Introduction

Ash dieback disease is threatening the common ash (Fraxineus excelsior) in central Europe. The invasive ascomycete Hymenoscyphus fraxineus is causing this disease leading to decreased growth, reduced wood quality and declining ash trees. Transmission occurs by wind-borne ascospores, first colonizing the leaves and at a later stage invading the petioles and shoots (Timmermann et al. 2011; Cleary et al. 2013; Fones et al. 2016; Haňáčková et al. 2017) but also via root system (Fones et al. 2016; Baxter et al. 2023). Subsequent expansion in tissue can lead to necrosis of cambium and bark, shoot death or stem foot necrosis (Kowalski and Holdenrieder 2008; Schumacher et al. 2010; Peters et al. 2023). The fungi remain in leaves, and in autumn it lives as a saprotroph degrading foliage and forming new spores for the infection cycle in the new year (Fones et al. 2016). In contrast, in its native environment East Asia the pathogen seems to colonize the local ash species Fraxinus mandshurica without any symptoms (Zhao et al. 2012). In Europe, the Mandshurian ash shows only mild symptoms (Drenkhan and Hanso 2010; Nielsen et al. 2017).

It remains unclear why some ash tree species and even different genotypes of the same species as shown for F. excelsior, react more susceptible and some more resistant to H. fraxineus (Stener 2013; Cleary et al. 2014). Therefore, previous studies analyzed the transcriptional response of F. excelsior trees infected with H. fraxineus. Initial evidence has been obtained from transcriptome analyses that molecular markers can be used to identify individual trees with a low level of susceptibility to ash dieback (Harper et al. 2016; Sollars et al. 2017). Also, numerous single nucleotide polymorphisms (SNPs) have been identified in the ash genome that are associated with tree susceptibility (Stocks et al., 2019). Another study focusing on transcriptional responses of F. excelsior trees to H. fraxineus infection showed limited differences in gene expression patterns between resistant and susceptible F. excelsior clones (Sahraei et al. 2020). However, a set of candidate genes likely involved in the containment of the necrosis had been identified. The mechanisms of resistance are not yet well understood.

On the other hand, also little is known on the pathogen side. To gain more insights into the pathogen’s genetic repertoire and pathogenic mechanisms, the genome and transcriptome of H. fraxineus has been compared with its close relative H. albidus which is native in Europe and non-pathogenic (Stenlid et al., 2017). It was shown that the two genomes are very similar and differences in gene expression may be responsible for the difference in interaction with the host. Especially, a pectin-degrading enzyme was identified in the H. fraxineus compared to H. albidus transcriptome that could be responsible for the degradation of cell walls and thus lead to a stronger host response. In addition, two small, secreted proteins were identified that were not found in H. albidus, one of which, cerato-platanin, could act as pathogen-associated molecular patterns (PAMPs).

In the genome of H. fraxineus a viridiol biosynthetic gene cluster had been identified (Elfstrand et al. 2021). Despite that viridiol showed necrotic activity on ash seedlings (Andersson et al. 2010; Cleary et al. 2014), its role in virulence is not clarified as it is also encoded and produced by non-pathogenic H. albidus (Junker et al. 2014; Elfstrand et al. 2021). Furthermore, the concentration of viridiol was also shown not to correlate with virulence (Junker et al. 2014). Another compound shown to be encoded only by H. fraxineus is hymenosetin (Elfstrand et al. 2021). It belongs to 3-decalinoyltetramic acid family, known to have broad spectrum antimicrobial and cytotoxic effects (Halecker et al. 2014). As it was isolated from crude extracts of virulent H. fraxineus strains, it is suggested as possible pathogenicity factor which could also be involved in defending competing fungi and bacteria.

As summarized above, several studies on transcriptional response of F. excelsior trees infected with H. fraxineus had been performed, and also H. fraxineus genome and gene expression in F. excelsior were analyzed and compared to H. albidus. Nevertheless, the underlying molecular processes in the pathogen–host interaction are not yet understood in detail. But it is supposed that the fungus expresses different genes—depending on the ash species. It was also shown that the growth rates of H. fraxineus on culture media supplemented with leaf tissue are largely consistent with the susceptibility of ash species to the pathogen (Carrari et al. 2015; Harper et al. 2016). However, the transcriptome of H. fraxineus strains during growth on different ash species has not been studied so far. This research is needed to gain insights into substrate and species-specific regulated genes, which contribute significantly to the differences in disease progression.

This study investigates the differential gene expression via RNA-Seq of H. fraxineus grown on media without and with leave supplements of F. excelsior and F. mandshurica. The aim was to identify species and substrate-specific regulated genes including virulence factors of the pathogen, and their implications on colonization, invasion and for the exploitation of nutrients.

We hypothesize that susceptible F. excelsior provides better growth conditions to H. fraxineus due to beneficial plant metabolite composition and that better growth conditions lead to a higher expression of genes for pathogenicity. To verify or disprove the hypothesis, changes in gene expression profile of H. fraxineus grown on media supplemented with plant material of susceptible (F. excelsior), more resistant (F. mandshurica) ash species and a control media were investigated.

Material and methods

Plant material and fungal culture

H. fraxineus isolates 1431 (2013–3-B1-21) and 1511 (2013–3-B15-6), provided from Northwest German Forest Research Institute (NW-FVA), Göttingen, Germany, were cultured on malt-yeast-peptone (MYP) agar. For RNA-Seq culture a liquid MYP media was supplemented with 36 g/l leave material of F. excelsior (EX) and F. mandshurica (MA) and MYP media without leave supplements was used as control media. Leave material of F. excelsior and F. mandshurica had been collected in August 2020 (isolate 1511) and 2021 (isolate 1431) from trees in the botanical garden at the University of Hohenheim, Stuttgart.

H. fraxineus isolates 1431 and 1511 were incubated at 20 °C for 28 and 24 days, respectively. 10 ml culture of isolate 1431 and 2 ml of isolate 1511 vortexed liquid culture had been centrifuged (10,000 × g for 10 min), and pellets were stored at − 80 °C.

RNA templates

The pelleted fungal mycelia were homogenized in FastPrep-24 5G (MP Biomedicals, Eschwege, Deutschland) with 8 m/sec, and total RNA was extracted by solid phase extraction of RNA using InviTrap Spin Universal RNA Mini Kit (Invitek, Berlin, Germany). The purified pellet was dissolved in 50 μl of double-distilled water and stored at -80 °C. The concentration of the total RNA was measured using Qubit® fluorometer with RNA HR Assay Kit (Invitrogen, Carlsbad, Germany) according to manufacturer’s instructions. In addition, integrity of RNA had been assessed in agarose gel (1%), stained with GelRed (Genaxxon bioscience, Ulm, Germany), and visualized with a UV transilluminator.

RNA sequencing and automated downstream analysis

Three replicates were selected of each EX, MA, and MYP media cultures of H. fraxineus isolate 1511 and isolate 1431. After quality control, H. fraxineus isolate 1511 replicates were pooled, and 1431 replicates were used as triplicate for transcriptome studies. RNA sequencing and bioinformatic analysis was conducted by Novogene (UK) Company Limited, Cambridge, UK. Construction of mRNA libraries was performed by poly(A) enrichment. Illumina sequencing was carried out on NovaSeq 6000 with paired-end reads and 150 nt length (6 G raw data/ sample). The raw reads had been filtered to remove (a) reads containing adapters, (b) reads containing over 10% not determined bases (N > 10%) or (c) reads with a q score of over 50% bases ≤ 5. The quality-passed reads were used for the subsequent analyses. The mapping on reference genome H. fraxineus nf4 (GenBank assembly accession: GCA_911649665.1) (Elfstrand et al. 2021) was done with HISAT2 (Kim et al. 2019). Total mapping rate has been calculated as follows: (mapped reads)/(total reads)*100. The assembly was done with StringTie (Pertea et al. 2015) and quantification of gene expression with featureCounts. DESeq2 (Love et al. 2014) was used for differential gene expression analysis with biological replicates and edgeR (Robinson et al. 2009) if no biological replicates had been included. For both methods a screening threshold of |log2(FoldChange)|> = 1 and adjusted p-value (padj) <  = 0.05 had been applied to all genes in the compare groups.

The original gene annotation (Elfstrand et al. 2021) (https://zenodo.org/record/4355824) had been supplemented by UniProtKB Swiss-Prot database analysis (The Uniprot Consortium 2023) and Pfam protein families and domains (Mistry et al. 2021).

Identification of genes involved in toxin production and analysis of upregulated genes of interest

Protein sequences of the known biosynthetic gene clusters for gliotoxin in Aspergillus fumigatus (AY838877.1), hypothemycin in Hypomyces subiculosus (EU520418.1), aflatoxin in Aspergillus parasiticus (AY371490.1), and depudecin in Alternaria brassicicola (FJ977165.1) were used to identify putative homologue proteins potentially involved in toxin production in H. fraxineus. BLASTp algorithm was used to identify sequences showing a high structure similarity indicated by high e-values. For all H. fraxineus genes of interest, a new protein BLAST search of UniProtKB Swiss-Prot database was performed (June 2023). Additionally, Pfam protein families and domains had been verified with InterProScan (Jones et al. 2014). The protein sequences used for all analyses were obtained from Elfstrand et al. 2021 (https://zenodo.org/record/4355824).

Putative effector genes

Differentially expressed genes (DEGs) had been analyzed in silico with Phobius (Käll et al. 2007) for transmembrane topology and signal peptides to identify putative effectors.

The comparative transcriptome analysis allows the identification of substrate-dependent differentially expressed genes of H. fraxineus to gain insights into the different courses of infection and identified key genes.

Results

Sequencing and mapping

RNA sequencing resulted in 89,081,550 total reads for isolate 1511 (pooled samples) grown on malt-yeast-peptone media (MYP) as control. For media supplemented with plant material from F. excelsior (EX) 93,701,440 and F. mandshurica (MA) 87,261,106 total reads had been obtained. Isolate 1431, which had been analyzed in triplicate, resulted in minimum of 39,856,930 total reads for MYP, 36,384,148 for EX and 40,040,864 for MA (Table 1). The percentage of total clean reads assigned to the H. fraxineus genome is higher than 92% for all samples.

Table 1 Summary of the total illumina sequence reads (clean) and amount of data obtained from Hymenoscyphus fraxineus isolate 1511 and 1431 grown on malt-yeast-peptone media (MYP) as control and MYP media supplemented with plant material from Fraxinus excelsior and Fraxinus mandshurica
Full size table

Differential gene expression (DEG) analysis

DEG analysis was conducted to detect H. fraxineus transcriptome changes between growth on media supplemented with leaves from susceptible and resistant ash species to identify host specifically expressed genes. The comparison to control media allows to assess the tropism and influence of leave material.

For strain 1431, during growth on F. excelsior supplemented media compared to control media, a total of 1,777 genes had been differentially expressed, including 916 upregulated and 861 downregulated genes. Compared to F. mandshurica supplemented media, there are 126 DEGs, of which 95 genes are upregulated and 31 downregulated. Between growth on F. mandshurica and control media, there are 1,783 DEGs, with 841 up- and 942 downregulated genes.

For strain 1511, there are 1,837 DEGs between growth on F. excelsior supplemented media and control media, of which 875 are upregulated and 962 downregulated. There are 1,273 DEGs between F. excelsior and F. mandshurica supplemented media, with 483 upregulated and 790 downregulated genes. Comparing transcripts from growth in F. mandshurica supplemented media and control media, results in 812 DEGs, of which 424 are upregulated and 388 downregulated. The number of predicted DEGs between different H. fraxineus comparison groups is shown in Table 2. To test the hypotheses in this study, only the upregulated genes of F. excelsior comparison groups are considered.

Table 2 Number of predicted DEGs from H. fraxineus isolate 1511 and 1431 grown on maltose-yeast-peptone media (MYP) and MYP media supplemented with plant material from Fraxinus excelsior (EX) and Fraxinus mandshurica (MA)
Full size table

Upregulated genes of H. fraxineus grown on F. excelsior supplemented media compared to F. mandshurica supplemented media and MYP control media

We hypothesized that H. fraxineus colonization on susceptible F. excelsior may be associated with better growth due to beneficial plant metabolite composition. This will promote pathogenicity and will be reflected by changes in gene expression profile compared to colonization on resistant F. mandshurica. Therefore, differential gene expression of H. fraxineus had been compared between growth on F. excelsior supplemented media (EX) and control media (MYP) to assess tropism and influence of leave material. The comparison between the both supplemented media (EX and MA) provided the host-specific DEGs. The comparison was made for both strains 1431 and 1511 (Table 3; Fig. 1). Additionally, the shared upregulated genes between EX vs. MYP and EX vs. MA had been assessed for both strains.

Table 3 Upregulated genes of Hymenoscyphus fraxineus isolate 1511 and 1431 grown in maltose-yeast-peptone media (MYP) supplemented with plant material from Fraxinus excelsior (EX) compared to MYP media and MYP media supplemented with plant material from Fraxinus mandshurica (MA). The mean of log2FoldChange and standard deviation (± SD) of upregulated genes is indicated
Full size table
Fig. 1
figure 1

Upregulated genes of Hymenoscyphus fraxineus isolates 1511 and 1431 grown in maltose-yeast-peptone media (MYP) supplemented with plant material from Fraxinus excelsior (EX) compared to MYP media and MYP media supplemented with plant material from Fraxinus mandshurica (MA)

Full size image

A total of 916 genes in H. fraxineus 1431 were found to be significantly upregulated in F. excelsior media compared to MYP control media (mean log2FoldChange of 1.76, ± SD 0.83) and 95 genes in F. excelsior media compared to F. mandshurica media (mean log2FoldChange of 2.52, ± SD 1.13). The comparison of F. excelsior media to MYP control media and F. mandshurica media resulted in 54 intersecting genes (Online Resource 1), whereof 22 genes (40.74%) are annotated as proteins of unknown function.

For strain 1511, there are 875 genes upregulated in EX compared to MYP (mean log2FoldChange of 2.04; ± SD 1.20) and 483 genes in EX compared to MA (mean log2FoldChange of 2.31; ± SD 1.62). There are 244 intersecting genes between both in EX upregulated gene sets (Online Resource 2). Among them are many genes (136 genes, 55.74%) which are encoding for proteins of unknown function.

Analysis of those two shared upregulated gene datasets (n = 54 and n = 244) identified no commonly upregulated genes between both strains. Probably the limited number of predicted upregulated genes (n = 95) for comparison group F. excelsior (EX) vs. F. mandshurica (MA) and thus the small set of shared genes between treatments of strain 1431 prevent the identification of shared genes between strains.

Oxidative phosphorylation

Among the upregulated genes in strain 1431, there are many predicted to belonging to mitochondrial oxidative phosphorylation enzymes, like NADH-ubiquinone oxidoreductase (complex I), ubiquinol-cytochrome c oxidoreductase or cytochrome bc1 complex (complex III), cytochrome c oxidase (complex IV), and ATP synthase (complex V). All the upregulated genes seem to be in a gene cluster encoded by mitochondrial DNA.

Many of the upregulated genes are predicted as subunits of the NADH-ubiquinone oxidoreductase respiratory complex like chain 1 (HYFRA_G00013582), chain 2 (HYFRA_G00013648), chain 4 (HYFRA_G00013619), chain 4L (HYFRA_G00013634), and chain 5 (HYFRA_G00013627). This complex I is the largest of the respiratory chain and the major entry point for electrons. Of complex III, the only protein located on mitochondrial DNA, cytochrome b (cob, HYFRA_G00013579) is predicted as upregulated. The genes cox1, cox2, and cox3 are the largest subunits of cytochrome c oxidase (complex IV) and the ones which are mitochondrial encoded. In the upregulated gene set, there is a predicted subunit 1 (HYFRA_G00013653) and two subunits 2, one containing a periplasmic domain (Pfam: PF00116, HYFRA_G00013638), and one containing a transmembrane domain (Pfam: PF02790, HYFRA_G00013639). Furthermore, two predicted subunits of ATP synthase (complex V) of FO part, ATP6/ subunit a (HYFRA_G00013644) and ATP9/ subunit c (HYFRA_G00013622, HYFRA_G00013581), which are encoded by the mitochondrial genome are upregulated.

In addition, in the same gene cluster as the mitochondrial oxidative phosphorylation enzymes, there are four upregulated genes, which are predicted as intron-encoded endonucleases (HYFRA_G00013577, HYFRA_G00013578, HYFRA_G00013628, HYFRA_G00013632). Two of those genes (HYFRA_G00013577 and HYFRA_G00013578) have a GIY-YIG catalytic domain (Pfam: PF01541), whereas HYFRA_G00013577 additionally shows a LAGLIDADG endonuclease (Pfam: PF00961) and cytochrome b (C-terminal)/b6/petD domain (Pfam: PF00032). The LAGLIDADG endonuclease domain is also encoded at three further genes in this cluster (HYFRA_G00013617, HYFRA_G00013629, HYFRA_G00013645).

‘GIY-YIG’ and ‘LAGLIDADG’ families belong to group I homing endonucleases (reviewed by Stoddard 2014). They are highly specific DNA cleaving enzymes catalyzing the transfer of their own sequence and of the mobile element in which they are located, often self-splicing elements as introns and inteins.

However, there is no evidence that these genes encoding for enzymes of oxidative phosphorylation and homing endonucleases are also upregulated in strain 1511.

Toxin production

There are several upregulated genes in both strains showing similarity to genes involved in toxin production of other Ascomycota. The deduced amino acid sequence of one upregulated gene identified in strain 1511 (HYFRA_G00004959, 322 aa) shows an identity of 45.7% to thioredoxin reductase GliT of Neosartorya fumigata (Aspergillus fumigatus) (AAW03299.1, 334 aa) and carries a pyridine nucleotide-disulfide oxidoreductase domain (Pfam: PF07992). Another upregulated gene (HYFRA_G00010095, 587 aa) in strain 1431, has a major facilitator superfamily domain (Pfam: PF07690) and shows 35% identity to major facilitator gliotoxin efflux transporter GliA of N. fumigata (AAW03302.1, 542 aa) but also to efflux pump encoding gene AflT of A. parasiticus (36.5% identity).

In A. fumigatus gliT is part of a gene cluster for the biosynthesis of the epipolythiodioxopiperazine (ETP) gliotoxin and catalyzes the closure of disulfide bond of dithiol gliotoxin to final gliotoxin (Scharf et al. 2010; Schrettl et al. 2010). It also has an important role in self-protection as its deletion leads to disruption of gliotoxin release and an increased sensitivity to exogenous gliotoxin. Like the putative homologue in H. fraxineus it has a pyridine nucleotide-disulfide oxidoreductase domain (Pfam: PF07992). Gliotoxin is subsequently effluxed by GliA (Wang et al. 2014).

Gliotoxin is the is the best-known representative of the class of ETPs and is shown to have immunosuppressive features (reviewed by Ye et al. 2021). The complete gene cluster for biosynthesis of gliotoxin in Aspergillus fumigatus originally comprises 12 genes encoding for a two modular non-ribosomal peptide synthase (gliP), a glutathione S-transferase (gliG), and a hypothetical protein (gliK) shown to encode for a γ-glutamate cyclotransferase by Scharf et al. 2013, dipeptidase (gliJ), an 1-aminocyclopropane-1-carboxylic acid synthase (ACCS) (gliI), thioredoxin reductase (gliT), an O-methyl transferase (gliM), a methyl transferase (gliN), cytochrome P450 monooxygenases (gliC and gliF), major facilitator transporter (gliA), a zinc finger transcriptional regulator (gliZ) (Gardiner and Howlett 2005) as well as gliH, a unknown protein adjacent to the gliotoxin cluster (Schrettl et al. 2010). The detailed function of the proteins has already been reviewed elsewhere (Scharf et al. 2012, 2016; Dolan et al. 2015; Downes et al. 2023). For the proteins described to belong to the gliotoxin biosynthesis pathway putative homologs in H. fraxineus had been searched (Table 4). Based on similarity and domain structure, a putative homologue had been found for all proteins. For GliP, GliZ, GliI, GliC, GliG, and GliM the putative homologue protein hit showed low e-values (< 10–20) and/ or minimal identity (below 30%). However, the corresponding Pfam family/domain entry could be found in all potential homologous proteins, except for GliK of A. fumigatus (AAW03303.1). But GliK it shares the same Interpro homologous superfamily (Gamma-glutamyl cyclotransferase-like superfamily; IPR036568) like its putative homologue in H. fraxineus (HYFRA_T00006955). For GliH (EAL88826.2) of A. fumigatus no homologous protein can be identified (data not shown).

Table 4 BLASTP results of gliotoxin gene cluster proteins of Aspergillus fumigatus (AY838877.1) and predicted homologs in Hymenoscyphus fraxineus, incl. predicted function, length (aa), Pfam hits, e-values, and identities
Full size table

Since the description of the biosynthetic gene cluster in A. fumigatus (Gardiner and Howlett 2005), two more transcription factors have been reported. One is the transcription factor GipA (XP_747886.1), which has two C2H2 zinc finger binding domains (Schoberle et al. 2014). In H. fraxineus there is a putative homologue protein encoded by a protein of unknown function HYFRA_T00002564 (BLASTP identities = 96/196 (49%), e-value = 4e-49) also showing two zinc finger, C2H2 type domains (Pfam: PF00096).

The other transcription factor is RglT of A. fumigatus (B0XQ41.1) (Ries et al. 2020), which has a BLASTP hit at HYFRA_T00002891 (annotated as protein of unknown function) with an identity of 92/287 (32%) and an e-value of 9e-25. They have a common fungal Zn(2)-Cys(6) binuclear cluster domain (Pfam: PF00172). However, both genes are not upregulated in our strains grown in F. excelsior supplemented media compared to F. mandhurica supplemented or control media.

Among the upregulated genes, there are two for strain 1431 and three for 1511 which are similar to genes belonging to the gene clusters of the fungal metabolites aflatoxin (A. parasiticus) and hypothemycin (Hypomyces subiculosus). In strain 1431, one of these genes is HYFRA_G00005709 encoding a predicted cytochrome p450 domain (Pfam: PF00067). As best BLAST hit against UniProt KB SwissPro database it shows an amino acid identity of 37% (e-value 6.4e-36) to cytochrome P450 monooxygenase hpm1 of H. subiculosus (Nectria subiculosa) (B3FWR7.1), which is part of the gene cluster that mediates the biosynthesis of hypothemycin. However, it is shorter (170 aa) than Hpm1 (502 aa) and lacks a predicted transmembrane region. Compared to A. parasiticus aflatoxin cluster it shows 35% identity (e-value 3e-29) to averantin hydroxylase, a cytochrome P450 monooxygenase (AflG, AAS66008.1), and 35% identity (e-value 5e-29) to AflL (AAS66013.1), a versicolorin B desaturase. In A. parasiticus, AflG is catalyzing the conversion of averantin to 5'-hydroxyaverantin and aflL the conversion of versicolorin B to versicolorin A (Yu et al. 2004). Also, in comparison to AflG (495 aa) and AflL (500 aa) of A. parasiticus it is shorter and lacks a transmembrane region at N-terminus.

The second upregulated gene in strain 1431 (HYFRA_G00010188, 392 aa) encodes an O-methyltransferase domain (Pfam: PF00891), and compared against UniProt KB SwissPro database, it shows an 51% amino acid identity (e-value 5.5e-130) to O-methyltransferase PenC (A0A1B2CTA7, 398 aa) of Penicillium thymicola. Compared to the hypothemycin gene cluster of H. subiculosus it shows 29% identity (e-value 7e-35) to O-methyltransferase Hpm5 (ACD39764.1). Compared to A. parasiticus aflatoxin cluster it has 28% identity (e-value 2e-51) to AflO, a demethylsterigmatocystin 6-O-methyltransferase (AAS66016.1, Q9UQY0; 386 aa) and 25% identity (e-value 2e-28) to AflP, a sterigmatocystin 8-O-methyltransferase (AAS66017.1, 418 aa). All proteins (PenC, Hpm5, AflO, and AflP) have a O-methyltransferase domain (Pfam: PF00891) like HYFRA_G00010188. PenC has a rather unknown function in secondary metabolite biosynthesis of the alkaloid penigequinolone, whereas Hpm5 in interaction with the FAD-binding monooxygenase Hpm7 and cytochrome P450 monooxygenase Hpm1 is involved in biosynthesis (conversion of trans-7’,8’-dehydrozearalenol (DHZ) to aigialomycin C) of hypothemycin (Reeves et al. 2008).

The protein AflO (syn: OmtB) is described to be responsible for the conversion of demethylsterigmatocystin (DMST) to sterigmatocystin (ST) and of dihydrodemethylsterigmatocystin (DMDHST) to dihydrosterigmatocystin (DHST) (Yabe et al. 1989, 1998; Motomura et al. 1999). AflP (syn: OmtA) catalyzes the next step after OmtB, the conversion of both sterigmatocystin (ST) and dihydrosterigmatocystin (DHST) to O-methylsterigmatocystin (OMST) and dihydro-O-methylsterigmatocystin (DHOMST) (Yabe et al. 1989; Yu et al. 1995). Sterigmatocystin is a stable metabolite produced during aflatoxin B1 production and described to have cytotoxic effects in animals (reviewed by Zingales et al. 2020).

In strain 1511, there are two other upregulated genes encoding an O-methyltransferase domain (Pfam: PF00891). One is HYFRA_G00007274 (396 aa), which has best similarity to AflO (OmtB) demethylsterigmatocystin 6-O-methyltransferase of Aspergillus flavus (Q9P900, 386 aa) with 34% identity (e-value of 5.9e-50). The second-best hit is AflO of A. parasiticus (Q9UQY0, 386 aa) with an identity of 34% (e-value 2.3e-48). Both hits have the same O-methyltransferase domain (Pfam: PF00891) like the query. The other upregulated gene (HYFRA_G00003929, 422 aa) shows best similarity to grayanic acid biosynthesis cluster O-methyltransferase (46% identity; e-value 4.4e-121; E9KMQ4). Grayanic acid is the only known secondary metabolite known in Cladonia grayi (Armaleo et al. 2011). An additional BLAST of the amino acid sequence also shows that this gene is the best hit in H. fraxineus for AflP (OmtA) sterigmatocystin 8-O-methyltransferase (AAS66017.1, 418 aa) of A. parasiticus aflatoxin cluster (32% identity, e-value 9e-40). Like the upregulated predicted O-methyltransferase (HYFRA_G00010188) in strain 1431 both upregulated genes in 1511 show deduced amino acid similarity to O-methyltransferase Hpm5 (ACD39764.1) of H. subiculosus (identity of 26%; e-value 2e-28 for HYFRA_T00007274 and e-value 9e-14 for HYFRA_T00003929).

Another product of an upregulated gene HYFRA_G00013707 (255 aa) is similar to norsolorinic acid ketoreductase of Dothistroma septosporum (Nor1, M2Y1A3, 268 aa) with 44.5% identity (e-value 3e-56) and A. parasiticus (AflD, AAS66005, 271 aa) with 43% identity (e-value 5e-60). Both hits are having the same short chain dehydrogenase domain (Pfam: PF00106) like the query. Norsolorinic acid ketoreductase (aflD) catalyzes the dehydration of norsolorinic acid to form (1'S)-averantin in aflatoxin biosynthesis pathway of A. parasiticus (Zhou and Linz 1999). There is no similarity to any gene product of hypothemycin cluster of H. subiculosus.

Apart from the upregulated genes, there are more genes encoded in the genome of H. fraxineus showing similarity to genes of the aflatoxin biosynthesis cluster of A. parasiticus (Table 5) and hypothemycin biosynthesis cluster of H. subiculosus (Table 6).

Table 5 BLASTP results of aflatoxin gene cluster of Aspergillus parasiticus (AY371490.1) and predicted homologs in Hymenoscyphus fraxineus, incl. e-values and amino acid (aa) identities
Full size table
Table 6 BLASTP results of hypothemycin gene cluster of Hypomyces subiculosus (EU520418.1) and predicted homologs in Hymenoscyphus fraxineus, incl. e-values, and amino acid (aa) identities
Full size table

In strain 1511 there are more upregulated genes, showing similarity to genes of known toxin clusters. Two upregulated genes show similarity to genes for depudecin synthesis (DEP cluster) in Alternaria brassicicola (Wight et al. 2009). Depudecin is a polyketide that acts as a histone deacetylase (HDAC) inhibitor (Kwon et al. 1998) and contributes to virulence of A. brassicicola (Wight et al. 2009). This cluster consists of six genes (DEP1–DEP6), including four genes likely coding for enzymes that catalyze the synthesis of secondary metabolite, one gene encoding a MFS transporter (DEP3) and a gene that encodes a transcription factor (DEP6).

DEP5 is described to be the central enzyme in depudecin biosynthesis and belongs to type I reducing polyketide synthase. DEP2 and DEP4 are both predicted to encode monooxygenases. Whereas DEP2 belongs to class A and might be responsible for the epoxidations or hydroxylations of depudecin, DEP4 is of class B and likely can catalyze epoxidation. DEP1 is a protein with unknown function in depudecin biosynthesis.

The amino acid sequence of upregulated gene HYFRA_G00013234 (566 aa) is similar to FAD-dependent monooxygenase DEP4 of A. brassicicola (ACZ57547.1, 581 aa) with 50% identity. It has a pyridine nucleotide-disulfide oxidoreductase domain (Pfam: PF13738) like DEP4.

The amino acid sequence of the other upregulated gene (HYFRA_G00013226; 571 aa) and the DEP3 MFS transporter (ACZ57546.1) of depudecin cluster share 52% identity. Both proteins are predicted to belonging to the major facilitator superfamily of transporters (Pfam: PF07690) and having several predicted transmembrane domains.

Also, for the other proteins of A. brassicicola depudecin cluster a homologue sequence can be identified in H. fraxineus and amino acid sequences shared between 28 and 53% identity (Table 7).

Table 7 BLASTP results of depudecin gene cluster of Alternaria brassicicola (FJ977165.1) and predicted homologs in Hymenoscyphus fraxineus, incl. predicted function, length (aa), Pfam hits, e-values, and identities
Full size table

DEP1, the protein with unknown function in depudecin biosynthesis, shows 31% identity to HYFRA_T00013233. Both proteins have no detectable conserved domains, but several predicted transmembrane regions. DEP2 monooxygenase shows 37% identity to HYFRA_T00013230, sharing a conserved FAD-binding domain (Pfam: PF01494).

DEP5 (ACZ57548.1) polyketide synthase shows 53% identity to HYFRA_T00013232. They share six predicted Pfam domains (PF02801, PF00698, PF16197, PF08659, PF14765, PF00109); however, two domains predicted in H. fraxineus protein are missing in DEP5 of A. brassicicola. These are an alcohol dehydrogenase GroES-like domain (Pfam: PF08240) and Phosphopantetheine attachment site domain (Pfam: PF00550). The regulatory protein DEP6 (ACZ57549.1) exhibits 28% identity to HYFRA_T00013229, and both proteins share the same domains, a fungal Zn(2)-Cys(6) binuclear cluster domain (Pfam: PF00172) and a fungal-specific transcription factor domain (Pfam: PF04082). A BLAST of the identified H. fraxineus protein sequences against UniProtKB Swiss-Prot database shows that they are having best hits on proteins of depudecin cluster of A. brassicicola and Fusarium langsethiae.

In H. fraxineus the genes are located in in close proximity at scaffold 62 but exhibited gene arrangement (Fig. 2). The genes are flanked by HYFRA_T00013225 and HYFRA_T00013235. HYFRA_T00013225 is a predicted transmembrane protein, showing similarity to H( +)/Cl(-) exchange transporters and belongs to voltage-gated chloride channel family (Pfam: PF00654). HYFRA_T00013235 is a protein of unknown function. There are three more coding sequences located between the identified homologous genes which cannot be assigned to a function: HYFRA_T00013227 belongs to a protein family of unknown function (DUF3237), HYFRA_T00013228 has a predicted signal peptide region, and HYFRA_T00013231 has a NAD(P)H-binding domain (Pfam: PF13460). Beside HYFRA_G00013234 (predicted as DEP4) and HYFRA_G00013226 (predicted as DEP3), also HYFRA_G00013227 (unknown function) is upregulated in strain 1511.

Fig. 2
figure 2

Gene order in H. fraxineus showing homology to depudecin cluster (DEP 1–6) of Alternaria brassicicola (blue, continuous outline) and genes located inside or next to the cluster (gray, dashed outline), upregulated genes in strain 1511 are marked with an asterisk

Full size image

In strain 1431 none of those genes is predicted as upregulated during growth on F. excelsior compared to F. mandshurica and control media.

Another upregulated gene identified in strain 1511 is HYFRA_G00011677 (393 aa). It has a similarity to several trans-enoyl reductases, like ACTTS2 in Alternaria alternata (BAJ14523.1; BAJ09790.1; 343 aa) with 41.1% identity, e-value 2.6e-82. They share a predicted zinc-binding dehydrogenase domain (Pfam: PF00107) and belong to trans-enoyl reductase-like family (IPR047122). In A. alternata tangerine pathotypes this protein is essential for ACT-toxin biosynthesis and pathogenicity (Kohmoto et al. 1993; Ajiro et al. 2010).

Identification of differentially expressed genes involved in pathogenicity

We also hypothesized that better growth conditions in a susceptible host like F. excelsior go along with the higher expression of genes for pathogenicity in H. fraxineus. To identify genes with a high potential for involvement in pathogenicity, intersection of upregulated DEGs during growth in F. excelsior media (EX) versus F. mandshurica media (MA) of strain 1431 and 1511 had been analyzed. The comparison of 95 upregulated genes in EX compared to MA for isolate 1431 and 483 upregulated genes in EX compared to MA for isolate 1511 resulted in an overlapping set of four genes (Table 8; Fig. 3). Those four genes show a higher transcription rate in both strains grown in EX compared to MA.

Table 8 Upregulated genes of Hymenoscyphus fraxineus isolate 1511 and 1431 grown in maltose-yeast-peptone media (MYP) supplemented with plant material from Fraxinus excelsior (EX) compared to MYP media supplemented with plant material from Fraxinus mandshurica (MA)
Full size table
Fig. 3
figure 3

Upregulated genes of Hymenoscyphus fraxineus isolate 1511 and 1431 grown in maltose-yeast-peptone media (MYP) supplemented with plant material from Fraxinus excelsior (EX) compared to MYP media supplemented with plant material Fraxinus mandshurica (MA)

Full size image

One of those four shared genes (Table 9) is HYFRA_G00008114 (579 aa), which encodes a Baeyer–Villiger monooxygenase previously assigned to vir7 of viridiol biosynthetic gene cluster (Elfstrand et al. 2021) based on homology to vidF of Nodulisporium spp. (Wang et al. 2018). It has been demonstrated that VidF, together with VidP (esterase), and VidH (dehydrogenase), is involved in demethoxyviridin (furanosteroid) biosynthesis by catalyzing the pregnane side-chain cleavage (C17 − C20).

Table 9 Shared upregulated genes of Hymenoscyphus fraxineus isolate 1511 and 1431 grown in maltose-yeast-peptone media (MYP) supplemented with plant material from Fraxinus excelsior compared to MYP media supplemented with plant material Fraxinus mandshurica
Full size table

A further gene (HYFRA_G00001588, 398 aa) encodes for a protein of unknown function, for which Phobius predicts a signal peptide indicating secretion. Pfam shows a neprosin domain (PF03080), first described in pitcher plants as peptidase in digestive fluid (Lee et al. 2016).

Another gene (HYFRA_G00009022, 577 aa) has a similar protein sequence (30.2% identity; e-value 2.4e-62) to FAD-dependent monooxygenase DEP4 of F. langsethiae (A0A0M9ER62, 580 aa). It has a pyridine nucleotide-disulfide oxidoreductase domain (Pfam: PF07992), and also DEP4 protein of F. langsethiae belongs to the pyridine nucleotide-disulfide oxidoreductase family (Pfam: PF13738). For gene HYFRA_G00013234, upregulated in strain 1511 grown F. excelsior compared to F. mandshurica supplemented and control media, a similarity to an FAD-dependent monooxygenase DEP4 protein of A. brassicicola has been previously mentioned.

In the shared gene set, there is one more upregulated gene (HYFRA_G00004959; 322 aa) which is assigned to thioredoxin reductase gliT. It appeared already in the upregulated gene set of strain 1511 grown in F. excelsior compared to F. mandshurica supplemented and control media.

Discussion

The invasive ascomycete H. fraxineus is threatening F. excelsior populations throughout Europe. The exact mechanisms of pathogen–host interaction have not yet been elucidated. The comparison of the transcriptome during growth on media supplemented with F. excelsior and F. mandshurica leaf material performed in this study, shows that there are differentially expressed genes.

Transcriptome during growth on media supplemented with leaf material from F. excelsior compared to F. mandshurica and control media

In particular, the increased expression of genes involved in oxidative phosphorylation during growth of strain 1431 in F. excelsior supplemented media indicates better growth compared to F. mandshurica and control media. It was previously shown in Carrari et al. 2015 that growth rates were highest on F. excelsior supplemented media. This may be due to different or additional metabolites like carbon or nitrogen which leads to better growth. One may speculate why these genes are not upregulated in strain 1511 during growth in F. excelsior supplemented media. Probably strain 1511 was already in another physiological state at the timepoint of sampling, or it is missing the genes to utilize the metabolites which F. excelsior is providing. Also homing endonuclease genes, which are located on the same scaffold, are only predicted to be upregulated in strain 1431, but not in strain 1511.

Beside the upregulated genes for oxidative phosphorylation in strain 1431, in both strains there are genes upregulated with similarities to genes of the gliotoxin pathway, precisely a predicted gliA (strain 1431) and gliT (strain 1511). GliA is a transporter responsible for gliotoxin efflux (Wang et al. 2014) and GliT, an oxidoreductase, converts between the reduced (dithiol gliotoxin) and oxidized (gliotoxin) form (Scharf et al. 2010). In A. fumigatus both genes are involved in self-resistance, as deletion of gliA and gliT has led to enriched intracellular gliotoxin and increased sensitivity to exogenous gliotoxin (Scharf et al. 2010; Schrettl et al. 2010; Wang et al. 2014). Beside those two genes gliA and gliT, there are further genes for gliotoxin biosynthesis and regulation described in A. fumigatus (Scharf et al. 2012, 2016; Dolan et al. 2015), for which a putative homologous counterpart in H. fraxineus was identified. The identified putative homologous genes for gliotoxin biosynthesis do not seem to be organized in a gene cluster, as it is described for A. fumigatus (Gardiner and Howlett 2005) and for other ascomycetes (Patron et al. 2007). However, for Trichoderma spp. a separation of gliotoxin biosynthetic genes is known (Bulgari et al. 2020) and also for secondary metabolite coding genes in Fusarium spp. (Proctor et al. 2009). So far, it is unclear how gliotoxin production is stimulated, but for A. fumigatus it is suggested that different factors play a role like pH, temperature, composition, and aeration of the culture media (Scharf et al. 2012).

Probably, leave substrate of F. excelsior is stimulating the production of a unknown toxin similar to gliotoxin in H. fraxineus, which acts as a virulence factor and causes damage the host plant, which could be reflected in an upregulated gene expression of genes homolog to self-resistance genes gliT and gliA.

One may consider that the primary role of those two upregulated genes is not protection against own toxins, but the defense against extracellular toxins, like it has been demonstrated for T. virens GV29-8 (Bulgari et al. 2020). In this case one may speculate that elevated transcription of gliA and gliT could be induced due to a toxin from another fungus that colonizes the leaves of F. excelsior, which withstanded the autoclaving of the culture media. It is not clear if the toxin is stable during autoclaving the culture media, but it has been previously shown that the immunosurface of a A. fumigatus spore—released toxin is heat stable at least 5 min at 100 °C (Mitchell et al. 1997).

Another explanation for elevated toxin production and secretion, probably indicated by elevated gliA and gliT transcripts, is the prevention of side effects of intracellular oxidative stress for A. fumigatus (discussed in Schrettl et al. 2010; Owens et al. 2014). In this case, gliotoxin would act as electron acceptor and/or would help as buffer against endogenous and exogenous oxidants and is subsequently secreted. The cytotoxic effects on host cells therefore would be only a side effect. Probably, the increased growth of H. fraxineus in F. excelsior supplemented media leads to oxidative stress and stimulates the production of a molecule as part as an intracellular antioxidant defense system which unintentionally acts as toxin.

The comparison with the gene clusters for aflatoxin (A. parasiticus) and hypothemycin (H. subiculosus) shows that several genes with high similarities to genes involved in toxin production are encoded in the genome. Some of them are upregulated during growth on F. excelsior supplemented media compared to F. mandshurica and control. Those comprise genes with an O-methyltransferase domain (Pfam: PF00891) (HYFRA_G00010188 in strain 1431; HYFRA_G00007274 and HYFRA_G00003929 in strain 1511), but also one gene with a predicted cytochrome p450 domain (Pfam: PF00067) (HYFRA_G00005709 in strain 1431) and a short-chain dehydrogenase domain (Pfam: PF00106) (HYFRA_G00013707 in 1511). Like for gliotoxin, the growth conditions, especially pH and nutrients (especially of carbon and nitrogen), are described as parameter influencing aflatoxin production (Caceres et al. 2020). Oxidative stress is also proposed to induce aflatoxin production. Not only final aflatoxin but also intermediate compounds like versicolorin A could have strong cytotoxic effects, as shown for human intestinal cells (Gauthier et al. 2020). It has been shown previously that in H. fraxineus compared to H. albidus there are more homologs of versicolorin B synthase genes (Elfstrand et al. 2021).

The depudecin gene cluster is widely found within the Ascomycota, but it is discontinuously distributed (Reynolds et al. 2017). A. brassicicola depudecin is leading to lesions on cabbage leaves (Brassica oleracea). For all genes of depudecin cluster in A. brassicicola homologous genes organized in a cluster can be identified in H. fraxineus. One may consider that these genes are responsible for a depudecin-like secondary metabolite in H. fraxineus. Depudecin of A. brassicicola turned out to be a minor virulence factor (Wight et al. 2009), but there are also other HDAC inhibitors, like HC-toxin of C. carbonum, acting as strong virulence factor on maize (Brosch et al. 1995).

In summary, in H. fraxineus strains 1511 and 1431 grown in F. excelsior supplemented media compared to F. mandshurica and control, there are upregulated genes predicted to have similar functions and pathways, especially for gliotoxin, hypothemycin and aflatoxin, despite that there are no shared genes between both datasets.

However, genes predicted to be involved in oxidative phosphorylation as well as endonucleases are only predicted to be upregulated in 1431, and genes of depudecin pathway only in 1511. The different sample preparation methods of three pooled samples (strain 1511) versus triplicates (strain 1431), could account for a certain amount of differentially expressed genes predicted for the strains despite the same treatment. This is reflected in the DEGs set of strain 1431 grown in F. excelsior compared to F. mandshurica media (126 DEGs) in contrast to the same comparison group for 1511 (1,273 DEGs), which could be due to a triplicate outlier of strain 1431 grown in F. excelsior which can be seen in principal component analysis figure (not shown). The comparison results on the transcriptome of the two strains are also limited to the respective time point of sampling. Thus it may not consider the different growth rates of the strains and the physiological state e. g. exponential growth phase probably reflected by the oxidative phosphorylation. Beside those thoughts, it could be also that the strains have a unique transcriptome profile.

Shared upregulated genes between both strains during growth on media supplemented with leaf material from F. excelsior compared to F. mandshurica

Four genes are upregulated in both strains during growth on F. excelsior supplemented media, compared to F. mandshurica.

One gene encodes a protein with an unknown function (HYFRA_G00001588). It has a predicted neprosin domain (Pfam: PF03080) and a signal peptide. Neprosin is described as a protease with prolyl endopeptidase (PEP) activity (Lee et al. 2016). The exact function of this protein in H. fraxineus is not clear. As it possesses a signal peptide, it likely has an extracellular peptide cleavage activity.

Another gene (HYFRA_G00009022) shows similarity to FAD-dependent monooxygenase DEP4 (A0A0M9ER62) of F. langsethiae. In Alternaria brassicicola this gene belongs to the depudecin gene cluster, an inhibitor of histone deacetylase (Wight et al. 2009). The exact function of this protein in H. fraxineus is not clear, but as discussed previously it is probably involved in production of a toxin similar to depudecin.

The upregulated gene HYFRA_G00004959 has an identity to thioredoxin reductase gliT of gliotoxin cluster of A. fumigatus, which has already been discussed previously. In the upregulated gene set of F. excelsior supplemented media compared to F. mandshurica and control media it was only upregulated in strain 1511. However, it is predicted to be upregulated in both strains grown in F. excelsior supplemented media compared to F. mandshurica.

The upregulated gene HYFRA_G00008114 is encoding vir7, a Baeyer–Villiger monooxygenase, of viridiol biosynthetic gene cluster (Elfstrand et al. 2021). Viridiol can induce necrosis on ash tissues (Andersson et al. 2010; Cleary et al. 2014), but it is unlikely to be the main responsible agent for ash dieback symptoms, as it is also produced by the non-phytopathogenic H. albidus and higher concentrations do not seem to come with higher virulence (Junker et al. 2014). Interestingly, for both strains this gene is only upregulated during growth in F. excelsior compared to F. mandshurica, but not control media. This result may be interpreted with an acceleration of foliage senescence and suppression of other competing microbial organisms. Such interaction is largely harmless for the host at the end of the growing season in Asia but may promote disease in Europe. In strain 1511 there are additional genes of this cluster predicted to be upregulated during growth in F. excelsior media compared to F. mandshurica: vir4 (HYFRA_G00008111), vir6 (HYFRA_G00008113), vir8 (HYFRA_G00008115), and vir10 (HYFRA_G00008117).

Hymenosetin is another described secondary metabolite described for H. fraxineus (Halecker et al. 2014). It is discussed to have antimicrobial effects to provide an advantage in terms of substrate availability, but shown to have no or only moderate cytotoxic effects. Genes belonging to the biosynthetic pathway of hymenosetin, do not seem to be upregulated in our analyzed datasets.

It has been shown that the defense mechanisms in F. mandshurica against H. fraxineus take place in the leaves (Gross and Holdenrieder 2015). This suggests that the leaves of the different ash species have an influence on the growth and virulence mechanisms and therefore gene expression profile of H. fraxineus. This is reflected by our results, which show differentially expressed genes between growth in media supplemented with F. excelsior and F. mandshurica leaf material. It is very likely that some of the higher expressed genes are involved in virulence mechanisms.