Discovery of a new species of the Hypoxylon rubiginosum complex from Iran and antagonistic activities of Hypoxylon spp. against the Ash Dieback pathogen, Hymenoscyphus fraxineus, in dual culture

Abstract During a survey of xylarialean fungi in Northern Iran, several specimens that showed affinities to the Hypoxylon rubiginosum complex were collected and cultured. A comparison of their morphological characters, combined with a chemotaxonomic study based on high performance liquid chromatography, coupled with diode array detection and mass spectrometry (HPLC-DAD/MS) and a multi-locus phylogeny based on ITS, LSU, rbp2 and tub2 DNA sequences, revealed a new species here described as Hypoxylon guilanense. In addition, Hypoxylon rubiginosumsensu stricto was also encountered. Concurrently, an endophytic isolate of the latter species showed strong antagonistic activities against the Ash Dieback pathogen, Hymenoscyphus fraxineus, in a dual culture assay in our laboratory. Therefore, we decided to test the new Iranian fungi for antagonistic activities against the pathogen, along with several cultures of other Hypoxylon species that are related to H. rubiginosum. Our results suggest that the antagonistic effects of Hypoxylon spp. against Hym. fraxineus are widespread and that they are due to the production of antifungal phomopsidin derivatives in the presence of the pathogen.


Introduction
Hypoxylon Bull., 1791 is one of the largest genera of the Xylariales and comprises more than 200 species, which are mainly associated with angiosperm trees as saprotrophs and endophytes and are predominant in all forest ecosystems of the world (Daranagama et al. 2018;Helaly et al. 2018).
It traditionally belonged to the Xylariaceae until a recent phylogenetic study has resulted in a re-arrangement of the genera of stromatic Xylariales and the resurrection of the family Hypoxylaceae (Wendt et al. 2018). In this study and during the followup work by Lambert et al. (2019), genera like Hypomontagnella and Pyrenopolyporus were segregated from Hypoxylon, but the genus remained paraphyletic, indicating that further taxonomic segregation will eventually become necessary.
While the type species of Hypoxylon, H. fragiforme, belongs to a relatively small clade in the phylogeny of Wendt et al. (2018), the largest clades were comprised of the species of the "Hypoxylon rubiginosum complex" sensu Ju and Rogers (1996). Of these species, many had been lumped in H. rubiginosum, according to the broad concept established in the first monograph of the genus by Miller (1961). Miller's concept was mainly based on teleomorph morphology. In their revision of Hypoxylon, Ju and Rogers (1996) later recognised that anamorph characters, stromatal pigments and the micromorphology of ascospores and asci (in particular the apical apparatus) constitute valuable diagnostic characters. Modern concepts of the genus combine holomorph morphology with molecular phylogenetic data (Hsieh et al. 2005;. Moreover, secondary metabolite profiles generated by high performance liquid chromatography coupled to diode array detection and mass spectrometry (HPLC-DAD/MS) not only proved highly useful for segregation of species, but even led to the discovery of numerous novel natural products with prominent biological activities (see overviews by Hellwig 2005a andHelaly et al. 2018).
Hypoxylon rubiginosum and related taxa have been studied rather well on their stromatal secondary metabolites and, in many cases, morphologically similar species may contain entirely different pigments and other compounds Fournier et al. 2010). Interestingly, several species of the H. rubiginosum complex are known to frequently colonise Fraxinus species in the temperate Northern hemisphere. In some cases (e.g. H. cercidicola, H. fraxinophilum and H. petriniae), stromata are even almost exclusively found on dead wood of ash trees. They have also been frequently reported as endophytes of the same trees where they produce their stromata (Petrini and Petrini 1985) and are widespread endophytes of other host plants on which their stromata do not even occur (U'Ren et al. 2016). Therefore, the modern concepts of the taxonomy of the Hypoxylaceae take this fact into account and are based on the One Fungus-One Name concept. Some species have even been recognised on the basis of their anamorphic traits (Pažoutová et al. 2013) or their life cycle has been elucidated, based on a polythetic approach, i.e., by comparison of morphological, chemotaxonomic and molecular data of ascospore-derived cultures with endophytic isolates (see Bills et al. 2012 for H. pulicicidum andKuhnert et al. 2014 for H. griseobrunneum).
The Ash Dieback disease caused by the introduced apothecial ascomycete Hym. fraxineus (Leotiomycetes) has become one of the greatest problems in European forestry and the majority of common ash trees have succumbed to the fungal pathogen. We have recently studied the secondary metabolism of Hym. fraxineus (previously also known under the synonyms, Hym. pseudoalbidus or Chalara fraxinea) and its non-pathogenic domestic relative, Hym. albidus, for secondary metabolite production (Halecker et al. 2014(Halecker et al. , 2018Surup et al. 2018a). In parallel, we have also isolated endophytic fungi from apparently resistant ash trees in order to find natural antagonists that may be able to combat the devastating disease. One of the best candidates was identified as H. rubiginosum and, as reported recently (Halecker et al. 2020), it was found to produce the anti-fungal beta-tubulin inhibitor phomopsidin in dual culture with virulent strains of the pathogen. This compound was first reported from a marine-derived fungus that was originally assigned to the genus Phomopsis (Kobayashi et al. 2003). However, it has since then been found in other, terrestrial strains of the same genus, which should now be referred to as Diaporthe (Chepkirui and Stadler 2017), a large genus of the order Diaporthales. Interestingly, phomopsidin derivatives have never been reported from cultures of Xylariales before Halecker et al. (2020) found the compound in dual antagonist assays in agar cultures as described above. Moreover, they do not constitute major detectable metabolites of H. rubiginosum in the culture media that were used to study the chemotaxonomy of the genus before (cf. Bitzer et al. 2008).
Concurrently, we were about to study the taxonomy of new collections of Hypoxylon species originating from Iran that also belong to the Hypoxylon rubiginosum complex. Since mycelial cultures of these fungi had just become available, it appeared practical to combine the description of their taxonomy with an evaluation of their antagonistic potential to combat Hym. fraxineus. We have also included a number of other Hypoxylon species that colonise Fraxinus in Europe. The current study therefore provides new evidence on both, the taxonomy and chemical ecology of Hypoxylon.

Sample sources
Samples were collected from Guilan and Mazandaran provinces (Northern Iran) during 2015-2017. Parts of corticated branches and trunks bearing Hypoxylaceae stro-mata were transferred to the laboratory. Details of the specimens used for morphological investigations are listed in the Taxonomy section under the respective descriptions. Specimens have been deposited in the fungarium of the Department of Plant Protection, Faculty of Agricultural Science, University of Guilan, Guilan, Iran (GUM). Living cultures have been deposited in MUCL (Louvain, Belgium).

Morphological characterisation
Microscopic characters of the teleomorph were observed in distilled water and 10% potassium hydroxide (KOH). Melzer's reagent was used for staining of the apical ascus apparatus. The numbers of perithecia, ascospores, asci, conidia and conidiophores that were measured for size in the descriptions are 10, 30, 10, 30 and 5, respectively. Specimens were cultured from single ascospore isolates, using 2 % malt extract agar (MEA). For examination of culture macro-morphology, the strains were grown on Difco Oatmeal Agar (OA), following the protocols by Ju and Rogers (1996). Pigment colours were determined as described in the latter monograph, with colour codes following Rayner (1970). Macrophotographs were obtained with a Keyence VHX-6000 microscope. Light microscopy with Nomarski differential interference contrast (DIC) was done using a Zeiss Axio Imager A1 compound microscope, equipped with a Zeiss Axiocam 506 colour digital camera. SEM of ascospores were recorded using a fieldemission scanning electron microscope (FE-SEM Merlin, Zeiss, Germany), in a similar fashion as reported previously (Kuhnert et al. 2017).

DNA extraction, PCR and sequencing
DNA extraction of fresh cultures and amplification of the ITS (nuc rDNA internal transcribed spacer region containing ITS1-5.8S-ITS2), LSU (5' 1200 bp of the large subunit nuc 28S rDNA), rpb2 (partial second largest subunit of the DNA-directed RNA polymerase II) and tub2 (partial β-tubulin) loci were performed as described by Wendt et al. (2018). Sequences were generated by an in-house Sanger capillary sequencing solution on campus. Sequences were processed with Geneious 7.1.9 (http:// www.geneious.com).

Molecular phylogenetic analyses
The newly generated sequences were aligned with selected sequences from Wendt et al. (2018) and a combined matrix of the four loci (ITS, LSU, rpb2 and tub2) was concatenated for phylogenetic analyses, with four species (Biscogniauxia nummularia, Graphostroma platystomum, Xylaria arbuscula and Xylaria hypoxylon) added as the outgroup. The GenBank accession numbers of sequences are listed in Table 1. Sequences were aligned with the server version of MAFFT (http://mafft.cbrc.jp/alignment/server/, Katoh et al. 2019), checked and refined using BioEdit v. 7.2.6 (Hall 1999). After exclusion of ambiguously aligned regions and long insertions, the final combined data matrix contained 4369 characters, i.e. 578 nucleotides of ITS, 1301 nucleotides of LSU, 1017 nucleotides of rpb2, and 1473 nucleotides of tub2.
Maximum Parsimony (MP) analyses were performed with PAUP v. 4.0a165 (Swofford 2002). All molecular characters were unordered and given equal weight; analyses were performed with gaps treated as missing data; the COLLAPSE command was set to MINBRLEN. MP analysis of the combined multilocus matrix was done using 1000 replicates of heuristic search with random addition of sequences and subsequent TBR branch swapping (MULTREES option in effect, steepest descent option not in effect). Bootstrap analyses with 1000 replicates were performed in the same way but using 10 rounds of random sequence addition and subsequent branch swapping during each bootstrap replicate.
Maximum Likelihood (ML) analyses were performed with RAxML (Stamatakis 2006) as implemented in raxmlGUI 1.3 (Silvestro and Michalak 2012), using the ML + rapid bootstrap setting and the GTRGAMMA substitution model with 1000 bootstrap replicates. The matrix was partitioned for the different gene regions. In the Results and Discussion, bootstrap values ≤ 70% are considered low, between 70-90% intermediate and ≥ 90% high.

HPLC profiling
Stromata of Hypoxylon specimens were extracted as described by Kuhnert et al. (2017) and subsequently analysed by high performance liquid chromatography, coupled with diode array and electrospray mass spectrometric detection (HPLC/DAD-ESIMS) instrument settings as described by Halecker et al. (2020). The resulting UV/Vis and mass spectra were compared with an internal database (cf. Bitzer et al. 2008), comprising standards of known Hypoxylaceae.

Dual culture experiments
Dual cultures of Hypoxylon spp. and Hym. fraxineus (STMA 18166) were co-incubated on barley-malt agar by inoculation at opposite sites on 9 cm Petri dishes (cf. Halecker et al. 2020) with Hym. fraxineus being inoculated one week prior the beginning of the dual culturing due to its slow growth. Axenic cultures, containing only one fungus, were inoculated in parallel as a control group. Growth was documented and observed weekly after incubation in the dark for a maximum of four weeks. Thereafter, the agar plates were extracted with acetone following the method described by Halecker et al. (2020), except that the entire agar plate was extracted instead of the fungal interaction zone. Table 1. Isolates and accession numbers of sequences used in the phylogenetic analyses. Type specimens are labelled with HT (holotype) ET (epitype) and PT (paratype). Isolates/sequences in bold were isolated/sequenced in the present study.

Phylogenetic analyses
Of the 4369 nucleotide characters of the combined matrix, 1618 are parsimony informative (298 of ITS, 156 of LSU, 487 of rpb2 and 677 of tub2). Fig. 1 shows a phylogram of the best ML tree (lnL = −63870.651550) obtained by RAxML. Maximum parsimony analyses revealed one MP tree comprising 14,014 steps (data not shown). All major groups and deeper, highly supported nodes were consistent between the ML and MP analyses, but topologies of deeper unsupported nodes differed in the MP tree; as these differences are not relevant within the context of our new species, they are not further considered here. The phylogenies reveal a paraphyly of Hypoxylon, with the genera Annulohypoxylon, Daldinia, Entonaema, Jackrogersella, Hypomontagnella, Pyrenopolyporus, Rhopalostroma, Ruwenzoria and Thamnomyces embedded within the former. All of the latter genera appeared monophyletic except for Daldinia (Fig. 1). All of our new Iranian species and records described below are contained within the highly supported Hypoxylon clade H5. The new species H. guilanense clustered together with H. texense with 100% BS support, while sequences of two additional strains (Hypoxylon aff. rubiginosum MUCL 57724) and (Hypoxylon aff. rubiginosum MUCL 57725) formed a highly supported (100% BS support) clade that is the sister group of H. rubiginosum (Fig. 1). The sequences of the Iranian collection of H. rubiginosum (MUCL 57727) are almost identical to those of the ex-epitype culture (MUCL 52887) and they clustered together with maximum support. As in previous studies, the position of H. griseobrunneum and H. trugodes could not be resolved within the family. The remaining clades are in accordance with previous results of Wendt et al. (2018). Etymology. Guilanense, refers to its origin in Guilan province, Iran. Teleomorph. Stromata superficial, hemispherical to pulvinate, up to 2 cm long × 0.1-0.7 cm wide, with conspicuous perithecial mounds, surface Sienna (8), Umber (9) to Buff (45); Scarlet (5) to Orange (7) granules beneath the surface and between the perithecia, with Orange (7) KOH-extractable pigments. Perithecia spherical to obo-  void, 0.33-0.66 high × 0.3-0.55 mm wide. Ostioles umbilicate, inconspicuous. Asci not seen. Ascospores smooth, unicellular, brown to dark brown, ellipsoid, inequilateral with narrowly rounded ends, 12-15 × 5-6 µm, with straight germ slit spore-length on convex side; perispore dehiscent in 10% KOH, conspicuous coil-like ornamentation in SEM; epispore smooth.
Cultures and anamorph. Colonies on OA covering a 9 cm Petri dish in 4 wk, at first white, becoming Buff (45), cottony, slightly zonate with diffuse margins; finally, becoming Honey (64). Anamorph not produced in culture.
Secondary metabolites. Orsellinic acid, rubiginosin A and an unknown isomer thereof, as well as mitorubrinol acetate as prevailing stromatal components; cultures produce yet unidentified compounds on barley-malt agar.
Notes. The description of this taxon is based on a single specimen, which shows the salient features of the teleomorph and can be discriminated easily from all previously described species of the H. rubiginosum complex. The stromata of the holotype specimen differ from H. texense (i.e. the closest relative in the phylogeny), in having stromata with hemispherical to pulvinate shape, Orange (7) KOH-extractable pigments and larger ascospores [12-15 × 5-6 vs. 9.1-10.8 (-11.5) × (4.0-) 4.5-5.4 (-5.7) µm with straight germ slit.
Hypoxylon guilanense can also be easily differentiated from H. rubiginosum sensu stricto and H. petriniae in the peculiar stromatal shape and it also has larger ascospores. H. cercidicola differs from H. guilanense in having erumpent stromata with discoid shape and smaller ascospores [(9-) 9.5-12 × 5-6 µm)] with straight to slightly sigmoid germ slit. Table 2 compares morphological characters of some other taxa that may be confused with H. guilanense.
Secondary metabolites. Rubiginosin A and an unknown compound of the mitorubrin / rubiginosin azaphilone family prevalent; cultures produce phomopsidin and unidentified compounds on barley-malt agar.
Specimens examined. Iran, Guilan Province, Siahkal County, Deilaman forest, 36°57'25"N, 49°51'54"E, 1100 m elev., on fallen branch of Quercus castaneifolia,    ) and may occur in subtropical areas, such as Florida, USA (Ju and Rogers 1996). Most of the characters of the Iranian specimens are in accordance with previous descriptions , aside from insignificant variations in the size of ascospores.

Additional potentially new species of the H. rubiginosum complex
Below, we describe two collections that may eventually be recognised to represent new species. They appear phylogenetically different from the type specimen, as well as from Iranian records of H. rubiginosum, but share salient features with the latter species. It is explained in the Notes why we hesitate to describe them as new taxa in this complicated species complex.
Cultures and anamorph. Colonies on OA covering a 9 cm Petri dish in 3 wk, at first white, becoming whitish, cottony, azonate with entire margins; remaining mainly uncoloured with Pale Luteous tinges. Anamorph not produced in culture.
Due to the availability of well-studied strains of H. rubiginosum, H. perforatum and H. petriniae in public culture collections, a pre-screening was conducted to confirm production of 5 and 6 (with 13, 7 and 4 strains each, respectively (cf. Table 3, Fig.  11H). Out of these 24 strains, 16 emerged as producers of compound 5 and partially 6 (12 strains). Compound 6 was not detected in the absence of 5.  Fig. 7, showing the dual cultures after 1-4 weeks of incubation. The chemical structures are shown in Fig. 8 and selected chromatographic data are depicted in Figs 10, 11. Strikingly, during evaluation and comparison of the HPLC UV/Vis chromatograms with our internal database, the mitorubrin derivatives 2, 4 and 7 were identified by direct comparison of chromatograms derived from extracts of stromata and cultures of the ex-type strain and the holotype of H. texense Figs 7 I-L, 11B, D).

Discussion
The present study dealt with the identification of Hypoxylon species from Northern Iran based on morphological, chemotaxonomic and phylogenetic data, focusing on  (5), 10-hydroxyphomopsidin (6), orthosporin (9), daldinone B (10), 1,8-dimethoxynaphthalene (11), daldinin F (12), 5-methylmellein (13), viridiol (14) and an unidentifiable compound (UC 6) after comparison of data with internal databases. The UV signal of UC 6 was enhanced in the dual culture extract. the H. rubiginosum complex. The specimens encountered appeared morphologically and chemotaxonomically related to H. rubiginosum sensu stricto, as revealed from their morphology and secondary metabolite profiles. While the majority of specimens were assigned to typical H. rubiginosum, we have encountered a new taxon that significantly deviates from the complex in both stromatal and ascospore morphology and appears most closely related to a species that was so far only reported from the southern USA ). Furthermore, we found two specimens that slightly differed in one or two characters from typical H. rubiginosum and also showed deviating positions in the phylogenetic trees, but are so far only known from single collections. Attempts should be made to encounter additional specimens of these fungi, which may eventually lead to their recognition as new species. The recent study on intragenomic polymorphisms in Hypoxylaceae has suggested that molecular data alone may be misleading in this family and new taxa should be based on multiple records sharing the same genotypic and phenotypic features ). Hsieh at al. (2005) have already established that protein-coding genes provide a better resolution in the Hypoxylaceae than ITS and finally even omitted this locus from the phylogeny and rather decided to focus on tub2 and alpha-actin sequences. Kuhnert et al. (2014) also found tub2 to be more suitable than ITS in their phylogeny, based on material from the Caribbean. Our phylogenetic analyses confirmed previous results (Wendt et al. 2018;Lambert et al. 2019;Sir et al. 2019), suggesting that the genus Hypoxylon appears paraphyletic in Hypoxylaceae, with a relatively small clade comprising the type species H. fragiforme as "core group" to which members of the Hypoxylon rubiginosum complex form a sister clade. The genus will eventually need to be further subdivided, but molecular data for the majority of known species remain incomplete and such a task should only commence as the phylogenetic data matrix has increased. Our study further contributed to this monumental task by adding some data on representatives from the Middle East, a geographic area that has certainly not been as well explored as Western Europe and other parts of the world.
A main objective of this work was to assess the antagonistic potential of the newly isolated cultures and some strains of related species against an important pathogen, following the recent discovery that an endophytic isolate of H. rubiginosum from a resistant ash tree inhibited the growth of the alien pathogen, Hym. fraxineus (Halecker et al. 2020). Assessment of axenic cultures of the Hypoxylon species in a single medium (barley-malt) led to the detection of phomopsidin in one out of five strains of H. petriniae, two out of seven strains of H. perforatum and ten out of 13 strains of H. rubiginosum. The stromata of these three taxa have been frequently reported from Fraxinus and it is plausible that they all occur as endophytes in this host and only form the stromata on dead host tissues. On the other hand, phomopsidin was not detected in other related, but apparently rare species like H. texense, H. crocopeplum and H. carneum. Only the Table 4. Identified secondary metabolites in dual culture (barley-malt medium with Hymenoscyphus fraxineus) of the surveyed strains listed in Table 3. Identified compounds: 5: phomopsidin; 6: 10-hydroxyphomopsidin; 8: rickiol A; 9: orthosporin; 10: daldinone B; 11: 1,8-dimethoxynaphtahlene; 13: 5-methyl-mellein. Identified stromal azaphilone groups detected in culture: MI = Mitorubrin type; NA = Naphthalene type; DA = Daldinin type. For chemical structures, see Fig. 8 two latter species, however, were represented in our study by cultures that were isolated from stromata growing on Fraxinus wood. In addition, our results need to be further validated because we cannot exclude that some of the strains, which have been kept in culture collections for many years, may have degenerated. In any case, our results suggest that phomopsidin is not a specific marker for the species complex or for H. rubiginosum sensu stricto. As the compound is preferentially observed in dual cultures, its biosynthesis may be under control of epigenetic effectors. Therefore, in the future, it would be useful to evaluate a broader range of ascospore-derived cultures of Hypoxylon for their potential as biocontrol agents against the ash dieback pathogen and to define the genetic mechanisms encoding phomopsidin biosynthesis. Last but not least, the current study also revealed some interesting aspects for potential follow-up projects. For instance, the examination of H. fuscum (a species that has never been isolated from Fraxinus, but is actually associated with Corylus and other Betulaceae) in the antagonism assay, revealed the production of several hitherto unknown compounds whose production was significantly enhanced in the presence of Hym. fraxineus. This observation suggests that it will be worthwhile to further study the secondary metabolism of Hypoxylon species in other scenarios using the dual culture approach. The first step would be to scale-up the production of the unknown molecules, isolating enough for structure elucidation and biological studies. This should not be expected to be a trivial task, but it appears doable using the methodology that is presently available.
The production of known and yet unidentified azaphilones (i.e. a compound class that is normally found in high concentrations in the stromata of various Hypoxylaceae, but was rarely observed in their mycelial cultures) in H. rubiginosum and allies, is another interesting observation relating to the differential expression of biosynthetic genes encoding secondary metabolites. It should be rewarding to evaluate the regulation mechanisms that lead to the production of the pigments, aided by genomic and transcriptomic studies.