An Ammonium Transporter Gene Contributes to the Aggressiveness of the Dutch Elm Disease Pathogen Ophiostoma novo-ulmi

Abstract

Molecular mechanisms determining pathogenicity of the Dutch elm disease fungus Ophiostoma novo-ulmi are poorly understood. Prior identification of the pathogenicity locus pat1 prompted a chromosome walking approach to elucidate gene function in this region. Among 17 identified genes, ONUg0282 (amtA) was predicted to encode a high-affinity ammonium transporter. In silico analyses confirmed the presence of four additional amt genes (amtB, amtC, amtD, and amtE) in both O. novo-ulmi and the less aggressive O. ulmi and that amtA and amtB belong to the Saccharomyces cerevisiae mep2 clade. The predicted amtA gene product showed features of Mep2-type transceptors, including amino acid residues corresponding to His-168 and His-318 in Escherichia coli AmtB protein, 11 transmembrane helices, and a conserved 22 amino acid motif immediately downstream of the last transmembrane helix. A knockdown amtA mutant with 25% residual expression was significantly less aggressive than wild-type O. novo-ulmi strain H327 when infecting Ulmus americana × U. parvifolia saplings. Predicted AmtA transporters from two CRISPR-Cas9 knockout mutants contained only five intact transmembrane helices. The ΔamtA mutants retained several wild-type phenotypic traits, including yeast–mycelium dimorphism, but were significantly less aggressive than H327 towards U. americana saplings. We concluded that ONUg0282 is an important determinant of aggressiveness in O. novo-ulmi.

1. Introduction

The invasive ascomycete fungi Ophiostoma ulmi and O. novo-ulmi are responsible for two pandemics of Dutch elm disease (DED), a vascular wilt that has strongly impacted elm populations worldwide [1]. Ophiostoma ulmi is a moderately aggressive species, which caused the first pandemic, before it was gradually replaced by the highly aggressive O. novo-ulmi, which is associated with the current ongoing pandemic [2]. Genetic analyses of controlled crosses between O. ulmi and O. novo-ulmi indicated that the aggressiveness was under polygenic control [3,4]. Sequencing of O. ulmi and O. novo-ulmi nuclear genomes [5,6,7,8] confirmed the occurrence of over 1700 genes with homologs in the PHI-Base database of genomic sequences encoding proteins presumed to be involved in host–pathogen interactions [9,10]. Comparative genomics of over 75 Sordariomycete species, including 20 wilt-causing species, suggested that the pathogenicity of DED fungi was not associated with large-scale gene expansions and was more likely driven by a few large-impact genes. These genes might have evolved from existing genes or were acquired through duplications, either de novo or via horizontal gene transfer [11].

The latter origin was supported by in silico analyses of secondary metabolite biosynthetic gene clusters in Ophiostoma spp. that contained the fujikurin-like cluster OpPKS8, which appears to have been horizontally acquired and may contribute to pathogen fitness and host interactions [12]. At present, only three candidate genes for pathogenicity and aggressiveness have been identified and validated through the inoculation of null mutants. Two of these genes, ade1 (ONUg6779) and ade7 (ONUg7434), encode enzymes involved in the biosynthesis of adenine. Ophiostoma novo-ulmi deletion mutants for ade1 or ade7 are auxotrophic and non-aggressive [13], similar to what has been reported in adenine auxotrophs in other plant pathogenic fungi [14,15,16]. The third gene, ogf1 (ONUg1537), encodes a homolog of the GPF1 transcription factor [17], and its inactivation resulted in a significant decrease in aggressiveness when Ulmus americana saplings were inoculated with low inoculum doses (3.12 × 10³ or 1.25 × 10⁴ cells) of O. novo-ulmi [18].

In fungi, morphogenetic transitions, such as the yeast-to-mycelium (Y-M) switch, are tightly regulated by conserved signaling pathways, including the cyclic AMP-dependent protein kinase A (cAMP/PKA) pathway, mitogen-activated protein kinase (MAPK) cascades, and the Rim/PaI pathway [19,20,21]. The switch to mycelial growth is often mediated by both MAPK and cAMP pathways in response to nitrogen starvation [21], whereas the RIM pathway is pH-dependent [20]. Yeast–mycelium dimorphism is not uncommon in fungal pathogens of plants, insects, and mammals [21,22,23], including the DED pathogens in which the Y-to-M switch is documented both in vitro [22,23] and in planta [24]. The loss of ability to transition from yeast to mycelial growth has been shown to induce the loss of virulence in the thermophilic Ascomycete human pathogens Candida albicans, Blastomyces dermatitidis, and Histoplasma capsulatum [25,26], as well as in the Basidiomycete plant pathogen Ustilago maydis [27] and the Mucoromycete human pathogen Mucor circinelloides [28]. In O. novo-ulmi, genes involved in cAMP/PKA signaling are highly expressed during the Y-M transition in vitro, suggesting that modulation of these pathways could contribute to pathogen aggressiveness and adaptation to elm hosts [29].

Previous analysis of sexual progeny recovered from a controlled genetic cross between highly aggressive O. novo-ulmi ssp. novo-ulmi H327 and introgressant strain AST27 displaying moderate aggressiveness towards U. procera showed a 1:1 segregation for the high and moderate aggressiveness phenotypes, suggesting that these were controlled by two different alleles at a single nuclear locus, designated pat1 [30]. The alleles present in strains H327 and AST27 were designated pat1-h and pat1-m, respectively. Several O. ulmi-type Random Amplified Polymorphic DNA (RAPD) polymorphisms co-segregated with the pat1-m allele in the H327 × AST27 progeny, as well as in backcross progeny between H327 and F₁ strain A2P30 carrying the pat1-m allele [30]. These observations, made before the advent of next-generation sequencing (NGS), motivated our initial efforts to (1) identify the pat1 locus through chromosome walking and (2) subsequently validate its role in aggressiveness by targeted mutagenesis. Here, we report the partial characterization of the pat1 gene region and present experimental evidence that candidate sequences for pat1 include a gene encoding a high-affinity ammonium transporter (Amt). Although amt genes have been implicated in aggressiveness for a number of plant pathogenic fungi [31,32,33], they have not been studied in DED pathogens. We therefore investigated whether the candidate gene, denoted amtA, contributes to the Y-M transition and influences the aggressiveness in O. novo-ulmi.

2. Materials and Methods

2.1. Fungal Strains

The highly aggressive Ophiostoma novo-ulmi ssp. novo-ulmi strain H327 carrying the pat1-m and mat1-1 alleles [30], moderately aggressive introgressant strain AST27 (pat1-m, mat1-2), and O. ulmi Q412T (pat1-m, mat1-1) were used. Additional strains included O. novo-ulmi laboratory strain F5-28 (pat1-h, mat1-2), strain A3P11 (pat1-m, mat1-1) from the H327 × A2P30 backcross [30,34], as well as mutants derived from strain H327 by RNA interference (RNAi) or CRISPR-Cas9, as described below.

2.2. Chromosome Walking to the Putative pat1 Locus

We attempted to identify and clone the DNA sequence corresponding to locus pat1 by using a series of overlapping DNA clones prepared from RAPD polymorphisms that co-segregated with pat1 alleles. Total genomic DNA was recovered from lyophilized yeast-phase cultures of O. ulmi and O. novo-ulmi according to the protocol of Zolan and Pukkila [35]. RAPD primers OPK3 and OPJ20 (QIAGEN Operon, Alameda, CA, USA; Table S1) were used to amplify DNA polymorphisms closely linked to pat1 [30] in DNA from strains H327, AST27, and Q412T. The amplicons were separated by electrophoresis in gels containing 0.8 to 2% agarose. Bands with a length corresponding to the expected marker size (1050 nt and 540 nt for OPK3 and OPJ20, respectively; Table S1) were reamplified by PCR and cloned using Vector pGEMt-easy (Promega, Madison, WI, USA) and One Shot^® TOP10 competent cells (Invitrogen, Carlsbad, CA, USA). Cloned Ophiostoma DNA was recovered through minipreps using QIAQuick (QIAGEN, Mississauga, ON, Canada) columns prior to Sanger sequencing at the Laval University Genomic Analysis Platform (PAG; Québec City, QC, Canada). Full-length clones were first used as probes in Southern hybridization with O. novo-ulmi chromosomes separated by pulsed-field gel electrophoresis (PFGE) [36]. Since each probe hybridized with more than one site, shorter probes (ca. 500 nt) were produced by PCR amplification (Table S1), eluted from electrophoresis gels, labeled with ³²P using the Rediprime II random prime labeling system (GE Healthcare, Piscataway, NJ, USA), and purified with the QIAGEN QIAQuick nucleotide removal kit prior to being used in Southern hybridization with genomic DNA.

A genomic library for strain H327 was prepared with the Lambda FIX^® II/Xho I Partial Fill-in Vector Kit (Stratagene, Cedar Creek, TX, USA), according to the manufacturer’s instructions and propagated in XL1 Blue MRA (Invitrogen) bacterial cells [37]. Phage plaques were transferred onto Duralose-UV^TM (Stratagene) circular membranes and screened with ³²P. A total of 10 membranes (Total titer: 500,000 phages) were screened in order to identify several positive clones for each probe. The restriction fragment profiles of five or ten clones per hybridization were analyzed in order to select clones with a large insert and ensure the latter targeted a single genomic region. Phage DNA was extracted using Lambda Midi Kit columns (QIAGEN) and Ion-Exchange Pre-swollen DE52 cellulose columns (Whatman, Kent, UK) [38]. Restriction analyses were carried out on 2 µg phage DNA with NotI, XbaI, and SpeI restriction enzymes (New England Biolabs, Pickering, ON, Canada) used individually or in combination. The longest clone was digested with one or two restriction enzymes, so that 1 to 6 kb-long fragments from the O. novo-ulmi DNA insert could be subcloned into bacterial vectors. Inserts were sequenced on an ABI 3730xl automated DNA sequencer (Applied Biosystems, Foster City, CA, USA) at the Plateforme de séquençage et de génotypage du CRCHUL (Québec, QC, Canada). One DNA strand was sequenced in all clones (2% error rate). Nucleotides at the extremities of the fragment sequenced were used to design primers for additional rounds of sequencing. The translated sequences were submitted to BlastX 2.2.4 searches of non-redundant protein sequences (nr) in the GenBank database at the National Center for Biotechnology Information (NCBI: https://www.ncbi.nlm.nih.gov/, accessed between 1 March 2004 and 30 November 2004).

2.3. In Silico Characterization of O. novo-ulmi ONUg0282

Nucleotide sequences of ONUg0282 cloned from O. novo-ulmi H327 and introgressant strain AST27 were first compared to sequences for orthologs in other ascomycetes, which were retrieved from the NCBI website, except for the sequence from O. ulmi strain W9, which was downloaded from http://www.moseslab.csb.utoronto.ca/o.ulmi/ (accessed on 17 September 2025). Paralogs of ONUg0282 were also identified in O. novo-ulmi H327, O. ulmi W9, and other ascomycetes. Amino acid sequences for orthologs and paralogs of ONUg0282 DNA sequences were predicted using EMBOSS Transeq (https://www.ebi.ac.uk/jdispatcher/st/emboss_transeq, accessed on 20 September 2025) [39]. Nucleotide sequences and predicted amino acid sequences were aligned using CLUSTAL Omega (https://www.ebi.ac.uk/jdispatcher/msa/clustalo, accessed on 20 September 2025) [39]. The phylogenetic relationships among 43 Amt/Mep-encoding sequences from Ascomycete species, including the DED fungi, were inferred by the Neighbor-Joining method [40] implemented in MEGA12 utilizing up to 8 parallel computing threads [41]. The confidence intervals of the phylogenetic trees were tested using the bootstrap statistical test [42] with 1000 resampling iterations. The DeepTMHMM 1.0 algorithm [43] available on the Danish Technical University website (https://services.healthtech.dtu.dk/services/DeepTMHMM-1.0/, accessed on 25 September 2025) was used to verify whether the predicted protein encoded by ONUg0282 had the characteristic topology of transmembrane proteins. The segregation of O. novo-ulmi- and O. ulmi-type alleles for ONUg0282 in H327 × AST27 F₁ progeny [30] was determined by PCR amplification of allele-specific primers (Table S2).

2.4. Production and Recovery of Targeted Mutants in O. novo-ulmi ssp. novo-ulmi H327

We first attempted to recover knockout O. novo-ulmi mutants using a transformation protocol we had used previously for tagging potential pathogenicity genes in this species [44]. Knockout vector Tranhmb was constructed by PCR amplification of the first 799 nucleotides of ONUg0282, the 1.4 kb of the Hmb cassette, and the 3’ end of ONUg0282 (bases 929 to 1596) and inserted into pDrive (QIAGEN, Mississauga, ON, Canada). The linearized vector was used to transform protoplasts of strain H327. Colonies growing on Ophiostoma Complete Medium (OCM) agar [45] supplemented with 0.6 M sucrose and hygromycin (200 µg/mL) were retrieved and assessed by PCR for homologous recombination. We also inserted the O. novo-ulmi ONUg0282 allele into the genome of Pat1-m strains Q412T and A3p11. The complementation vector S2 was constructed in pDrive by cloning a H327 genomic DNA fragment spanning ONUg02822 and 5166 nt upstream of the ATG start codon. Protoplasts of strain A3P11 were co-transformed with vector S2 and plasmid pPS57 harboring the bacterial hygromycin phosphotransferase gene (hph), and transformants were retrieved on OCM agar with 0.6 M sucrose and hygromycin (200 µg/mL).

Knockdown mutants for ONUg0282 in O. novo-ulmi were produced by RNA interference (RNAi) according to the procedure used previously with DED pathogens [46,47]. Subsequently, deletion mutants for ONUg0282 were recovered using a CRISPR-Cas9 [48] protocol adapted for Ophiostoma species [49]. This protocol is based on the transfection of protoplasts from a pink adenine-requiring mutant [50] with RNA guides targeting the deficient adenine biosynthetic gene (ONUg7434) and the gene of interest. RNA guides were assembled by mixing equimolar amounts of Alt-R CRISPR-Cas9 tracrRNA and Alt-R crRNA oligonucleotides (both obtained from Integrated DNA Technologies, Coralville, IA, USA) targeting ONUg7434 (ade7: ACCGCAATCAGTACCACCGC) and ONUg0282 (amtA: AACCGAACCAGCCGATCCAC). Lipofectamine RNAiMAX (Thermo Fisher Scientific, Ottawa, ON, Canada) was used to increase the transfection efficiency of protoplasts prepared from a pink Ade^- mutant of O. novo-ulmi H327 [51]. Following the overnight incubation of protoplasts to allow for cell wall regeneration, adenine prototrophy-reverted transformants (Ade⁺) were recovered on adenine-free medium, and gene ONUg0282 was amplified by colony PCR. The PCR products were then digested in vitro using the CRISPR-Cas9 RNP complex. Candidate mutants were validated by Sanger sequencing at the U. Laval PAG. The sequences of the PCR primers for DNA sequencing are provided in Table S2. Mutants were stored at −80 °C in 15% glycerol.

2.5. Phenotyping of Targeted Mutants In Vitro for Growth, Mating Capacity, and Target Gene Expression

Mycelial growth was assessed at 21, 28, and 30.5 °C on Oxoid (Thermo Fisher Scientific, Ottawa, ON, Canada) Malt Extract Agar (MEA), as well as on MEA supplemented with 1.0 M sodium chloride (NaCl). Colony morphology was recorded on Ophiostoma Minimal Medium (OMM) agar [45] containing different concentrations of nitrogen in the form of ammonium sulphate. The mycelial growth was also assessed on OMM agar supplemented with methylammonium (MilliporeSigma, Oakville, ON, Canada) at concentrations up to 200 mM. Yeast cultures were produced by incubating plugs of mycelium in liquid OMM with L-proline (1.15 g/L) as the nitrogen source [23], and placing cultures on a rotary shaker at 150 rpm at 21 °C. The capacity of mutants to mate with a sexually compatible strain was tested by conducting reciprocal crosses on elm sapwood agar [52] supplemented with linoleic acid at 6 ml/L (ESAL) [45]. A plug of agar-bearing mycelium of the parent that would act as recipient was first allowed to grow on ESAL before it was fertilized with a yeast suspension from the donor parent. Unless stated otherwise, all growth experiments included at least three biological replicates.

Hyphal development and morphological responses were analyzed using light microscopy after spore inoculation on various media conditions. Spores were inoculated on solid minimal medium and incubated for 24 h under standard growth conditions. For nutrient limitation experiments, spores were grown on nutrient-depleted OMM agar lacking sucrose and ammonium sulfate. Carbon source utilization assays were performed using modified OMM agar supplemented with alternative carbon sources (galactose or maltose) at equimolar concentrations. Microscopic images were captured at constant magnification (40×) using a Zeiss (Oberkochen, Germany) Axio-Imager Z2 upright widefield fluorescence microscope at the University of Montreal Bio-Imaging Platform. Hyphal morphology parameters including elongation, branching patterns, growth organization, and biomass development were qualitatively assessed by comparing WT strain H327 with CRISPR-Cas9 ΔamtA-3 deletion mutant under each experimental condition. Images were acquired with standardized exposure settings.

Reverse Transcription-Quantitative PCR (RT-qPCR) was used to assess ONUg0282 expression in RNAi mutants. Mycelium from strain H327 and RNAi mutants (5 replicates/strain) was harvested after two days of incubation on OMM agar with or without nitrogen and ground with a Mixer Mill MM300 (Retsch, Newtown, PA, USA) in 1.5 mL micro tubes (Sarstedt, Montreal, QC, Canada) with tungsten beads (QIAGEN). The RNAs were extracted with the RNeasy plant mini kit (QIAGEN). The RNA concentration was calculated by spectrophotometry with Multiskan Spectrum (Thermo Electron, Milford, MA, USA). RNAs (2 ug) were treated with DNAse I, RNase-free (Roche Diagnostics GmbH, Mannheim, Germany) in 25 mM Tris-HCl, 5 mM MgCl₂, and 0.1 mM EDTA pH 7.2.The RNA quality was determined with an RNA 6000 Nano Assay kit in a bioAnalyser 2100 (Agilent Technologies, Mississauga, ON, Canada). First-strand cDNAs were synthesized from 2 ug RNA with SuperScript^TM II Reverse Transcriptase and oligo-dT primer, according to the manufacturer’s instructions (Invitrogen, Carlsbad, CA, USA). The cDNA was diluted 1:5 in RNase-free water and stored at −80 °C.

Complementary DNAs (15 ng) were used as template in a RT-qPCR assay using the QuantiTect^® SYBR^® Green PCR kit (QIAGEN) with a final volume of 15 ul. Primer pairs were designed for gene ONUg0282 and gene ONUg8051 encoding actin (Table S2). Reactions were performed on the model 7500 real-time PCR cycler (Applied Biosystems, Carlsbad, CA, USA) with the following protocol: 15 min at 95 °C, followed by 40 cycles of 10 sec each at 94 °C and 2 min at 62 °C. Melting curves were obtained at the end of the 40 cycles to determine the production of dimers. All reactions included duplicate technical replications. The PCR efficiency of each primer pair was determined using standard curves constructed with serial dilutions of linear copies of plasmid pGEM-t-easy (Promega, Madison, WI, USA) containing either ONUg0282 or the reference gene ONUg8051.

2.6. Pathogenicity Tests

Fungal strains were first assayed on Golden Delicious apples [53]. Plugs of actively growing mycelium on MEA were inserted, mycelium facing inwards, into wounds made with a 9 mm diameter sterile cork borer, with 5 to 8 biological replicates per strain tested. A negative control treatment consisted of inserting a plug of sterile MEA into apples. Diameters of necrotic lesions developing on apples kept at room temperature (21 °C) were assessed at 28 days post inoculation (dpi). Two independent experiments were conducted. Fungal strains were also inoculated onto moderately resistant Ulmus americana × U. parvifolia hybrid clone 2213 [54,55] or onto two- or three-year old saplings of susceptible Ulmus americana grown from a local seed source, using the procedure described previously [13]. Inoculations were carried out in a growth chamber or in a PPC1-certified compartment with 16 h of daylight and temperatures kept at 24 °C during the day and 18 °C during the night [55]. Unless otherwise stated, the fungal inoculum consisted of 6.25 × 10³ yeast cells grown in liquid OMM with proline as the nitrogen source. Non-inoculated control saplings were injected with sterile distilled water. Each treatment included at least eight biological replicates. Saplings were observed at frequent intervals over up to 63 dpi, and foliar symptoms (wilting, yellowing, browning and defoliation) were quantified. Four independent assays were conducted with CRISPR-Cas9 mutants: three assays with mutant Δamt-3 and one with mutant Δamt-35. The development of Ophiostoma strains within the xylem of inoculated saplings was evaluated through their re-isolation on a semi-selective medium containing cycloheximide, streptomycin sulfate, and chloramphenicol [52,56].

3. Results

3.1. Identification of Genes in the pat1 Locus Region Through a Chromosome Walk

The molecular identification of the chromosomal region containing the pat1 locus was attempted before a reference genome was available for the DED fungi. Hence, a chromosome walk was launched with probes for markers OPK3₁₀₅₀ and OPJ20₅₄₀ linked to the pat1 locus previously identified by Mendelian co-segregation in the progeny of the H327 (pat1-h) × AST27 (pat1-m) cross [30]. The 500-nt probes designed from RAPD markers OPK3₁₀₅₀ and OPJ20₅₄₀ each hybridized to a single genomic region and were thus suitable for initiating the chromosome walk. The largest (16 kb) overlapping insert from genomic clones identified with the OPK3₁₀₅₀ probe was sequenced. Primers T7 and T3 permitted the identification of five additional genomic clones for sequencing and thus the extension of the walk upstream (80 nt) and downstream (11 kb) of the initial 16 kb of sequence around the OPK3₁₀₅₀ locus. The walk initiated with the OPJ₅₄₀ probe enabled us to sequence 19.57 Kb of genomic DNA.

Attempts to close the gap between the genomic regions identified by probes OPK3₁₀₅₀ and OPJ20₅₄₀ through chromosome walking were unsuccessful. Initial bioinformatic analysis of 46 927 bp from two non-overlapping genomic clones identified five open reading frames (ORFs) on fragment OPJ20 and eight ORFs on fragment OPK3 with strong homology to genes from other fungal species. Once an annotated reference genome became available for O. novo-ulmi ssp. novo-ulmi strain H327 [6,7], we reassessed the results from the original bioinformatic analysis. Our analysis established that the genomic regions identified using markers OPK3₁₀₅₀ and OPJ20₅₄₀ included 17 genes in total and were located 157.5 kb from each other near one end of chromosome I (Figure 1). With the exception of ONUg0281 encoding a DNA polymerase, at least three or more transcripts per gene were detected in Ophiostoma transcriptome datasets (Table S3) from previous studies conducted in planta [18,47].

The 17 ORFs included three genes with orthologs in the PHI-Base curated database of genes involved in host–pathogen interactions [9,10]: ONUg0229 encoding a malate dehydrogenase (PHI:2959; Reduced virulence), ONUg0278 encoding transcription initiation factor TFIID subunit 3 (PHI:1316; Unaffected pathogenicity), and ONUg0282 predicted to code for a member of the ammonium transporter/methylammonium permease (Amt/Mep) family (PHI:2710; Reduced virulence). ONUg0229 was the one gene for which the most transcripts were detected in U. americana inoculated with Ophiostoma species including O. novo-ulmi, O. ulmi, and the saprobe O. quercus (Table S3). In contrast, transcripts of gene ONUg0278 were detected in planta in low amounts for the DED fungi but not for O. quercus. Lastly, gene ONUg0282 was expressed after O. novo-ulmi was inoculated onto susceptible U. americana saplings but not onto the resistant ones, whereas transcripts were barely detectable in O. quercus transcriptomes in planta (Table S3). Gene ONUg0282 was therefore considered one of several strong candidates for pat1 and retained for further analyses.

3.2. In Silico Analyses of ONUg0282

The ONUg0282 gene contains 1596 nucleotides and has no intron (Figure S1). The allele present in the introgressant strain AST27 is highly similar to the allele observed in O. ulmi W9, except for close to 50 single nucleotide polymorphisms (SNPs) dispersed throughout the sequence. The most notable difference between the AST27/W9 alleles and the H327 allele is the presence of a 21-nt insertion near the 3’-end of the former (Figure S1). Alignment of the predicted amino acid (aa) sequence for ONUg0282 with orthologs and paralogs in other ascomycete fungi (Figure 2) confirmed that it had high homology with genes encoding Amt/Mep proteins and that it belonged to the S. cerevisiae mep2 clade (Figure 3A). The latter also contained two mep genes from Collelotrichum gloeosporioides, including the mepB gene deposited in PHI-Base [31], and two amt/mep genes from the ophiostomatoid species Sporothrix schenckii (Figure 3A). Since ONUg0282 was the first Amt/Mep-encoding gene identified in O. novo-ulmi, it was designated amtA. A thorough search of the O. novo-ulmi ssp. novo-ulmi H327 reference genome [7] for paralogs of ONUg0282 led to the identification of ONUg1681 (amtB), ONUg3050 (amtC), ONUg3952 (amtD), and ONUg4730 (amtE). Transcripts for these four genes were present in previously published transcriptomes of O. novo-ulmi H327 recovered in vitro [29,57] and in planta [18,47] (Table S4).

Homologs of all five O. novo-ulmi Amt-encoding genes were found in the genome of O. ulmi W9. According to the phylogenetic analysis, sequence amtB (ONUg1681) was also part of the S. cerevisiae mep2 clade (Figure 3A). The predicted protein encoded by ONUg0282 exhibited structural characteristics typical of Amt/Mep ammonium transporters, notably the presence of residues corresponding to His-168 and His-318 of the reference Escherichia coli AmtB protein (Figure 2), the presence of 11 transmembrane helices with an N_out-C_in topology (Figure 3B), and the occurrence of a conserved 22 aa motif (PGLHLRASEEAEILGIDDSEIG) immediately downstream of TMH 11 (Figure 2). The insert present in the C-terminal portion of the predicted amino acids encoded by ONUg0282 alleles found in O. ulmi W9 and the introgressant strain AST27 included three contiguous serine residues. Nevertheless, the predicted topologies of the predicted proteins encoded by the AST27 and W9 amt alleles were similar to those determined by the DeepTMHMM algorithm for the O. novo-ulmi allele (Figure 3B). The O. ulmi- and O. novo-ulmi alleles for amtA co-segregated with pat1-m and pat1-h alleles in the F₁ progeny from the cross H327 × AST27. Comparison of the predicted AmtA and AMtB amino acids for O. ulmi, O. novo-ulmi, S. schenckii, and C. gloeosporioides based on BLOcks SUbstitution Matrix (BLOSUM) analysis [58] highlighted the amino acids that differed between the two transporters (Figure 2).

3.3. Recovery and Analysis of Mutants for g0282

No knockout mutants were obtained using insertional mutagenesis [44], whereas the O. novo-ulmi allele for amtA was successfully integrated into the genome of the backcross strain A3P11 but not into O. ulmi Q412T. A first set of seven O. novo-ulmi knockdown mutants for ONUg0282 was obtained by RNAi of the WT strain H327. Prior to phenotyping the mutants, the expression of amtA in strains H327 (Pat1-h) and A3P11 (Pat1-m) grown for two days on OMM agar was evaluated by RT-qPCR. The expression in both strains was lowest in the presence of a high concentration (10 mM) of ammonium (Figure 4A). Transcript abundance of amtA in the mycelium of O. novo-ulmi RNAi mutants grown for two days on OMM agar without ammonium was comparable to that of the Pat1-m progeny strain A3p11 but significantly lower (p < 0.05) than the level measured in strain H327 (Figure 4B).

Complementation of the Pat1-m progeny strain A3P11 with a H327-type amtA allele (strains A3P11/6, A3P11/8 and A3P11/10) resulted in the partial restoration of gene expression. Mutant strain T2 showed the lowest level of transcription and was inoculated onto U. americana × U. parvifolia hybrid clone 2213, along with strains H327, A3P11, and complemented strain A3P11/8. The inoculum consisted of 1.75 × 10³ yeast cells. The mutant induced significantly fewer (p < 0.05) leaf symptoms than strain H327 and progeny strain A3P11 after 3 and 6 weeks (Figure 4C). The aggressiveness of strain A3P11 did not differ from that of the complemented strain A3P11/8. Vascular browning was observed in the xylem of saplings inoculated with fungal strains tested but not in saplings injected with sterile water. The four strains were reisolated from the area surrounding the inoculation point, at mid-stem (20–30 cm from the infection point), and in the distal portion of the stem (40 cm and up).

Four O. novo-ulmi ΔONUg0282 mutants were obtained using CRISP-Cas9, and two of them were analyzed further. A single nucleotide at position 743 in the WT allele was deleted to obtain mutant ΔamtA-35, whereas a stretch of 14 nucleotides (positions 734 to 747) was deleted for mutant ΔamtA-3 (Figure S2 and

Table 1. Genotype and phenotype of CRISPR-Cas9 ΔamtA mutants of Ophiostoma novo-ulmi H327.

Genomic or Phenotypic Trait	H327	ΔamtA-3	ΔamtA-35
Position of CRISPR-induced deletion(s) 1	NA 2	743–747	743
Functional alpha helices in predicted AmtA protein	11	5	5
Mycelial growth rate on MEA at 21 °C (mm/day)	3.01	3.82 *	3.83 *
Mycelial growth rate on MEA at 30.5 °C (mm/day)	0.82	0.82	0.88
Mycelial growth rate on MEA + 1.0 M NaCl at 21 °C (mm/day)	0.91	0.67 **	0.80
Mycelial growth rate on MEA + 1.0 M NaCl at 30.5 °C (mm/day)	0.23	0.24	0.17
Mycelial growth rate on MEA + 0.2 M methylammonium at 21 °C	Yes	Yes	Yes
Male-fertile	Yes	Yes	Yes
Female-fertile	Yes	Yes	Yes

¹ Position of amtA gene nucleotide(s) deleted by CRISPR-Cas9 gene editing. ² No deletion in wild-type strain H327. * and ** indicate significant differences from W H327 at p < 0.05 and p < 0.01, respectively, according to ANOVA.

). The mutations had dramatic effects on the amino acid sequence and three-dimensional structural topology of the predicted proteins, as evidenced by the conformational alteration of the 6th alpha helix and loss of alpha helices 7 to 11 in the predicted proteins synthesized by these mutants (Figure 3B and Table 1). The two mutants produced yeast cells in shake liquid MM with proline as the nitrogen source and mycelium on solid media. On MEA, mutants had a slightly but significantly higher (p < 0.05) mycelial growth rate than WT H327 at 21 °C but not at 30.5 °C. Mutant ΔamtA-3 was significantly (p < 0.05) more sensitive to the presence of 1.0 M NaCl in MEA when incubated at 21 °C. Wild-type and mutants showed similar responses to methylammonium (up to 200 mM) in OMM agar (Table 1). When incubated at 21 °C on OMM agar, mycelial colonies of WT H327 and ΔamtA mutants included a waxy-looking central zone, in which yeast cells were observed, and an outer zone where only mycelium was observed (Figure 5A). When ammonium was omitted, colonies of the ΔamtA-3 mutant exhibited a significantly (p < 0.05) larger waxy zone than WT H327 colonies after four days of incubation. Mycelial growth in the outer zone of colonies was also significantly (p < 0.05) more pronounced in the ΔamtA-3 mutants during the first two days of incubation. In both WT and ΔamtA strains, the development of mycelium was inhibited during the first two days of incubation when ammonium was provided at the highest concentration (50 mM) tested (Figure 5B). In controlled sexual crosses on ESAL medium, both mutants could be used as either donor or recipient.

We investigated the role of amtA in morphological transitions and nutrient utilization through qualitative preliminary microscopic examination of spore germination and hyphal development of WT H327 and ΔamtA-3 mutant after 24 h of incubation on OMM agar with different carbon or nitrogen sources. Both strains exhibited comparable germination patterns on sucrose-containing medium supplemented with ammonium sulfate (Figure 6A,B). The substitution of sucrose with maltose resulted in similar hyphal elongation patterns between WT and mutant strains, though the overall biomass appeared reduced compared to that in sucrose-grown cultures (Figure 6C,D). Notably, galactose as the sole carbon source promoted enhanced hyphal elongation and branching in both strains, with no apparent difference between the WT and ΔamtA mutant (Figure 6E,F). Varying the nitrogen revealed subtle phenotypic differences between the WT and mutant strains. Under nitrogen starvation conditions (sucrose without nitrogen), both strains showed restricted hyphal development with increased spore aggregation (Figure 6G,H). When ammonium sulfate was replaced with amino acids as the sole nitrogen sources, distinct morphological responses were observed. Proline supported moderate hyphal growth in both strains with no discernible differences (Figure 6I,J). However, the use of cysteine (Figure 6K,L) and arginine (Figure 6M,N) as nitrogen sources resulted in markedly reduced spore-to-hypha transition in both WT and mutant strains, with predominantly ungerminated spores visible after 24 h. Strikingly, under complete nutrient deprivation (no carbon or nitrogen source), the ΔamtA-3 mutant displayed enhanced hyphal formation compared to WT, producing longer and more branched hyphae despite the absence of exogenous nutrients (Figure 6O,P).

When mutants ΔamtA-3 and ΔamtA-35 were inoculated onto Golden Delicious apples, their aggressiveness was similar to that of H327 (Figure 7A). However, when the mutants were inoculated onto potted U. americana saplings kept in a greenhouse, they were found to be significantly less aggressive (p < 0.05) than WT H327 in all inoculation trials (Figure 7B–E). Saplings inoculated with ΔamtA mutants developed foliar symptoms at a slower rate than those inoculated with WT strain H327. In one experiment, in which O. ulmi strain Q412T was included, its aggressiveness was comparable to that of mutant ΔamtA-3 (Figure 7D). The ΔamtA mutants always induced significantly less disease than WT H327, whether they were inoculated at low (6.25 × 10³ yeast cells) or high (50 × 10³ yeast cells) dose onto the host.

4. Discussion

The current pandemic of DED, which was first detected in the 1960s in southern England [59,60], is caused by the highly aggressive O. novo-ulmi [2]. The pathogen includes two subspecies designated as novo-ulmi and americana [61], and these have rapidly replaced the less aggressive O. ulmi, which was responsible for the first pandemic. The global expansion of O. novo-ulmi is threatening natural elm populations in the northern hemisphere [1,62,63], elms that were introduced to other regions of the world where they are not native [64], as well as species of Zelkova in their native range [65]. As in many fungal pathogens of plants, aggressiveness in DED fungi is under polygenic control, as demonstrated initially by the phenotypic analysis of progeny from controlled genetic crosses between O. ulmi and O. novo-ulmi [4]. These results have since been supported by in silico observations, where more than 1700 of the 8642 genes annotated in the O. novo-ulmi nuclear genome are predicted to code for proteins that may be involved in host–pathogen interactions [7].

In the late 1990s, we reported the results of Mendelian and molecular analyses of strain AST27 [30], initially believed to be an O. novo-ulmi ssp. novo-ulmi strain with unusually low aggressiveness towards U. procera [66]. We found that the lower aggressiveness of strain AST27 was associated with introgressed O. ulmi DNA in chromosome I; this was supported by its co-segregation with O. ulmi-type RAPD markers in the progeny of controlled crosses between AST27 and the highly aggressive O. novo-ulmi ssp. novo-ulmi strain H327 [30]. We postulated that the introgressed DNA included one locus, labeled pat1, that contributed to aggressiveness. We subsequently concluded that a quantitative trait locus (QTL), associated with mycelial growth rate at 28 °C, was also tightly linked to pat1 [34]. Concurrently, we successfully completed a 46.9 Kb-long chromosome walk based on cloned RAPD amplicons OPK3₁₀₅₀ and OPJ20₅₄₀ linked to pat1, as reported herein.

Based upon sequence comparisons to a high-quality reference genome for O. novo-ulmi ssp. novo-ulmi H327 [7], we are now confident that the portion of chromosome I that was cloned and analyzed by Sanger sequencing contains a total of 17 genes. However, the two sequenced regions are not contiguous and are separated by a stretch of 157.5 kb containing 41 annotated genes. Based on these analyses, there are 58 candidate genes for pat1 distributed over the 201 kb of O. ulmi DNA bounded by the OPK3₁₀₅₀ and OPJ20₅₄₀ markers. However, the number of candidates genes is likely higher, as we do not know the total length of O. ulmi DNA introgressed into this portion of the AST27 nuclear genome located outside the boundaries set by ONUg0228 and ONUg0284.

Nonetheless, the 17 genes identified by chromosome walking provide valid targets for identifying pat1. We focused our attention on ONUg0282, predicted to encode an ammonium transporter/methylammonium permease, for the following reasons: (1) ONUg0282 was one of three genes with an ortholog in PHI-Base (PHI:2710; Reduced virulence), namely the mepB gene shown to contribute to the aggressiveness of the plant pathogen C. gloeosporioides [31]; (2) ONUg0282 and several other genes in the cyclic AMP (cAMP)-dependent protein kinase A (PKA) pathway [33,67] were highly expressed during the yeast–mycelium (Y-M) transition in O. novo-ulmi grown under axenic conditions [29]; and (3) ONUg0282 was also expressed in O. novo-ulmi colonizing susceptible U. americana saplings, whereas transcripts were barely detectable in the transcriptomes of the saprotroph O. quercus under the same experimental conditions [18,47]. Based on these observations, we hypothesized that ONUg0282 might be implicated in both theY-M transition and strain aggressiveness in O. novo-ulmi.

Phylogenetic analyses of the predicted aa sequence encoded by ONUg0282 confirmed it belongs to the Amt/Mep family of transmembrane proteins and is part of a clade that includes the S. cerevisiae Mep2 protein [68,69,70]. Members of this clade of ammonium transporters are also known as transceptors, since they act as receptors to regulate downstream signaling pathways [71]. The Mep2 transceptor occurs as a stable trimer in the cytoplasmic membrane of prokaryotes and eukaryotes. Each monomer has a conserved core of typically 400 to 450 aa and contains 10 to 12 transmembrane helices (TMHs) with N_out–C_in topology. It harbors a region of 22 residues located in the cytoplasmic C-terminal region (CTR), immediately downstream of the last TMH, which is highly conserved in bacteria, archaea, fungi, and plants [72,73]. Based on X-ray crystallography of the reference AmtB protein of Escherichia coli, the TMHs in each monomer form a right-handed helical bundle around a central channel through which ammonium is transported [74,75]. Molecular features of the predicted protein encoded by ONUg0282 support its function as an ammonium transceptor in DED fungi; we observed the presence of conserved histidine residues, an N_out-C_in topology consisting of 11 THMs, and the occurrence of a conserved region in the CTR portion of the protein.

The orthologs of ONUg0282 in O. ulmi W9 and introgressantAST27 are predicted to encode proteins with a topology similar to that of ONUg0282, since the additional motif composed of seven residues at positions 513–519 is distant from the last alpha helix (positions 396–424). The high level of expression by ONUg0282 when O. novo-ulmi H327 is grown on OMM agar without nitrogen is typical of the response described in other fungal species [68,71,76]. Whether the O. novo-ulmi and O. ulmi alleles for amtA respond similarly to nitrogen starvation remains to be determined by additional studies; these may include the phenotyping of H327 × AST27 F1 and backcross progeny strains.

Three amt/mep genes occur in the yeasts S. cerevisiae [67,69] and S. pombe [77]. However, four amt/mep genes have been reported in most filamentous Ascomycota studied so far. The current study determined that the genomes of DED pathogens O. ulmi and O. novo-ulmi harbor five genes encoding ammonium transporters; the search for paralogs of ONUg0282 in these genomes led us to identify four additional ammonium transporter-encoding genes, designated amtB to amtE. Examination of the transcriptomes of O. novo-ulmi H327 from previous studies confirmed that the five amt genes were expressed in vitro [29,57] and in planta [18,47]. According to phylogenetic analysis, the AmtB transporter encoded by ONUg1681 belongs to the S. cerevisiae Mep2 clade, which also includes the MepA and MepB transporters from the anthracnose-causing fungus C. gloeosporioides, as well as two ammonium transporters from the dimorphic human pathogen Sporothrix schenckii, the type species for the genus Sporothrix within the Ophiostomatales [78]. The occurrence of paralogs from the same species in the S. cerevisiae Mep2 clade is not unusual. For instance, it has also been reported for Aspergillus flavus, A. oryzae, and Fusarium fujikuroi [70]. The XP_01659096 NCBI sequence of S. schenckii showed strong identity with the Amt2 sequence (NP_973634) within the set of A. thaliana Amt/Mep sequences used as an outgroup to root the phylogenetic tree (Figure 3A). When the S. schenckii and A. thaliana sequences were queried against the NCBI Clustered NR database, the top 100 sequences retrieved by the blastp algorithm were either all fungal sequences or all plant sequences, respectively. We, therefore, conclude that the findings of our phylogenetic analyses are not an artefact but rather suggest that the two sequences originate from the same bacterial Amt sequence [79,80].

Genes encoding transporters occur as members of large families [81]. In fungi, the occurrence of multiple genes encoding ABC transporters in the wheat pathogen Mycosphaerella graminicola [82] or inorganic phosphate (Pi) transporters in the ectomycorrhizal symbiont Hebeloma cylindrosporum [83] has been associated with functional redundancy. On the other hand, functional diversity was suggested to account for the presence of multiple ammonium transporter-encoding genes with different characteristics. For instance, in S. cerevisiae, the Mep1 and Mep3 transporters differ in their affinity for NH4⁺, whereas transceptor Mep2 can also sense nitrogen limitation and activate the MAPK and cAMP/PKA cascades [69,71,84]. Likewise, the three ammonium transporters characterized in the endomycorrhizal fungus Geosiphon pyriformis were found to differ in their cellular location and physiological role [85]. Whether differences in the expression of O. novo-ulmi H327 expression profiles of amt genes [18,29,47,57] result from functional diversity remains to be verified by thorough functional studies of all five genes.

The results observed in vitro and in planta for a set of five RNAi mutants of O. novo-ulmi strain H327 were encouraging. Mutant strain TF2, selected on the basis of its low level of ONUg0282 transcripts, was significantly less aggressive towards the U. americana × U. pumila hybrid clone 2213 than strain H327. The high level of aggressiveness observed for the pat1-m strain A3P11 did not permit conclusive results when assessing the aggressiveness of the complementation mutant A3P11/8 carrying a H327-type allele for ONUg0282. In a previous study [30], strain A3P11 was considered moderately aggressive towards U. procera. The observed discrepancy between these studies is likely due to the fact that U. procera is more resistant to DED than the U. americana × U. pumila hybrid clone 2213 [55]. Following the accidental loss of the initial set of RNAi mutants, new knockdown mutants were produced, and one of them, AmtA-B, was inoculated onto U. americana, on which, it induced 100% crown symptoms [47]. This was unexpected, as the mutant had retained low (1.8%) residual gene transcription activity in vitro. The observed host response was attributed to the high susceptibility of U. americana to the pathogen [47]. The higher residual gene transcription occurring in planta and the higher inoculum load used in the previous study (50 × 10³ vs. 1.75 × 10³ yeast cells) were considered to be contributing factors.

The targeted mutagenesis of DED pathogens may be accomplished by OSCAR-type vectors [13], or CRISPR-Cas9 gene editing [49]. We chose CRISPR-Cas9 gene editing, since (1) it does not require the construction of deletion cassettes for each gene targeted, as is the case for OSCAR-based transformation, and (2) custom RNA oligonucleotides required for the assembly of RNA guides are commercially available. Screening of potential mutants was greatly facilitated by the co-transformation of an adenine-requiring auxotroph of O. novo-ulmi strain H327 with RNA guides for the ade7 (ONUg7434) and amtA (ONUg0282) loci and by selecting white colonies formed by Ade⁺ prototrophs for further verification of the nucleotide sequence of the amtA locus. Sanger sequencing confirmed that the target gene of ΔamtA mutants was altered in the same region between nucleotides 734 and 747 but differed in the number of nucleotides that had been lost (only nucleotide 743 in ΔamtA-35 and nucleotides 734 to 747 in ΔamtA-3). Both mutants displayed frameshift mutations that strongly altered the predicted amino acid sequence of the AmtA protein in which only five of the 11 alpha helices present in the wild-type protein remained intact. Nevertheless, the mutants retained several wild-type phenotypic traits: they produced mycelium on OMM agar without nitrogen, tolerated the presence of relatively high doses of sodium chloride, were male- and female-fertile, and underwent mycelium–yeast transition under growth conditions that support the transition in WT strains.

While both mutants induced necrotic lesions on Golden Delicious apples, they showed decreased aggressiveness towards U. americana saplings in controlled greenhouse experiments. It is noteworthy that, in the case of CRISPR-Cas9 mutants, the quantity of inoculum injected into elm saplings (6.25 × 10³ yeast cells or 50 × 10³ yeast cells) did not influence significantly the outcome of the interaction. Therefore, the results from the controlled inoculations of U. americana saplings support our initial hypothesis that a fully functional ONUg0282 locus is an important contributor to aggressiveness in O. novo-ulmi. Ammonium transporters may contribute to the fitness of fungal pathogens that colonize the plant vascular system, since the concentration of nitrogen in this environment is limiting. For instance, the concentration of total nitrogen in the sap of healthy U. americana was found to vary from 1.02 to 1.65 mM throughout the summer [86], which would place this species in the lower portion of the range for tree species [87]. Therefore, inactivation of the O. novo-ulmi amtA gene may have a more pronounced effect on aggressiveness when the pathogen develops in nitrogen-poor elm tissues, compared to Golden Delicious apple fruit tissue in which nitrogen concentrations are 15 to 40 times higher than in elm xylem [88,89].

To our knowledge, this is the first time a gene encoding an ammonium transporter has been shown to contribute to the aggressiveness of a tree pathogenic fungus. This finding was not unexpected, in light of previously published studies reporting the role of Amt/Mep-encoding genes in the aggressiveness of several plant pathogenic fungi. These include the mepB gene in C. gloeosporioides [31], gene mep2 in Fusarium graminearum, causal agent of Fusarium head blight of wheat and barley [32], and genes ump2 and mepA in the smut fungi U. maydis and Microbotryum violaceum, respectively [33]. However, phylogenetic analyses suggest that the O. novo-ulmi ortholog of the C. gloeosporioides mepB entry in PHI-Base is ONU1681 (AmtB) rather than ONUg0282 (AmtA). Transcriptomics results in planta support a role for ONU1681 in pathogenesis, as it was the most expressed Amt-encoding gene when O. ulmi and O. novo-ulmi were inoculated onto susceptible U. americana elm saplings [18,47].

In addition to their role in pathogenicity, Amt/Mep-encoding genes are implicated in fungal morphogenesis. Ammonium transceptors have been shown to modulate the Y-M transition in several dimorphic fungal species. For instance, the activation of Amt/Mep-encoding genes in response to nitrogen starvation results in the formation of pseudohyphae in S. cerevisiae [67] and hyphae in the fission yeast S. pombe [77], in the human pathogens Candida albicans [90] and Cryptococcus neoformans [91], and in the plant pathogens U. maydis and M. violaceum [33]. The unexpected enhancement of mycelial formation in ΔamtA mutants under complete nutrient starvation suggests that AmtA participates in regulatory circuits extending beyond its primary role in ammonium transport. In O. novo-ulmi, the Y-to-M transition involves differential expression of over 22% of the genome, with genes involved in cAMP/PKA and MAPK signaling pathways being highly expressed during this morphological switch [29]. The enhanced mycelial growth observed in ΔamtA mutants under nutrient deprivation may be the result of dysregulation in these signaling cascades. Understanding these regulatory connections may reveal how O. novo-ulmi coordinates nutrient sensing with morphological adaptations critical for survival in the nutrient-limited environment of elm xylem vessels.

5. Conclusions

The current study allowed us to identify the predicted Mep2 transceptor gene ONUg0282 as a major contributor to the high aggressiveness of O. novo-ulmi ssp. novo-ulmi when colonizing U. americana saplings. Available experimental data do not support the conclusion that ONUg0282 is indeed the pat1 locus, since we did not systematically test all candidate genes in this region of the O. novo-ulmi chromosome I. Rather, the results reported herein open up new research perspectives on nitrogen metabolism in the DED pathosystem [86,92]. Work is now underway to recover CRISPR-Cas9 mutants for all five Amt-encoding genes detected in the genome of O. ulmi and O. novo-ulmi. They will be used to assess the role of these genes in the Y-M transition and aggressiveness of DED fungi through exhaustive phenotyping of single- and multiple-gene deletion mutants.