endophytes of Brachipodium hosts (E. typhina or E. sylvatica) (Leuchtmann and Oberhofer 2013). The second ancestor was until now believed to be either extinct or not yet found.
So far, there are 20 genomes of Epichloë available, representing 15 species. Only the genome of Epichloë festucae Fl1 is assembled on a chromosome level. It consists of seven chromosomes and one mitogenome with a genome size of 35 Mb and a GC content of 43.9% (Winter et al. 2018).
Melica uniflora, a perennial rhizomatous cool-season grass of tribe Meliceae that grows in shady places in Europe, northwards to Scotland and Southwest Finland and eastwards to Moldavia (Tutin et al. 2010) was until now not reported to be a host for Epichloë (White and Baldwin 1992;Wilson et al. 1991).
In the "Butterberggelände" nature reserve in Bad Harzburg, Germany, several M. uniflora individuals showing stromata were found. The plants were sampled to isolate the fungal candidate causing the infection. Here we describe a new haploid Epichloë species isolated from M. uniflora, and report its telomere-to-telomere de novo genome assembly.

Biological materials/fungal isolation
Two M. uniflora individuals bearing stromata were collected in June 2020 in the "Butterberggelände" nature reserve (NSG BR 004) 51° 53′ 13.5″ N, 10° 34′ 36.9″ E. This is a melic grass / beech forest characterized by stony limestone weathered soils, which are well supplied with nutrients. One plant individual was sampled with attached roots and brought to the greenhouse, while the other only consisted of aerial tissue. The grasses were identified based on the flowering tillers and identity was confirmed using DNA-based identification techniques (see below).
Endophytes were isolated from M. uniflora pseudostems according to "Basic Protocol 4", described by Florea et al. (2015). Another isolation was made directly from the stroma of the greenhouse specimen by surface swab, followed by a series of subculturing steps.

Morphological examination
Fungal structures were examined and photographed using a Zeiss Axioskop 2 plus compound microscope and an Olympus SZX 12 stereo microscope equipped with a Jenoptik ProgRes ® digital camera. Images were recorded using CapturePro 2.8 software (Jenoptic, Jena, Germany). Growing mycelia mounted in water, as well as slide cultures (Gams 1998) were used to illustrate fungal structures in different developmental stages. Nomarski Differential Interference Contrast (DIC) optics were used for observation and measurements. All measurements were obtaind from cultures growing on potato dextrose agar (PDA, Merck) and are given as × 1-× 2 (× 3 ± SD), with × 1 = minimum value observed, × 2 = maximum value observed, × 3 = average, and standard deviation (SD), followed by the number of measurements (n). Color changes of fungal structures formed in culture were checked using 3% potassium hydroxide (KOH) watery solution. Color codes used in the description were determined according to https:// www. ral-farben. de/ en/ all-ralcolou rs.

Growth rate studies
Growth rates were determined at various temperatures from 5 to 35 °C at 5 °C intervals in the dark. Agar disks of 4 mm diam, excised from the margin of a young PDA culture were placed onto four replicate plates of PDA, cornmeal agar (CMA, Fluka), and yeast malt agar (YM: 3 g yeast extract, 3 g malt extract, 5 g peptone, 10 g glucose, 20 g agar, 1L deionized water). The colony diameter was measured weekly for a 4 wk period.

Confocal laser-scanning microscopy
Stromata and substromata samples were cut in approx. 0.1 cm cross sections by hand with a scalpel blade. Leaf sheath and blade samples were cut in 2 cm long subsamples. All samples were stained with Wheat Germ Agglutinin, Alexa Fluor ™ 488 Conjugate (ThermoFisher Scientific, MA, USA) and Aniline Blue diammonium salt (C 37 H 32 N 5 O 9 S 3 ) and eventually with propidium iodide (both: Sigma-Aldrich Chemie, Traufkirchen, Germany) as described by Becker et al. (2018). Samples were transferred to microscopic slides and embedded in staining solution for 10 min before microscopic examination. Confocal laser scanning microscopy (CLSM) was done using a Leica TCS SP8 as described in Becker et al. (2018).

Molecular studies
DNA extraction from single host plants and seeds was performed using DNeasy ® Plant Mini Kit (QIAGEN, Germany) following the standard procedure according to Page 3 of 18 Thünen et al. IMA Fungus (2022) 13:2 the manufacturer. DNA extraction from individuals used for the distribution analysis was performed using the 96 well plate nexttec ™ 1-Step Plant DNA Extraction kit (nexttec ™ Biotechnologie, Germany) following the standard procedure provided by the manufacturer. Genomic DNA from fungal isolates was extracted by transferring roughly 1 cm 2 of fresh mycelium grown on PDA to lysis buffer (150 mM EDTA, 50 mM Tris-HCl, 1% sodium lauroyl sarcosine). After incubation at 70 °C for 30 min, DNA was isolated from the aqueous phase by sequential precipitations with 5 M potassium acetate and isopropanol followed by a washing step with 70% ethanol. Final DNA was resuspended in 50 μl of PCR-grade water.
High molecular weight (HMW) genomic DNA (gDNA) for genome sequencing was extracted using the modified protocol of Mayjonade et al. (2016). 2 × 15 mg of freezedried mycelium (grown as liquid culture in PD broth) were transferred to one 1.6 ml bead-tube filled with 140 mg ceramic beads (Ø1.4-1.6 mm) and 4 ceramic beads (Ø 2.6-2.8 mm) each. Tubes were cooled down in liquid nitrogen before and between 2 bead-beating steps for 15 s in BeadRupter (Biolabproducts, Germany) at speed 4.00. After addition of 600 µl of lysis-buffer (preheated to 65 °C) including RNase A, samples were incubated at 65 °C for 5 min until all mycelium was dissolved. Afterwards, samples were incubated at 50 °C for 30 min and mixed by inverting (20 times) every 10 min. After addition of 200 µl 5 M potassium acetate, samples were placed on ice to cool down and centrifuged at 5,000 × g for 10 min at 4 °C. Supernatant was transferred to a fresh tube and 1 vol. of binding buffer and 1:18 (v:v) of Serapure beads solution were added followed by incubation at room temperature for 10 min in a rotary shaker. Tubes were placed on a magnetic rack for 1 h, supernatant was removed and beads were washed 2 times with 1 ml of wash solution. DNA was eluted by adding 100 µl of preheated (50 °C) elution buffer. DNA was cleaned using 1 vol. of bead-solution according to the protocol by Schalamun et al. (2019). DNA concentration was measured using Qubit and integrity of the DNA was checked on a 0.5% Agarose gel.
For confirmation of identity of the host species, PCR reactions were performed to amplify ribulose-bisphosphate carboxylase (rbcL) in a total volume of 30 µl 1 × Green GoTaq ™ Reaction Buffer, containing 2 ng DNA, 1 U GoTaq ™ DNA Polymerase (Promega), 0.3 mM of each dNTP (Biozym Scientific, Germany), and 0.5 µM each of rbcL_1_for and rbcL_1388_rev primers (Christin et al. 2008) using the following cycling parameters: 94 °C for 3 min, then 40 cycles of 94 °C for 1 min, 48 °C for 30 s, and 72 °C for 2 min followed by 72 °C for 7 min.
PCR products of tefA were cloned using pGEM ® -T Easy Vector System I (Promega). For fungal isolates studied here (DSM111774, DSM111775 and DSM112488), a minimum of eight colonies per isolate containing the tefA sequences were sequenced with primers SP6 and T7. PCR products of ITS, tubB, calM and rbcL were sequenced using the same primers used for the PCR. Sequences obtained were assembled using CLC Main Workbench v20.0.4 (QIAGEN Aarhus A/S). Sequences obtained were deposited in Gen-Bank under accession numbers MZ147888, MZ147889, MZ147863, MZ198217-MZ198219, MZ224334-MZ224338, MZ438661-MZ438663.

Molecular phylogenetic analyses
In order to create the dataset for a multigene phylogeny a search was conducted using the NCBI Nucleotide database. The search resulted in 800 and 1065 sequences of tefA and tubB, respectively. Sequences were extracted from the XML-files, combined in two separated FASTAfiles for each gene. An additional file was created containing species name, accession number, gene, strain, isolate, and sequence definition. This file was used to find matching pairs regarding species and strain/isolate. Information on allele was added manually based on the information given in the sequence definition. A total of 438 sequences could be identified showing matching pairs of tefA and tubB. Both previously created datasets were reduced to those sequences. A subset was created containing one available individual of each species, subspecies and variation based on sequence length and quality. Details of the sequences used for the alignment can be found in the table of taxa (Table 1). Additionally, an exclusively tubB sequences dataset (tubB-only) was prepared based on the alignment published by Leuchtmann et al. (2014). The dataset was expanded by the tubB sequences generated here, both alleles from E. alsodes strain NY 12-14 (Shymanovich et al. 2017), E. hybrida strain NEA11 (Campbell et al. 2017), E. schardlii var. pennsylvanica PA 10 (Shymanovich et al. 2017), three alleles of E. novae-zelandiae strain AL0725 (Leuchtmann et al. 2019), and sequences of both alleles of E. sinensis 57A (Tian et al. 2020). DNA sequences were aligned using the online version of MAFFT v7 (Katoh et al. 2019;Kuraku et al. 2013) adopting the iterative refinement methods L-INS-i for both tubB and tefA genome regions used for the combined dataset, and FFT-NS-i for the adapted tubB-only dataset. Alignments were visually examined and their starts and ends were manually trimmed using AliView v. 1.26 (Larsson 2014). Phylogenetic analyses were applied using Bayesian Inference (BI), maximum-likelihood (ML) and neighbor-joining (NJ) for all datasets. The best-fit model of DNA substitution was estimated using MrModeltest v2.2 (Nylander 2004) under hierarchical likelihood ratio test (hLRT) and the Akaike Information Criterion (AIC). The general time reversible model with gamma distributed substitution rates and invariate sites (GTR + I + G) and the symmetrical model with gamma distributed substitution rates and invariate sites (SYM + I + G) were selected as the best fitting model for the combined aligned dataset, under hLRT and AIC, respectively. Substitution models for the tubB-only dataset selected using MrModeltest were including K80 + G (hLRT) and K80 + I + G (AIC).
For the multigene phylogeny, Bayesian analysis was performed using Metropolis Coupled Monte Carlo Markov chains (MCMCMC) based on both best fitting models settings in MrBayes v3.2 (Ronquist and Huelsenbeck 2003). The process was run for 2,000,000 generations and trees were sampled every 500 Locality (Loc.) is provided as ISO 3166-1 alpha-2 code. N.A is used for North America. '-' in the Reference (Ref.) column indicates unpublished sequences, * marks sequences obtained during this study generations. A 50% majority rule consensus tree was computed only from trees of the plateau, and if, additionally, the split frequencies were below 0.01. A total of 1701 trees representing the "burn-in phase" were discarded and the remaining trees were used to infer posterior probabilities (PP) for the nodes of the majority rule consensus tree. ML and NJ analyses were conducted as previously described (Ashrafi et al. 2017 (Manni et al. 2021) was used to identify conserved gene orthologs independently in each species using the closest BUSCO reference database, hypocreales_odb10. Single copy ortholog protein sequences were passed to OrthoFinder v2.5.4 (Emms and Kelly 2019), which generated a maximum likelihood species phylogeny using RAxML-NG v1.0.3 (Kozlov et al. 2019) based on 2,828 orthologs with < 5% missingness across species. Branch support values were calculated using the Shimodaira-Hasegawa algorithm in FastTree v2.1.10 (Price et al. 2010) with 1000 × resampling.

Genome sequencing and assembly
The genome was sequenced using the previously isolated HMW gDNA following the Oxford Nanopore Technologies (ONT) Genomic DNA by Ligation protocol for the SQK-LSK110 Ligation Sequencing Kit (Version: GDE_9108_v110_revE_10Nov2020). Sequencing run was performed on a MinION 1B in a R10.3 Flow Cell for 72 h. Basecalling was performed on a NVIDIA Jetson AGX Xavier developer kit using ONT Guppy v5.0.11 and the r9.4.1 HAC model. Reads were corrected, trimmed and assembled using Canu (Koren et al. 2017) (snapshot v2.2development + 149 changes (r10258)). Genome size was set to 38 Mb and the correctedErrorRate parameter was set to 0.039 for the assembly step.
Polishing of the genome was performed by mapping high-coverage Nanopore long read data to the assembled genome using minimap2 v2.22 (Li 2018) and Pilon v1.24 (Walker et al. 2014). The genome was submitted to Gen-Bank under the accession numbers CP083245-CP083252 and the corresponding BioProject PRJNA756890 and BioSample SAMN20929797 numbers.

Distribution analysis
To determine the distribution of the fungus studied here in the natural reserve "Butterberggelände", M. uniflora grasses from seven areas evenly distributed along the reserve (sample sites B to H in Fig. 1) were collected at the end of August 2020. Each area had a radius of 5 m. Additionally an area with a radius of 0.5 m was sampled at the site where the initial stroma-bearing individuals had been detected (sample site A in Fig. 1). Within those areas, ten M. uniflora individuals with inflorescence and no visible stromata were collected and checked for the presence and mating type of Epichloë using tubB and mtAC/mtBA PCR amplification. Seeds of the samples were separated and checked for the presence of Epichloë using the same methodology mentioned earlier.

Results
Two fungal strains DSM111774 and DSM111775 were isolated from surface sterilized pseudostems of the two Melica uniflora individuals sampled from sampling sites A and E (Fig. 1). Identity of the plant host was confirmed by rbcL gene sequence blast against the NCBI database.
The specimen with roots was kept in the greenhouse of the Julius Kühn Institute, Braunschweig, Germany. Aerial tissue of this individual died off, but daughter ramets emerged from the rhizome, bearing developing stromata (Fig. 2b). The fungal strain DSM112488 was isolated directly from a developing stroma.
Partial as well as complete choking of the inflorescence on Melica uniflora was observed in the field and bundle cells (Fig. 3g, h). Endobiotic hyphae formed plant exit structures, expressoria, to establish restricted epibiotic hyphal net on the leaf surface (Fig. 3b, e, f ). During stroma formation of Epichloë spp. hyphae switch to prolific endobiotic and epibiotic growth and colonize host vascular bundles. This was also observed in the species described here (Fig. 3c, d).

Sequence comparison and phylogenetic reconstructions
The combined dataset consisted of tefA (927 sites) and tubB (789 sites) partial sequences with a total length of 1716 base pairs (bp). The alignment comprised 65 taxa representing 33 species in the genus Epichloë and the species Claviceps purpurea as the outgroup. The DNA sequences of tubB and tefA obtained from the analyzed specimens of the Epichloë species studied here were identical between isolates. The topologies of the phylogenetic trees were identical without any conflict in supported groupings using Bayesian inference (Fig. 4), neighbor-joining or maximum likelihood. Strains of the fungus described here were highly supported as a monophyletic species group in all analyses, and clustered with E. disjuncta allele 1 in a well-supported clade. Tree topologies of BI analyses conducted based on substitution models were identical. Comparison of tubB and tefA between the strains studied here and E. disjuncta allele 1 showed a 100% identity for tubB and 99.2% and 99.0% identity for tefA of E. disjuncta strains A1_1 and C5a_1 and strain A4_5, respectively.
The tubB-only alignment consisted of 211 sequences with 599 sites. The sequences analyzed represented 39 known Epichloë species, 11 specimens of Epichloë spp., as well the Epichloë species described here. This analysis also showed that the strains examined here formed a highly supported clade together with E. disjuncta, providing sufficient phylogenetic resolution to distinguish this group from all other Epichloë groupings (Fig. 8).
The phylogentic analysis based on 2,828 BUSCO single copy protein orthologs showed an early branching of the species described here after branching of the E.   Description: Infested plants bearing immature stromata. Stromata cylindrical, variable in size, 13-29 mm long, white to lemon-yellow with age. Colonies moderately growing, on PDA at 20 °C reaching 7-8 mm diam (7 d), 14-16 mm diam (14 d), and 26-27 mm diam (21 d); optimum temperature for growth 20 °C; at 5 °C 1 mm (21 d), at 30 °C 3-4 mm (21 d). Optimum temperature for growth on other examined culture media at 20 °C, reaching 30-31 mm diam (CMA, after 21 d), 38-32 mm (YM, after 21 d); no growth observed at 35 °C. Colonies on PDA elevated centrally, surface smooth with dense aerial mycelium, cottony white in the central part to pale creamy at the margin, margins wide and flattened, reverse natural yellow (RAL 095 85 50) in the central part to asparagus-yellow (RAL 095 80 30) at the margin, no exudates, no medium staining (Fig. 2f, g). Hyphae hyaline, thin-walled, septate, forming strands or coils, occasionally anastomosed, bearing the conidiogenous cells.
Distribution: At the end of August 2020, 80 individuals of M. uniflora with inflorescences but no visible stromata were collected at eight sites along a transect in the "Butterberggelände" nature reserve (Fig. 1). The survey resulted in discovery of 18 M. uniflora individuals infected with Epichloë scottii. The distribution is shown in Fig. 1 with the numbers showing the percentage infection rate observed at each sample site. Beyond the initial sample sites A and E, infected M. uniflora was only observed at sites G (10% infection rate) and H (20% infection rate). Because sampling was conducted quite late in the year, the majority of seeds had already fallen. Only four fully developed seeds were recovered from infected individuals. Sequencing results of the seed samples confirmed the presence of E. scottii.
Discussion: According to phylogenetic inference (Fig. 4), E. scottii is closely related to but distinct from E.
(See figure on next page.) Fig. 4 Bayesian inference of the phylogenetic relationship of the fungus described here among Epichloë isolates based on tubB and tefA sequences using SYM + I + G as the nucleotide substitution model. Numbers above nodes are estimates of a posteriori probability (BIpp, ≥ 0.9), and bootstrap values of maximum likelihood (ML) and neighbor-joining (NJ) (≥ 70%), respectively. The topology was rooted with the distantly related species Calviceps purpurea disjuncta. Both taxa form moderately growing colonies on PDA. Epichloë scottii develops ellipsoid to ovoid and small size conidia (4.5 × 3.0 µm) whereas E. disjuncta forms medium-sized conidia (6.9 × 2.7 µm) that are lunate to reniform and often bear an apiculum-like bulge at the base. They also differ in the length of the conidiogenous cells. Epichloë scottii develops conidiogenous cells (14.1 ± 2.8 µm) that are shorter that those of E. disjuncta (33.8 ± 7.3 µm). Although E. scottii was originally isolated from M. uniflora, E. disjuncta was reported from Hordelymus europaeus (Leuchtmann and Oberhofer 2013). Among non-hybrid species studied by Leuchtmann and Oberhofer (2013), E. scottii and E. sylvatica subsp. pollinensis form conidia and conidiogenous cells that are morphometrically similar, however the difference between these species is strongly supported by sequence and genomic comparison (Figs. 4,5).

Genome sequencing and assembly
Initial assembly of the genome resulted in 22 contigs. Six contigs resembled complete chromosomes with telomeres on both ends. Two contigs, each with one telomere showed ribosomal DNA (rDNA) repeats at the broken ends. These two contigs therefore represent a single chromosome broken on the rDNA locus and were joined together. In order to clean up the rDNA locus two partial rDNA units at the flanks of the rDNA locus were identified, together with two complete rDNA repeats in between. One contig was identified as a broken copy of the rDNA locus and could therefore be excluded. The mitochondrial DNA (mtDNA) was present in one contig, which the assembler had identified as a circular DNA. The sequence was reordered to start at the canonical mtDNA start site. The contig contained two copies of the mtDNA genome, so the first copy was retained and the second copy deleted. The assembler identified 11 contigs as bubbles. Mapping of these contigs to the chromosomes identified high identity matches and the contigs could therefore be excluded. One small contig (3510 bp) could not be matched to the other contigs. An NCBI blast search showed that this contig is a contaminant from Escherichia coli, likely part of a cloning vector. This contig can therefore be safely excluded, as it is not an Epichloë sequence. Telomeres were trimmed to the nearest canonical repeat (TAA CCC ) at both ends of each chromosome. Chromosomes were sorted by length from largest to smallest. The overall length of the genome after the polishing step is 37.4 Mb distributed over seven chromosomes and one mitogenome. The overall GC content of the chromosomes is 42.9%. The length of the individual chromosomes as well as their GC content can be found in Fig. 7. The length of the mitogenome is 94,060 bp and its GC content 27.7%.

Genetic variation
All isolates showed the same genotype regarding key genes required for alkaloid production and mating type idiomorphs (Table 2). No genes involved in the biosynthesis of ergot alkaloids and lolines could be detected by Multiplex PCR according to the method of Charlton et al. (2014). Only two indole-diterpene genes (idtF and idtQ) showed a PCR fragment of the expected size. Regarding peramine, all strains were positive for the perA 5′ region and the perA reductase domain, but the perA T2 domain was not detected. PCR markers for determination of the mating-type identified all strains as mating-type B. Presence and absence of genes involved in the alkaloid biosynthesis could be confirmed by crosscheck with the genome assembly of E. scottii.

DISCUSSION
Here we report the first occurrence of an Epichloë endophyte in Melica uniflora. Comparative morphological and molecular phylogenetic evidences indicates it is a new species, given the name Epichloë scottii here. Variation in tubB and tefA sequences compared to other haploid Epichloë species strongly supported that E. scottii is a distinct and monophyletic species in the genus. Phylogenetic analysis and sequence comparison suggests that E. scottii is one of the progenitor species of E. disjuncta (Leuchtmann and Oberhofer 2013), formerly known as Neotyphodium sp. HeuTG-3, and first described by Oberhofer and Leuchtmann (2012). Epichloë disjuncta is a hybrid Epichloë species isolated from the European woodland grass Hordelymus europaeus. Melica uniflora and H. europaeus share the same habitat. In a survey of 44 Querco-Fageta woods north of the Harz Mountains in Germany, H. europaeus could be identified in 16 areas, all of which harbor M. uniflora as well (Zacharias 1993). The second ancestral species of E. disjuncta is related to endophytes of Brachypodium hosts (E. typhina or E. sylvatica) (Leuchtmann and Oberhofer 2013). In the survey of Zacharias (1993), B. sylvaticum was identified in 43 of the 44 woods. This places both hosts of the putative ancestral species and the host of the resulting hybrid in the same habitat.
A third phylogenetic analysis based on a total of 2,828 single copy protein orthologs was conducted. Accordingly E. scottii was placed as an early branching species within the genus Epichloë. The genus of the host of E. scottii, Melica, is itself an early branching genus within the subfamily Pooidae (Saarela et al. 2018) suggesting a co-evolution of fungus and its host as described by Schardl et al. (2008). Sequencing results of PCR products showed only single alleles of the genes tubB, tefA, calM and the ITS region in all isolates. Therefore, E. scottii is considered a haploid Epichloë species. The BUSCO results from the genome analysis labeled almost all genes (> 99.7%) as single copy, also supporting this new species being haploid. The size of the conidia, which is comparable to other haploid Epichloë species (Kuldau et al. 1999) is another strong indicator for the haploid nature of E. scottii. Epichloë species are heterothallic. For successful fertilization of the stromata they require spermatia or ascogonia from individuals of the opposite mating type, which are transmitted via vectors such as female flies of the genus Botanophila (Bultman and Leuchtmann 2003). No stromata on M. uniflora discovered during this study showed any sign of fertilization (no sign of embedded perithecia). Molecular analysis of all E. scottii isolates and of individuals found in the infected grasses identified them as mating type B. No individuals of mating type A have been observed in this population so far. This explains the lack of fertilized stromata. No genes for the biosynthesis of ergot alkaloids and lolines could be detected. As the presence of these marker genes is highly correlated to alkaloid production in planta (Charlton et al. 2014), E. scottii is unlikely to produce either of these alkaloids. Only two genes involved in the biosynthesis of indole-diterpenes could be detected (idtQ and idfF): idtG, the first gene involved in the biosynthesis pathway (Schardl et al. 2013), is missing which makes it unlikely that E. scottii is able to produce any indole-diterpenes. For peramine, the marker for the T2 domain is missing. Preliminary results of the draft genome analysis showed a deletion of 2130 bp, which includes the T2 domain but also the M domain and the C-terminal subunit of the A2 domain. Due to this, neither the selection and activation of the arginine substrate, which requires the A2 C-subunit, nor the tethering or methylation of the activated arginyl residue, which requires the T2 and M domains, can be performed by the encoded protein (Tanaka et al. 2005). Based on this, the fungus described here is most likely not able to produce peramine or any other pyrrolopyrazine product.
In addition to horizontal transmission, most Epichloë endophytes can transmit vertically via the seeds of their hosts. DNA extraction from seeds of infected M. uniflora individuals, followed by PCR and sequencing showed the presence of E. scottii. This suggests that E. scottii is capable of vertical transmission. Further investigations, including microscopy of seeds of E. scottii infected plants is necessary to verify these findings.
The original host of E. scottii strain DSM111774 was brought to the greenhouse where it developed several daughter ramets, which subsequently all developed stromata. This shows that the endophyte is capable of transmission via rhizomes. For rhizomatous Festuca ruba, clonal growth via rhizomes can reach up to 220 m (Harberd 1961). Sample sites A and E were only 15 m apart, which makes it possible that E. scottii strains DSM111774 and DSM111775 originated from the same host individual.
This study showed that E. scottii can be transmitted vertically via clonal growth of its host and seeds. It further develops stromata in its host plants, indicating horizontal transmission of the fungus. However due to the lack of the opposite mating type development of ascospores and sexual life-cycle were not observed. Collectively, this evidence suggests E. scottii is a pleiotropic symbiont (Schardl et al. 1997), which can be classified as a Type II endophyte according to Clay and Schardl (2002).

CONCLUSION
In this study we described a new, haploid and stroma forming species within the genus Epichloë, Epichloë scottii sp. nov. which was isolated from Melica uniflora growing in Bad Harzburg, Germany. Phylogenic analysis revealed E. scottii as the unknown ancestor species of the hybrid E. disjuncta and placed it as an early branching species within the genus Epichloë. We further released a telomere-to-telomere de novo assembly of all seven chromosomes and the mitogenome of E. scottii.

Appendix A
See Fig. 8.

Table 2 Alkaloid profile and mating type of Epichloë scottii
Genes for ergot alkaloid, indole-diterpene and loline biosynthesis are displayed in order of their involvement in the corresponding biosynthesis pathway. For peramine, which is encoded by a single gene, gene fragments are ordered from the 5′ to 3′ end of each gene part Alkaloid/mating type Gene E. scottii DSM112488  8 Bayesian inference of the phylogenetic relationship of tubB sequences of the taxa among Epichloë using K80 + I + G as nucleotide substitution model. Numbers above nodes are estimates of a posteriori probability (BIpp, ≥ 0.9). The strains studied here are in Bold. The tree was rooted as midpoint. Host species are provided after each strain