Two isolates of C. jacksonii, HRS10 and HRI11, which previously had draft de novo genome assemblies (Green et al., 2016), underwent additional sequencing using the Oxford Nanopore platform, followed by de novo assembly and annotation. HRS10 exhibits sensitivity to multiple fungicide classes, whereas HRI11 shows multidrug resistance to demethylation inhibitor fungicides as well as other fungicide classes including plant growth regulators (Green et al., 2016; Hulvey et al., 2012). The genome reads generated by Green et al. (2016) for these isolates are accessible on GenBank under accession numbers LNGN00000000 for HRS10 and LNKV00000000 for HRI11.

Mycelia stored previously were incubated in potato dextrose broth (PDB) (Difco Laboratories, Detroit, MI) in a 50ml Falcon tube for 4 days, followed by sub-culturing in a 250 ml flask with PDB for an additional 4 days. Each isolate’s mycelia were then divided into six 2 ml centrifuge tubes. After brief centrifugation, any remaining media was discarded. Genomic DNA was extracted using a modified cetyl trimethyl ammonium bromide (CTAB) method (Cubero et al., 1999), with an incubation temperature of 65°C maintained for 30 minutes. Following the initial washing step with chloroform: isoamyl alcohol (24:1 v/v), every two upper aqueous phases were combined into a single tube to concentrate the final product. To ensure gDNA purity, precipitation was achieved using 100% ethanol. DNA concentration was quantified using the Qubit dsDNA HS Assay (ThermoFisher SCIENTIFIC), and sample purity was assessed via gel electrophoresis and by measuring the OD₂₆₀/OD₂₈₀ ratio with a Nanodrop.

Genomic DNA was prepared for sequencing using the ONT SQK–RAD004 genomic rapid sequencing kit (Oxford, UK). Genomic DNA was fragmented using ONT Fragmentation Mix. To achieve this, 500ng of genomic DNA was gently mixed with 2.5 µL Fragmentation Mix in a PCR tube, then incubated at 30°C for 1 minute followed by 80°C for 1 minute. The tubes were subsequently cooled on ice. Next, the Rapid Adapter was ligated to the DNA ends by gently mixing 1 µL of Rapid Adapter with the DNA ends and incubated for 5 minutes at room temperature. The DNA library was then kept on ice until loading into the flow cell.

For flow cell preparation and library loading, reagents for priming and sequencing were thawed and mixed at room temperature and immediately transferred to ice upon thawing completion. The flow cell priming mix was prepared by pouring 30 µL of Flush Tether into the Flush Buffer tube, gently mixed by pipetting, and 800μl of the mix loaded into the flow cell through the priming port. After a 5-minute wait, the SpotON sample port cover was lifted and left open, and 200μl of priming mix was added through the priming port.

In a new tube, 34μl sequencing buffer, 25.5 µL of loading beads, 11μl of DNA library, and 4.5μl of nuclease-free water were combined. The mixture was added dropwise to the flow cell through the SpotON port. After adding the sample, the SpotON port cover was replaced, and the priming port was closed. The MinION lid was then replaced, and sequencing was initiated. The sequencing and base-calling processes were managed using the MinKNOW software. Base-called reads were subsequently downloaded for further analysis.

The Illumina HiSeq2000 and Pacbio CLR reads for HRS10 and HRI11 were retrieved from a public database (Green et al., 2016). Raw Nanopore ONT reads were generated after 12 hours of sequencing. Quality control of the raw genomic reads was performed using FastQC v0.11.5 (Andrews et al., 2015). Sequence duplication for Illumina reads were collapsed into a single copy using Tally-15-065 (Richard et al., 2016). Trim-galore v0.4.2 (Krueger, 2015) was utilized to detect and trim adapter contamination as well as low-quality bases. The Canu v1.8 pipeline (Danecek et al., 2011) with default parameters was employed to generate corrected sequence consensus and assemble Nanopore and PacBio reads.

The assembled contigs underwent polishing using Illumina paired-end reads mapped with bwa v0.7.17 (Cingolani et al., 2012) and Pilon v1.23 (Emms and Kelly, 2019). Redundans v0.14a (Nattestad and Schatz, 2016) was applied to reduce heterozygous regions by identifying and selectively removing alternative heterozygous contigs. Genome statistics were generated using QUAST v5.0.2 (Dobin et al., 2013). The completeness of the genome assemblies was assessed using BUSCO v4.0.4 (Simão et al., 2015),which compares the assemblies to universal single-copy orthologs.

The genomes of the two C. jacksonii assemblies were annotated using the Funannotate pipeline v1.8.15 (Palmer and Stajich 2020). Genome repetitive elements were soft-masked using RepeatMasker v4.1.6 (Smit et al., 2015) and RepeatModeler 2.0.5 (Smit and Hubley, 2015) with default parameters. Ab initio gene models were predicted using two protein predictors: AUGUSTUS 3.5.0 (Stanke et al., 2006) and GeneMark-ES (Borodovsky and Lomsadze, 2011). Due to licensing restrictions, GeneMark-ES was run separately, and its predictions were manually integrated. For AUGUSTUS, Funannotate’s fungal training set and default settings were applied.

Transcript gene models were predicted by PASA 2.5.3 (Haas et al., 2008) using de novo transcript assemblies generated by Trinity 2.8.5 (Grabherr et al., 2011)under default parameters. Protein-coding gene models were predicted by protein alignment to the genome using Diamond 2.1.8 (Buchfink et al., 2014) and polished by Exonerate 2.4.0 (Slater and Birney, 2005). EVidenceModeler 1.1.1 (Haas et al., 2008) was employed to build consensus gene models, filtering by length, spanning gaps, and considering transposable elements from both ab initio and evidence-based gene models with default weighting. Functional annotation of the gene models was performed using sequence similarity blast (Altschul et al., 1990) against several databases, including Pfam (v36.0) (Mistry et al., 2021), UniProt DB version 2023_05 (Bateman, et al., 2023), InterProScan5 (v97.0) (Jones et al., 2014), BUSCO (v 5.6.0) (Simão et al., 2015), EggNOG (v6.0) (Hernández-Plaza et al., 2023), MEROPS (v12.0) (Zhang et al., 2018), and Carbohydrate-Active enZYmes (CAZy) Database (Cantarel et al., 2008). Protein-coding genes were additionally annotated by their protein domains using InterProScan5. The Phobius (Käll et al., 2004) webserver predicted transmembrane topology and signal peptides. Secondary metabolite gene clusters were predicted using the antiSMASH v7 webserver (Blin et al., 2023). Gene Ontology (GO)terms (Kanehisa et al., 2016), eukaryotic orthologous groups, and Kyoto Encyclopedia of Genes and Genomes (KEGG) metabolic pathways (Yu et al., 2012) were used for gene classification. Lastly, the genome’s tRNA and anticodons were predicted using tRNAscan-SE (v2.0.12) (Chan et al., 2021).

Several analytical pipelines were employed to compare the structural differences between the two isolates. Whole-genome alignment and visualization of insertions/deletions in aligned regions were conducted using MUCmer from the MUMmer sequence alignment v4.0.0 package (Marçais et al., 2018). To assess structural variations (SVs), copy number variations (CNVs), and single nucleotide polymorphisms (SNPs), HRI11 reads were aligned to the HRS10 genome using Bwa v0.7.17 (Li and Durbin, 2010). Samtools v1.18 facilitated conversion to various file formats for subsequent variant discovery analysis (Li et al., 2009). All tools were run with default parameters unless stated otherwise.

Structural variations were detected using Sniffles2 (Smolka et al., 2024) and DELLY (Rausch et al., 2012), with annotation provided by SURVIVOR_ant v0.0.1 (Sedlazeck et al., 2017). CNVs discovery was complemented by CNVpytor (Panda et al., 2024) and the Control-freeC pipeline (Boeva et al., 2012). SNPs were identified using FreeBayes v1.3.8 (Garrison and Marth, 2012) and filtered using VCFtools v0.1.16 (Danecek et al., 2011) to remove indels, low coverage and/or quality sites. Filtered SNPs were analyzed and annotated by Snpeff v4.3 (Cingolani et al., 2012) to predict the coding effects of genetic variations. OrthoFinder (Emms and Kelly, 2019) provided comparative genomics statistics and phylogenetic orthology inference. Functional comparison was facilitated by the Funannotate pipeline. All tools were run with default parameters unless stated otherwise.

The resistome (Wright, 2007) refers to the full collection of genes in an organism that confer resistance to antimicrobial agents, a concept originally developed in bacterial systems and later adapted to fungi for identifying antifungal resistance determinants. In our study, we employed two fungal resistome databases, MARDy (Mycology Antifungal Resistance Database) (Nash et al., 2018) and AFRbase (Antifungal Resistance Database) (Jain et al., 2023), to compare resistance genes in our isolates.

To investigate the molecular mechanisms underlying DMI resistance, we performed transcriptomic profiling of two C. jacksonii isolates, HRS10 and HRI11 under both untreated and propiconazole-treated conditions (0.1 µg ml^-1 for 1 hour). For each isolate and condition, two independent biological replicates were used, yielding a total of eight RNA-seq libraries. Detailed methods for RNA extraction, cDNA library preparation, and sequencing are described in a previous publication (Sang et al., 2018). Each library generated over 75 million paired-end reads using the Illumina HiSeq 2000 platform.

Initial quality control of raw RNA-seq reads was conducted using FastQC v0.11.5 (Andrews et al., 2015), and quality and adapter trimming was performed using Trim-galore v0.6.10 (Krueger, 2015) with settings identical to those used for genomic reads. Given our focus on resistance mechanisms, all RNA-seq reads were aligned to newly assembled reference genome HRI11 using the STAR 2.7.11b (Dobin et al., 2013), and gene counts were quantified using HTSeq 2.0 (Putri et al., 2022). DESeq2 (Love et al., 2014) was employed for count normalization and detection of differentially expressed genes (DEGs) between two C. jacksonii isolates before and after propiconazole treatment. A gene is considered significantly differentially expressed if its adjusted P-value was < 0.05 as calculated by DESeq2.

Annotation data packages and databases for C. jacksonii were constructed using AnnotationForge with SQLite. GO enrichment and KEGG pathway enrichment analyses for DEGs were performed using enrichGO and enrichKEGG functions from the clusterProfiler v3.16.0 R package (Yu et al., 2012). These enrichment analyses were performed based on the custom annotation package created for this study.

Genome assemblies for HRS10 and HRI11 were produced using a hybrid sequencing approach, combing approximately 8 million previously published reads (200x coverage) from Illumina and PacBio platforms with newly obtained long-read data from the Oxford Nanopore MinION ( Table 1 ). The final assembly size for HRS10 was 39.6 Mb, comprising 100 scaffolds and an N50 of 1.39 Mb. In comparison, HRI11 yielded a larger cumulative assembly of 43.6 Mb across 51 scaffolds with an N50 of 1.57 Mb. Both assemblies showed significant improvements in contiguity and quality, with over 99.1% of assembled sequences incorporated into scaffolds exceeding 50 kb in length. Assembly completeness was evaluated using Benchmarking Universal Single-Copy Ortholog (BUSCO), revealing that 98% of single copy orthologs were present in both genomes, indicating high assembly integrity. Genome annotations using Funannotate identified 12,161 and 12,831 protein-coding genes in HRS10 and HRI11, respectively. Annotation datasets with a BUSCO transcriptome completeness score of 98% were obtained, indicating that the vast majority of conserved fungal genes were successfully identified and annotated. The high score reflects the high quality and completeness of the genome annotation. Additionally, 131 transfer RNA (tRNA) comprising 22 anticodons were predicted in HRS10, while HRI11 featured 130 tRNAs.

Compared to previous assemblies based solely on Illumina and PacBio reads, the hybrid assemblies showed substantial improvements, including fewer and smaller assembly gaps, larger contigs, and a 2-fold increase in both N₅₀ values and the longest scaffold lengths. Improved repeat filtering also reduced repetitive content, minimizing the potential for false-positive results in subsequent analyses.

Repetitive sequences, which often are a significant challenge in eukaryotic genome assemblies, complicating both assembly accuracy and functional prediction. To enhance genome quality, approximately 10% of these repetitive sequences were masked using the RepeatMasker program in both assemblies. As shown in Figure 1 , the core genome shared between the two isolates consists of 10,797 orthogroups, representing 94.8% of all predicted genes. In addition to the core set, HRS10 and HRI11 possess 643 and 653 strain-specific genes, respectively ( Supplementary Table S1 ). The majority of these unique genes encode hypothetical proteins with unknown functions, underscoring potential isolate-specific adaptions or functional divergence.

The observed differences in genome size and GC content may reflect natural intra-species variation, isolate-specific genomic features, or potential sequencing and assembly artifacts. Nonetheless, phylogenetic analysis strong evidence that that both isolates belong to the same species, C. jacksonii, supporting their suitability for comparative genomic and transcriptomic analyses.

Genome structural comparisons were performed using dedicated software to identify structural variation, copy number variation, and single nucleotide variation. Structural variations (SVs), which involve alterations in large segments of DNA, are significant drivers of genomic diversity. To analyze SVs between isolates HRS10 and HRI11, we employed DELLY and Sniffles2 pipelines with PacBio long reads. These pipelines revealed similar structural changes. Notably, HRI11 showed inversions on scaffolds 2 and 7 ( Figure 1b ), where these regions contained nine gene inversions. These genes include nucleoside triphosphate hydrolase (NTPases), two alcohol dehydrogenases, a member of the tetratricopeptide-like helical domain superfamily, and five hypothetical proteins. Insertion events affected two genes, causing a frameshift in the multidrug resistance gene MDR1_3. Moreover, the NUD1 gene (leucine-rich repeat protein) had a breakpoint in HRI11. In addition, HRI11 exhibited deletions in two genes, FSH3 (serine hydrolase) and FUN_011878 (hypothetical protein), and FUN_011921 (hypothetical protein) was duplicated.

CNVs between the two isolates, HRI11 and HRS10, were identified using both the CNVpytor and Control-FreeC pipelines, with cross-referencing to minimize false positives. HRI11 exhibited increased copy numbers for 31 genes and decreased copy numbers for 29 genes compared to HRS10 ( Supplementary Table S2 ). Of the genes with increased copy numbers in HRI11, more than half are tRNA genes, including multiple copies of tRNA-Lys, tRNA-Gly, tRNA-Arg, and others. Additionally, 12 mRNA-encoding genes showed heightened copy numbers, with annotations including 10 hypothetical proteins, the global transcription regulator ge1_2 (SGE1), and NADH:ubiquinone oxidoreductase subunit 5 (nad5). In contrast, all genes with decreased copy numbers in HRI11 are protein coding genes, including two ATP-dependent RNA helicases, DED1 and CHL1.

We identified 96,326 single nucleotide polymorphisms (SNPs) in the HRI11 isolate using FreeBayes, and these were analyzed for their effects using SnpEff. The majority of SNPs were located in intergenic regions (62.7%), followed by upstream (14.4%) and downstream (20.0%) regions ( Figure 2 ). Less than 3% of the SNPs (3,215) were found within coding regions. Among these coding SNPs, approximately half (1,620) were synonymous, while the rest comprised 1,557 non-synonymous and 38 nonsense mutations. Analysis with Bcftools corroborated these findings, helping to refine the initial results. We identified 33 non-synonymous SNPs that overlapped between FreeBayes and Bcftools, which could have significant impacts on gene function. These SNPs predominantly caused start codon losses and premature stop codon mutations. While many affected genes are annotated as hypothetical proteins, their roles often relate to cellular metabolism and signaling pathways. Notably, a SNP was also found in the global transcription regulator SGE1 within the HRI11 strain, underscoring its potential importance in observed genetic variations.

RNA-seq analysis was conducted on two C. jacksonii isolates before and after propiconazole treatment. The analysis mapped RNA-seq reads to 96.1% of HRI11 gene models and 96.9% of HRS10 predicted genes. Genes showing significant differential expression were identified based on criteria of log2fold change > 1 and adjusted P-value < 0.05, calculated using DESeq2. Following propiconazole exposure, 37 genes in isolate HRS10 were upregulated, while 17 genes were downregulated. Among the upregulated genes in HRS10 after DMI treatment, in ion transport pathways was observed (P = 0.0025). In contrast, HRI11 exhibited 17 upregulated genes and 14 downregulated genes post-DMI treatment ( Figure 3 ). Notably, gene induced by DMI treatment in HRI11 were enriched in processes related to ergosterol and lipid biosynthesis. The differences following DMI treatment between HRS10 and HRI11 suggest potential mechanisms contributing to resistance in HRI11.

In both isolates, six genes showed elevated expression following propiconazole treatment, including three ergosterol biosynthesis genes (ERG6, ERG11, and ERG24), the ABC transporter ShatrD, a potential histidine kinase/Hsp90-like ATPase, and a serine-type peptidase. Notably, the expression levels of these ergosterol biosynthesis genes were slightly higher in isolate HRI11 compared to isolate HRS10. Additionally, we observed the upregulation of the previously reported ABC transporter ShatrD (Sang et al., 2015) in both isolates after propiconazole treatment. However, another ABC transporter, ShPDR1 (Sang et al., 2018), was induced specifically in the resistant isolate HRI11 in the presence of propiconazole. Similarly, ERG5 and the previously reported ShCYP561 were induced exclusively in the resistant isolate following propiconazole treatment. Regarding down-regulated gens, no overlap was found between HRS10 and HRI11. The genes inhibited by DMI in HRI11 were enriched in the mycotoxin biosynthetic process (P = 0.002), suggesting an energy allocation strategy that may contribute to resistance mechanisms.

Gene expression patterns were compared between two isolates of C. jacksonii under normal conditions and after propiconazole treatment to understand their differences. Under normal growth conditions, isolate HRI11 exhibited 245 upregulated genes enriched in extracellular regions, serine peptidase activity, and hydrolase activity ( Figure 4a ), while 110 genes were downregulated and enriched in secondary metabolic processes ( Figure 4b ) compared to the sensitive isolate HRS10. These transcriptomic patterns may reflect shifts in cellular priorities or physiological states between the two isolates. Propiconazole treatment induced additional expression differences between the isolates, with 322 genes upregulated and 146 genes downregulated. GO-enrichment analysis showed similar results compared to non-DMI treatment conditions.

To identify genes specifically responsive to propiconazole treatment, we further analyzed genes that were uniquely differentially expressed in HRI11 after DMI treatment. After filtering, we identified 184 overexpressed genes and 76 suppressed genes that responded to propiconazole in HRI11 ( Figure 5 ). KEGG pathway enrichment analysis revealed that those overexpressed genes were significantly enriched in transporters [Ko02000].

The Clusters of Orthologous Groups of Proteins (COGs) database (Galperin et al., 2021) was used to classify protein activities phylogenetically, providing a comprehensive comparison between different isolates. Post-annotation, approximately 29% of protein functions remain unidentified. Known functional families predominantly include: (G) Carbohydrate transport and metabolism (28.5% in HRS10, 28.8% in HRI11), (O) Posttranslational modification, protein turnover, chaperones (7.5% in HRS10, 7.3% in HRI11), (Q) Secondary metabolite biosynthesis, transport (6.8% in HRS10, 6.7% in HRI11), and catabolism, (E) Amino acid transport and metabolism (5.6% in HRS10, 5.9% in HRI11), and (U) Intracellular trafficking, secretion, and vesicular transport (5.2% for both strains) ( Figure 6 ). Despite its larger genome size, HRI11 typically possesses either more or the equivalent number of proteins in all COG categories compared to HRS10, except for category (V) Defense mechanisms. HRS10 has 59 proteins in this category, while HRI11 has 58 proteins, a finding that prompts further investigation, as transporters are often involved in fungicide resistance.

Both major facilitator superfamily (MFS) transporters and ATP-binding cassette (ABC) transporters play crucial roles in maintaining cellular homeostasis, influencing various physiological processes such as nutrient uptake, drug resistance, and toxin elimination. Enhanced drug efflux via MFS and ABC transporters has been associated with DMI resistance in diverse fungi (Sirim et al., 2009; Ziogas & Malandrakis, 2015).

HRS10 harbored 353 MFS transporters, whereas HRI11 possessed 375 MFS transporters as identified by InterProScan (Jones et al., 2014), with no SNPs detected among them. The response of MFS transporters to DMI treatment varied between the isolates: in HRS10, two MFS transporters were upregulated and two were downregulated after DMI treatment. In contrast, HRI11 exhibited no significant differential expression of MFS transporters between untreated and DMI-treated conditions, suggesting a limited transcriptional response to the fungicide within this isolate. However, when comparing the two isolates under normal conditions, HRI11 exhibited three upregulated and nine downregulated MFS transporters compared to HRS10, indicating a potentially pre-existing resistance mechanism in HRI11 influencing MFS transporter expression. Post-DMI treatment, HRI11 showed 13 overexpressed and 12 inhibited MFS transporters. The greater number of MFS transporters and their diverse expression patterns in HRI11, particularly under DMI treatment, underscore potential regulatory mechanisms driving differential responses in the two C. jacksonii isolates.

In addition to MFS transporters, HRS10 and HRI11 possessed 65 and 77 ABC transporters, respectively. Two non-synonymous SNPs were identified in two ABC transporters, including ShatrD. Beyond genetic variation, the expression of ABC transporters varied across different conditions. ShatrD was upregulated in response to DMI treatment in both HRS10 and HRI11. Moreover, ShatrD showed higher expression in HRI11 than in HRS10 under both untreated and treated conditions, highlighting its pivotal role in responding to DMI and in resistance mechanisms. ShPDR1 exhibited upregulation, specifically in HRI11 following DMI treatment and maintained higher expression levels in HRI11 compared to HRS10 across all conditions, indicating its potential involvement in baseline resistance in HRI11. Additionally, FUN_007147-T1 (hypothetical protein) showed elevated expression in HRI11 compared to HRS10 regardless of DMI treatment, suggesting its potential involvement in the inherent resistance mechanisms of HRI11. Moreover, MDR1_1 and MDR1_3 were specifically upregulated in HRI11 under DMI treatment.

The Carbohydrate-Active enZYme (CAZy) database complies the enzymes involved in the synthesis, metabolism, and reorganization of complex carbohydrates. Predictions identified 512 and 528 CAZyme-encoding genes in HRS10 and HRI11, respectively ( Supplementary Figure S1 ). The putative CAZymes were assigned into six classes: Auxiliary Activities (AA), Carbohydrate-Binding Modules (CBM), Carbohydrate Esterases (CE), Glycoside Hydrolases (GH), Glycosyl Transferases (GT), and Polysaccharide Lyases (PL). Compared with HRS10, families of GT maintained the same and families of CBM contracted while the other families expanded in HRI11.

Transcription factors (TFs) play pivotal roles in diverse biological processes such as development, metabolism, stress response, and pathogenicity including drug resistance in filamentous fungi (Cacciotti et al., 2024; John et al., 2021, 2022). In HRS10 and HRI11, a total of 367 and 387 TFs were predicted, respectively. More than half of these TFs contain the fungal Zn₂-Cys₆ binuclear cluster domain and/or the fungal-specific TF domain, crucial for regulating drug efflux pump activity (MacPherson et al., 2006). A previous study (Sang et al., 2018) highlighted that HRI11 features an amino acid substitution in the Zn₂-Cys₆ TF ShXDR1, contributing to multidrug resistance by upregulating efflux pump encoding genes such as ShatrD and ShPDR1. Additionally, HRI11 possesses three unique TFs: one MADS-box superfamily TF, one CCHC-type zinc finger TF, and one TF with a Zn₂-Cys₆ fungal-type DNA-binding domain.

The expression level of TFs under different conditions were compared ( Figure 7a ), revealing distinct regulatory patterns in HRI11 following exposure to DMI fungicide. After propiconazole treatment, HRI11 showed no significant change in TFs expression, whereas HRS10 exhibited downregulation of two TFs. Under normal conditions compared to HRS10, HRI11 had four TFs upregulated and two TFs downregulated. Following DMI treatment, HRI11 displayed upregulation of 10 TFs and inhibition of three TFs. These findings highlight a differential regulatory system in HRI11, particularly in response to fungicide stress, potentially influencing its adaptive capabilities and drug resistance mechanisms.

Ergosterol, the primary fungal sterol on the cell membranes, significantly influences membrane integrity, permeability, and fluidity (Rodrigues, 2018). It is also a target for many antifungal medicines (Gulshan & Moye-Rowley, 2007). We investigated ergosterol biosynthesis genes at different levels to understand their role in drug resistance. Both isolates share 26 identical ergosterol synthesis genes ( Figure 8 ). HRI11 possesses an additional ERG12_2 paralog and a non-synonymous substitution in HMG1. The effects of propiconazole on three ergosterol biosynthesis genes ERG6, ERG11, and ERG24 were evaluated in both isolates.

Additionally, cytochrome P450 monooxygenases (CYPs) are widely recognized for their role in detoxifying xenobiotics and contributing to fungicide resistance (Lah et al., 2011). HRS10 and HRI11 harbor 106 and 126 CYPs, respectively. Apart from the higher count of CYPs in HRI11, it also exhibits a non-synonymous SNPs in FUN_010664 compared to HRS10. Previous studies have shown that ShCYP65 is upregulated in HRI11 compared to HRS10 under normal conditions and after DMI treatment, implicating its potential involvement in the fungicide resistance mechanism of HRI11. Similarly, ShCYP561 responds similarly to ShCYP65, showing upregulation after DMI treatment in HRI11. Several other CYPs, including FUN_000674-T2, are implicated in DMI resistance, with elevated expression levels following DMI treatments in both isolates and higher expression in HRI11 compared to HRS10 post-DMI treatment. In HRS10, six CYPs were up-regulated and three were down-regulated after propiconazole treatment, whereas HRI11 exhibited upregulation of seven genes and downregulation of two CYP genes. Compared to HRS10, FUN_010440 in HRI11 was repressed under normal conditions but overexpressed after propiconazole treatment. However, FUN_008349-T1 is not differentially expressed to DMI in HRI11 ( Figure 7b ).

Glutathione S-transferases (GSTs) constitute a diverse family of enzymes crucial for detoxifying various endogenous and xenobiotic compounds, including fungicides. These enzymes facilitate detoxification by catalyzing the conjugation of the tripeptide glutathione to a wide array of substrates, thereby neutralizing their toxicity and promoting their elimination from the cell (Sheehan et al., 2001). Both isolates were found to possess 26 GSTs. Among them, FUN_005805-T1 shows upregulation in HRI11 compared to HRS10 under normal conditions, while FUN_010740-T1 exhibits upregulation in HRI11 after DMI treatment. This differential expression suggests a possible role for these GSTs in augmenting detoxification capacity and potentially contributing fungicide resistance in HRI11.

Additionally, we examined 39 known genes, including Yap1, Ecm1, etc. which are implicated in drug resistance. However, no differences in expression were observed between the two isolates in terms of these genes.

Using a sequence similarity threshold of 60%, we identified 23 resistance genes that showed no differences between the two isolates, including the DMI target gene ERG11/CYP51. However, the beta-tubulin 2 gene, which is the target of methyl benzimidazole carbamate fungicides, exhibited an E198A amino acid substitution in the MDR isolate HRI11 compared to the sensitive isolate HRS10. This mutation, known to confer benzimidazole resistance, has been widely documented in other pathogenic fungi such as Venturia inaequalis, Gibberella zeae, and various C. jacksonii isolates.