The Folini strains under investigation were provided by the Federal Crop Research Institute, city of Torzhok, Tver’ region, Russia. Their virulence was assessed under greenhouse conditions using reference cultivars (Tost 3, Tvertza, A-29) with contrasting susceptibility to Fusarium wilt. Mitcherlich vessels were filled with healthy soil by 2/3 of the height. Next, a pure fungal culture was introduced into each vessel (40 grams per vessel), covered with soil and abundantly watered. The seeds of a reference cultivar were sown on the fifth – sixth day after the inoculation.

The pure culture inoculum of a strain was prepared by growing it on the beer-wort agar-agar medium with subsequent incubation on the oat grain substrate (the grain-to-water ratio of 1 to 1.75). After three, maximum four, weeks’ time, once the substrate had been completely colonized and the macro and microconidia had formed, the pathogen was introduced into the soil.

The severity of the disease in а reference cultivar was evaluated during harvest after the onset of the phase of early yellow ripeness. The DSS (Disease Severity Score) grades ranged from 0 to 3, where 0 stands for a healthy plant, 1 indicates a partial plant browning or stem browning from one side, 2 matches a fully browned plant with bolls, and, finally, 3 corresponds to a fully browned plant that collapses before the formation of bolls. Based on these grades the disease severity index (DSI) was calculated according to the formula generally accepted in the phytopathology field:

where a - is the number of plants with the same DSS; b – is the estimated DSS; A is the total number of plants and K is the highest DSS grade (i.e., grade 3). A strain was considered strongly virulent, if DSI value exceeds 50%, moderately virulent, if the DSI value is in the range from 20% to 50%, and, finally, weakly virulent in those cases, where the DSI ≤ 20%. Three flax cultivars representing contrasted disease susceptibility groups, namely highly resistant A-29, moderately resistant Tvertza and the highly susceptible Tost3 were used to classify Folini strains into strong, moderate and weakly virulent groups.

Vegetative Compatibility groups are determined in accordance with the adapted method first introduced by Puhalla (Puhalla, 1985). The modifications are as follows: to recover nitrogen-nonutilizing (nit) mutants we used PDA supplemented with potassium chlorate (15 g/l) and sucrose (20 g/l) as a chlorate-containing medium.

Fusarium oxysporum strains F200, F365, F418 were grown in Chapek broth medium (30 g/l sucrose, 2 g/l sodium nitrate, 1 g/l potassium monohydrogen phosphate, 0.5 g/l magnesium sulfate, 0.5 g/l potassium chloride, 0.01 g/l ferrous sulfate). DNA was extracted with NucleoSpin Plant II (Macherey-Nagel, Germany) with the addition of RNAse A. DNA sequencing of the F418 genome was conducted at BGI (Hong Kong, China), with 100x coverage by PacBio RS II System and 30x coverage by Illumina HighSeq. Both the remaining 2 isolates i.e., F200 and F365 were sequenced with 30x coverage using Illumina HighSeq (paired-end reads, 150 bp). Hybrid assembly of the F418 genome was performed in three steps: (1) the initial assembly of PacBio contigs was performed with Canu package (v. 1.4) using the default settings (Koren et al., 2017); (2) further polishing and correction of the assembly was done with Pilon software (v. 1.23) (Walker et al., 2014) using sequencing reads generated by Illumina HighSeq; (3) the guided chromosome-level assembly was constructed with MI39 reference genome using RagTag program (version 2.1) (Alonge et al., 2022).

An assembly of genomes sequenced with short reads (Illumina) only includes two steps: contigs assembly with abyss-pe program from the ABYSS package (version 2.1.5) (Simpson et al., 2009), and reference-guided assembly using MI39 reference genome. To ensure the uniformity of data processing the genomes of previously published strains F329, F324, F282, and F287 (Kanapin et al., 2020) were reassembled with RagTag package (Alonge et al., 2022). Besides the genomes of the strains sequenced in (Dvorianinova et al., 2021) (i.e., F456, F476, F482, F483, F525) and newly sequenced strains F200 and F365 were constructed with abyss-pe and RagTag using MI39 as a reference. The quality of every genome assembly was assessed with the QUAST program (version 5.0.2) (Gurevich et al., 2013). Short reads of 12 Folini strains were aligned to the MI39 reference genome using bwa-mem (Li and Durbin, 2009) with default parameters.

To accomplish gene prediction and functional annotation of the proteins, we used BUSCO (v 4.0.4) (Seppey et al., 2019) with parameters “–augustus” and “-l hypocreales_odb10”, respectively. Variant calling for the alignments was completed with NGSEP version 4.0 (Tello et al., 2022) Synteny analysis was done using Satsuma2 program (Grabherr et al., 2010).

Protein orthogroups were identified with silix software (version 1.3.0) (Miele et al., 2011), where the minimum percent of identity to accept blast hits for building families has been set to 0.8. The pangenome accumulation curves were plotted using R vegan package’s specaccum function (Oksanen et al., 2018)

Candidate secreted proteins were identified in three steps. First, secretory signal peptides were predicted using signalP tool (version 5.0) (Nielsen et al., 2019). Then, the TMHMM (v2.0) program was used to keep proteins without a transmembrane domain or with a single transmembrane domain in the N-terminal signal peptide (Krogh et al., 2001). Lastly, PredGPI was applied to remove sequences containing GPI (glycosylphosphatidylinositol) anchors (Pierleoni et al., 2008). Subsequently, the secretomes were screened for CAZymes using a stand-alone version of DBCan2 suite (Zheng et al., 2023) and for effector genes using EffectorP software (version 3.0) (Sperschneider and Dodds, 2022). SIX genes were not accounted for in the BUSCO program’s prediction, leading to their inclusion in the final count. Putative peptidases and peptidase inhibitors were predicted using the MEROPS database (Rawlings et al., 2017).

For the purposes of phylogenetic analysis, EF-1α gene sequences were selected from 53 strains. These include 13 Folini genomes from this study, 20 strains of Folini available from the NCBI repositories, and 19 strains of different F. oxysporum ff. spp. and F. solani, F. graminearum, F. verticelloides, and F. avenaceum were used as an outgroup ( Supplementary Table S1 ). First, the sequences were aligned with CLC Sequence Viewer tool (http://www.qiagenbioinformatics.com). Next, the maximum likelihood phylogeny tree was constructed with the Tamura-Nei model using 1000 bootstrap replicates, as implemented in the IQ-TREE tool (Minh et al., 2020).

A phylogenetic analysis of SIX genes was done using sequences from 13 Folini strains and other special forms of F. oxysporum ( Supplementary Table S2 ). All the sequences were obtained from the NCBI database (https://www.ncbi.nlm.nih.gov/nucleotide/) and aligned with the aforementioned the CLC Sequence Viewer. Again, a consensus tree was built using IQ-TREE (Minh et al., 2020) with bootstrap support (1000 replicates). The following substitution models for each of the trees constructed were in use: SIX1 -TNe+G4, SIX7 - K2P+I, SIX10 - K2P+G4, SIX12 - K2P, SIX13 - K2P.

The whole-genome SNP-based phylogeny (as identified in the 12 Folini genomes) was inferred by MEGAX software (version 10.1.8) (Tamura et al., 2021), where the MI39 genome served as a reference.

Seedlings of the Fusarium wilt-resistant flax cultivar Atalante, as well as the seedlings of the susceptible cultivar LM98, were infected with the fungus strain MI39. RNA isolation from infected and uninfected plants was performed in three independent replicates on days 3 and 5 post-inoculation. Thus, RNA was obtained from 21 samples in total, where each sample includes material collected from three to five roots. Total RNA from infected roots was isolated using the RNeasy Mini Kit (QIAGEN, Germany). The isolated transcripts were sequenced on the DNBSEQ platform at the Beijing Genome Institute (BGI) (paired-end, 100 bp reads). The resulting data was filtered using SOAPnuke (Chen et al., 2017). To assess read abundance for reference transcriptomes of both flax (GCA_000224295.2) and Folini (MI39 strain) (Bray et al., 2016) the Kalisto tool (v.0.44.0) with bootstrap support for 100 samples and a k-mer length of 31 nucleotides was applied. Further downstream analysis was conducted using the sleuth package (version 0.30.1) (Pimentel et al., 2016). Transcripts characterized with low coverage values (i.e., 5 reads in either half or more than half of the samples undergoing the comparison) were filtered out and excluded from further consideration.

Fusarium cultures were grown in Chapek Broth medium for a week in a shaker incubator at 25°C and 100 rpm. Then mycelia were concentrated and rinsed twice in sterile water by centrifugation (3200 rpm) for as long as 15 min at 4°C. Next, the washed mycelia were collected, placed into mortars and dried in a thermostat at 32°C for 24 hours. The product was then powdered with sterile pestles. To extract DNA from the mycelial powder, we used DNeasy Plant Mini Kit (Qiagen) in accordance with the manufacturer’s recommendations. DNA quality was assessed by Qubit 4.0 fluorimeter (Thermo Fisher Scientific) and electrophoresis in 1% agarose gel with subsequent visualization by GelDoc system (Bio-Rad). Finally, we ran PCRs with specific oligonucleotide primers ( Supplementary Table S3 ) to evaluate expression levels of SIX1 — SIX 14 genes in each strain. Likewise, PCR products were documented by electrophoresis in 1% agarose gel using the GelDoc system (Bio-Rad).

Nucleotide sequences of genes SIX1, SIX7, SIX10, SIX12 and SIX13 were extracted from Genbank. Homologous loci in Folini genomes were identified with the blast+ tool. The resulting homology regions were analyzed with the MIT GENSCAN web server (Burge and Karlin, 1998) and the FindORF tool from the NCBI web server. The predicted polypeptides were aligned to the homologous regions to visualize introns with CLC Sequence Viewer (QIAGEN Aarhus).

Ten Folini strains were isolated from flax straw or from vegetative plants cultivated in the Tver region (Russia). In addition, two isolates, F200 and F456, came from the Netherlands and France correspondingly, while the F365 strain was obtained from China ( Table 1 ). Three flax cultivars distinguished by contrasting disease susceptibility, namely A-29 - highly resistant, Tvertza - moderately resistant and, finally, Tost 3 – the highly susceptible type, were used to estimate strain virulence under greenhouse conditions by calculating the Disease Severity Index (DSI) ( Table 2 ), where the DSI is a normalized proportion of genotypes with identical disease symptoms (see Section 2.1). Thus, Folini strains F282, F287, F324, F329, F483 and MI39 were classified as strongly virulent, while strains F200, F365, F418, F456, and F482 belog to the weak virulence group. Finally, strains F476 and F525 developed a moderate response to infection.

The ability of the hyphae of two individual fungal isolates to fuse and form viable heterokaryons (i.e., vegetative compatibility) may be considered indirect evidence of genetic similarity (Kistler, 1997; Katan and Primo, 1999). Because of the above, we interrogated 11 Folini strains to determine Vegetative Compatibility (VCG) groups ( Table 2 ). Of the four VCGs revealed, the largest of them, i.e., group “004416” contains seven strains, including MI39. The next group by size, the “004417”, has three members, while groups “004415” and “004418” are represented by one strain each (F365 and F482, correspondingly).

The MI39 reference genome was constructed from PacBio long reads and was later polished and corrected using short sequencing reads generated by Illumina technology as described in (Kanapin et al., 2020). Out of the 29 scaffolds, 20 scaffolds of the resulting reference genome assembly have been identified as chromosomes based on their homology to the Fusarium oxysporum f. sp. lycopersici reference genome and the similarity of the genome assemblies of other F. oxysporum f. sp. lini isolates, as elaborated in (Kanapin et al., 2020). The provisional attribution of the Folini chromosomes into two genomic compartments has been facilitated by the syntenic analyses of the MI39 strain that were previously reported (Kanapin et al., 2020; Samsonova et al., 2021). Consequently, chromosomes 1 – 11 and 16 – 19 are assigned to the conservative part, whereas chromosomes 12 –15 and 20 are allocated to the variable compartment.

We used the MI39 genome as a reference to reassemble the genomes of 12 Folini strains sequenced by different methods ( Table 1 ). The sizes of assembled genomes vary from 53.2 to 69.5 Mbp, with average GC content equal to 48.1%. The N50 value ranges from 28,09 to 173,71 kbp. Both numbers of predicted chromosomes and gene models are similar across thirteen Folini genomes, with the latter ranging from 14,087 to 14,395. The majority of the strain genomes are comprised of twenty chromosomes, while F329 has nineteen of them in the predicted karyotype.

To gain insight into patterns of synteny and orthology across Folini isolates we compared MI39 and F418 strains as a) their genome sequences resulted from hybrid de novo genome assembly, and b) the isolates demonstrate contrasting pathogenic potential. Specifically, MI39 is a highly virulent strain, while F418 exhibit weak virulence. To analyze syntenic patterns, we ran GENESPACE (Lovell et al., 2022) software on MI39 and F418 genome assemblies. The conservative chromosomes of the above strains are similar in size, while the dispensable chromosomes of the MI39 isolate, namely, chromosomes 12–15, differ in length from their counterparts in F418 ( Figures 1A, B ). As expected, the conservative chromosome sets of MI39 and F418 strains demonstrate a high degree of similarity in gene order. Also, the conservation pattern did not change once F. oxysporum Fo47 endophytic strain was added to the analysis (Wang et al., 2020). However, we observed at least eight whole chromosome inversions in the Fo47 genome ( Figure 1A ).

The dispensable chromosomes of fungal pathogens are associated with virulence and host-specificity (Frantzeskakis et al., 2019; Torres et al., 2020). Syntenic analysis of dispensable chromosomes in MI39 and F418 isolates revealed chromosomes 15 and 14 as the most and least conserved ones, respectively. Chromosome 15 showed the highest level (64.3%) of similarity in gene order between the aforementioned strains. In contrast, chromosome 14 shares syntenic blocks with dispensable chromosomes 13, 14, and 15 and core chromosomes 1, 6, and 8. It is also longer in MI39 as compared to F418. ( Figure 1B ). Similarly, other accessory chromosomes of MI39 share syntenic blocks with core and non-homologous dispensable chromosomes of F418, as shown in Figures 1A, B .

We performed the phylogenetic analysis to resolve the relationships and estimate genetic distances between Folini strains. Two phylogenetic trees were constructed: (1) a maximum likelihood (ML) tree for EF-1α sequence data for a group of 50 isolates, including 13 Folini genomes from this study, 20 Folini isolates from NCBI database (https://www.ncbi.nlm.nih.gov/nucleotide/), and 26 different F. oxysporum ff. spp. isolates, while 4 other Fusarium species were used as an outgroup ( Supplementary Table S1 , Figure 2A ), and (2) a neighbor-joining (NJ) cladogram based on a whole-genome SNP set for 12 Folini strains and MI39 genome as a reference (see Figure 2B ). On both trees, most of the strains clustered into two distinct clades, while F482 appeared to form a separate branch. It is worth noting that the F365 isolate was in a separate branch on the NJ tree, but grouped together with other isolates obtained from flax and cotton plants on the ML tree, forming a clade.

To estimate the genomic diversity of Folini strains and gain insight into its gene repertoire, we constructed a pangenome by clustering 185,853 proteins identified in the 13 Folini genomes. We thus detected 17,731 non-redundant orthologous protein clusters (othogroups), of which 9388 (53%) are core orthogroups shared among all genomes ( Figure 3A ). Moreover, 4095 (23%) of all orthogroups are present in some but not all genomes (i.e., accessory orthogroups) and 24% of the orthogroups (4248) are singletons composed of genes found in one of the genomes only. Curiously, F482 and F365 strains have the highest proportion of singletons, 837 and 608 genes, respectively, while the lowest proportion of singletons (144 and 180 genes) is detected in F282 and F324 isolates, respectively. The rest of the strains contain comparable numbers of singletons in the interval sandwiched by 212 and 399 genes.

The gene accumulation curve ( Figure 3B ) demonstrates that the number of core genes continually decreases with the addition of new genomes. On the contrary, pangenome size, defined as a total number of non-redundant orthologous protein clusters, grows steadily with the iterative addition of new genomes, which eventually characterizes the Folini genome as open one.

The GO enrichment analysis (FDR< 0.001) of the core, accessory and singleton proteins encoded by the Folini pangenome reveals clear differences in their functions ( Figure 3C ). The core proteins are significantly enriched in terms associated with basic cellular and molecular functions, such as microtubule-based movement, ribosome, structural constituent of the ribosome, vesicle-mediated transport, spliceosomal complex, and chromatin remodeling. As expected, the accessory proteins demonstrate enrichment of GO categories linked to pathogenicity and defense mechanisms (e.g. channel activity, D−arabinono−1,4−lactone oxidase activity, hydrolase activity, phosphorylation, polygalacturonase activity, primary amine oxidase activity, protein ADP−ribosylation, serine-type peptidase activity, transmembrane signaling receptor activity), as well as to DNA transposition (e.g. DNA integration, DNA synthesis involved in DNA repair, transposition). Finally, singleton proteins are enriched in terms associated with pathogen recognition and immune signaling, such as cell adhesion, assembly of signal complexes, and protein tyrosine kinase activity.

Fungus interaction with the plant immune system, as well as the digestion of plant cells to get nutrients, are accomplished by the secretion of specific proteins that make up an individual secretome (de Sain and Rep, 2015; Jashni et al., 2015; Berlemont, 2017; Kameshwar and Qin, 2018), which, incidentally, is roughly of the same size in different Folini strains. The pansecretome was assembled from 621 orthogroups of secreted proteins (3,5% of the Folini pangenome). The core and non-core orthogroups of secreted proteins comprise 40,3% and 59,7% of the pansecretome, i.e., 250 and 371 clusters, respectively. Among the non-core orthogroups, 229 clusters (36,9% of the pansecretome) are assigned to accessory orthogroups, while 142 orthogroups (22,6% of the pansecretome) are singletons.

The pansectretome primarily comprises three functional groups: CAZYmes, effector proteins, and proteases (see Table 3 ), where the non-core orthogroups dominate within each functional class ( Figure 4A ).

We classified CAZYmes into six groups according to their functions and analyzed their abundance and distribution among pansecretome categories ( Figure 4B ). The core genes dominate in the most abundant classes, which are glycosyl hydrolases (GH), auxiliary activities (AA) and carbohydrate esterases (CE).

The average number of effector proteins identified in individual Folini genomes is about 114 ( Table 3 ). The Folini pansecretome contains 200 effector orthogroups. Of these, 88 are accessory, 40 belong to one isolate only and 72 are conserved between all strains ( Figure 4C ). The number of singletons agrees with the phylogenomic relationship between strains. For example, the F482 strain contains the maximum number of singletons (10 orthogroups), while the F365 strain and F456 isolate incorporate 7 and 5 singletons correspondingly. The number of singleton effectors in the remaining ten genomes ranges between 3 and 1.

The Folini genome has dual compartmentalization encompassing conserved and dispensable (i.e., lineage-specific) parts. Incidentally, the reference MI39 strain has 116 effector genes ( Table 3 ; Supplementary Table S6 ), of which 63 and 53 genes belong to core and accessory pansecretome, respectively. F418, the second strain sequenced with LRS technology, has the same number of core effectors and 7 non-core effectors fewer than MI39. The majority of genes encoding these proteins are located on conserved chromosomes, while only 12 and 9 non-core effector genes map to dispensable chromosomes of MI39 and F418 correspondingly.

The expression level of predicted effectors in MI39 mycelia was assessed in both liquid culture and in planta by observing the infection progress in seedlings from resistant (Atalante) and susceptible (LM98) flax varieties over time. While all MI39 effectors were expressed, their levels varied. One-third of the effectors showed low expression in both mycelium and infected plants. The remaining effectors are divided into two categories: (1) those present in both the mycelium and the plant and (2) those only found in the plant (as shown in Supplementary Figure S1 ). The effector orthogroups’ presence-absence heatmap in Folini strains indicates a clear separation between strongly and weakly virulent strains ( Figure 4D ). Notably, an orthogroup known as FOL004461, consists of eight proteins expressed only in planta. Of these proteins, five are present in MI39, as well as in other strongly virulent strains, while two proteins are found in weakly virulent strains, namely F365 and F200, and one is present in the medium virulent strain, F525. These proteins contain three specific PFAM domains: MANEC (PF07502), PAN_3 (PF08277), and PAN_4 (PF14295). PAN domains are ubiquitous in diverse proteins and are involved in protein-protein and protein-carbohydrate interactions (Tordai et al., 1999), while the MANEC domain may play a pivotal role in the formation of protein complexes that contain various protease activators and inhibitors (Guo et al., 2004).

We performed a genome-wide annotation of CAZYmes, effector proteins, and proteases using the PFAM database (Mistry et al., 2020). The results of hierarchical clustering of PFAM domain frequencies computed for strongly and moderately virulent Folini isolates in core, accessory and singleton categories highlight a remarkable difference in domain repertoire ( Figure 4D ; Supplementary Figure S2 ). It is noteworthy that certain PFAM domains are exclusive to specific virulence groups, with most of these domains being encoded by non-core genes. Among strongly virulent strains, the number of unique PFAMs found in CAZYmes and proteases is three times less than in weakly virulent strains (23 and 68 domains, respectively), as presented in Supplementary Table S4 . This disparity can be attributed to the close genetic relationship of strongly virulent strains, as evident from most of them being grouped in one clade of the phylogenetic tree ( Figure 2A ). Despite the significant genetic similarity between strongly and weakly virulent strains, the number of unique PFAM domains found in effectors is similar for both, with 26 and 35, respectively (see Supplementary Table S5 ). The PFAM domains found in CAZYmes of strongly virulent strains are commonly found in transmembrane and extracellular proteins, as well as proteins participating in glycoside bond hydrolysis (PF01120, PF07477, PF03632, PF17851, etc.), carbohydrate synthesis (PF05691), and fatty acid metabolism (PF02737). Additionally, weakly virulent strains also possess a functionally similar set of specific domains for CAZYmes, but with the added presence of domains characteristic of methyltransferases (e.g., PF08241, PF13489, PF13649) and host-pathogen recognition (PF13517). The functional diversity of PFAM domains found in effector proteins is particularly high ( Supplementary Table S5 ), with strongly virulent strains possessing domains for proteins involved in signaling (PF07645, PF13923) and regulation of transcription (PF05086), while weakly virulent strains exhibit protein domains involved in tyrosine and phenylalanine biosynthesis (PF01817) and chromatin remodeling (PF09011).

The pathogen’s virulence is modulated by cysteine-rich small effector proteins called Secreted in Xylem (SIX) proteins (Rep et al., 2004; Houterman et al., 2009; Kashiwa et al., 2013; Schmidt et al., 2013; Ma et al., 2015; van Dam et al., 2017; Widinugraheni et al., 2018). SIX proteins were first discovered in xylem sap of tomato plants (Houterman et al., 2007; Schmidt et al., 2013). The coding of these proteins is attributed to SIX genes belonging to 14 gene families and located on pathogenic-specific chromosomes. Nonetheless, formae speciales have diverse SIX protein profiles and gene sequences (de Sain and Rep, 2015; Jangir et al., 2021). Thus, SIX genes can be utilized as markers for identifying pathogens.

Surprisingly, we could not find SIX genes in two weakly virulent strains (i.e., F365 and F482). Yet, the rest of the 11 Folini genomes considered in this study exhibit identical sets of the SIX gene families, namely SIX1, SIX7, SIX10, SIX12 and SIX13, thus suggesting a conservation of the SIX gene repertoire ( Table 4 ; Supplementary Figures S3 , S7 - S20 ).

The SIX1 gene sequence is identical in all genomes except for F418 and F456. The Six1 protein in the F418 isolate has 12 amino acid substitutions, of which one, at position 238 (glutamate to glutamine), is also present in the Six1 protein of the F456 isolate. Also, there are two paralogs of the SIX7 gene (i.e., SIX7a and SIX7b) on chromosomes 12 and 15, respectively ( Table 4 ). Interestingly, all Folini SIX-containing strains have SIX7a, while the MI39 genome has three copies of this gene. We found SIX7b gene in genomes of the MI39, F418, F456, F476, F483 and F525 strains. The amino acid sequences of SIX7a and SIX7b are 82% identical. The SIX10 gene is situated on chromosome 12 and includes one intron. It encodes two polypeptide variants, one with arginine and the other with glutamine at the C-terminus. Two paralogs of the SIX12 gene, SIX12a and SIX12b, are present, with widely varying location and number of copies. Most Folini strains, M39, F282, F287, F324, F329 and F418 contain both paralogs. In all strains, except for F456 and F483, SIX12a maps to chromosome 12, while SIX12b is found on chromosome 13. In the F456 strain, two copies of SIX12a are located on chromosome 13, while in the F483, there are two SIX12a copies on chromosome 13 and only one copy on chromosome 12. In the F476 and the F525 strains, genes SIX12b and SIX12a are found on chromosomes 13 and 12, respectively ( Table 4 ). The F200 strain has three copies of the SIX12b gene, all of which are located on chromosome 13. The Six12a and Six12b proteins are almost identical, differing only in two positions: 83 and 91. At these positions, Six12a has glutamate and lysine, while Six12b has lysine and asparagine, respectively. The SIX13 gene has two introns and is always present as a single copy on chromosome 12 (as shown in Table 4 ). This gene is strongly conserved across strains.

We characterized the relative location of SIX genes in chromosome 12 using the genomes assembled with the LRS technology (M39 and F418 isolates, Figure 5A ). The analyses revealed several gene clusters of similar architecture.

Recent results show that SIX1 and SIX13 gene families demonstrate host-related clustering, as revealed by phylogenetic analyses (Samsonova et al., 2021). Given these findings, we examined the remaining SIX genes to check if the aforementioned host-related grouping holds. The SIX gene dataset included members from different F. oxysporum ff. spp. alongside three other Folini isolates from Australia and Canada ( Supplementary Table S3 ). The phylogeny analyses conducted on the SIX7, SIX10 and SIX12 genes demonstrate that the Folini isolates are separate from other ff. spp. (see Supplementary Figures S4 , S5 ). The phylogeny of the SIX7 ( Supplementary Figure S4A ) and SIX12 ( Supplementary Figure S5 ) gene families distinctly highlights the difference between SIX7a and SIX7b, as well as SIX12a and SIX12b sequence variants, since they occupy separate clades on the trees. We thus conclude that the SIX7 and SIX12 gene homologs are of polyphyletic origin. In contrast, as demonstrated in the respective phylogenetic tree, the origin of SIX10 is monophyletic.

We used RNAseq technology to assess the expression levels of SIX genes during the 5-day infection process of flax plants. We conducted this evaluation on seedlings from both resistant (Atalante) and susceptible (LM98) varieties, as shown in Figure 5B . SIX1 and SIX13 exhibit higher expression levels than other SIX genes. SIX13 expression in LM98 exhibits variable levels of activity over replicates. In contrast to the Atalante variety, in LM98, it is slightly lower and reaches high levels only by the fifth day. The remaining SIX genes (and their copies) are either expressed at low or moderate levels. Interestingly, the latter is detected in two copies of SIX7a, a copy of SIX12 and SIX10 on day five post-infection in the susceptible flax cultivar.

Thus the characterization of the SIX gene diversity in Folini strains revealed the conservation of their repertoire: most strains had the same set of SIX genes regardless of their virulence status. However two Folini isolates lacked any of the SIX genes in spite of weak virulence.

It is widely accepted that F. oxysporum is a species that reproduces asexually (Gordon and Martyn, 1997). However, the fungus contains a conserved set of genes involved in sexual development and reproduction in other fungi, including mating-type idiomorphs (Yun et al., 2000). We found that twelve of thirteen Folini strains have MAT1–2 idiomorph, while only the F365 strain is MAT1–1 positive ( Supplementary Figure S6 ). In all strains considered, the MAT1–2 locus contains two genes oriented in opposite directions at its ends. The first gene is MAT1–2-1, highly conserved across Sordariomycetes. The second gene encodes a protein exhibiting a high level of homology with Mat1–2-3 protein from F. avenaceum and hence was identified as MAT1–2-3 (Lysøe et al., 2014) The F482 strain has the truncated version of this gene, MAT1–2-3m, which lacks the whole third exon and part of the second intron. PCR analysis revealed that MAT1–2-3 and MAT1–2-3m transcripts in liquid mycelial cultures of MI39 were of the expected gene length.

The F365 strain’s MAT1–1 locus conforms to the architecture characteristic of heterothallic Sordariomycetes ( Supplementary Figure S6 ), apart from the fact that the MAT1–1-3 gene lacks an intron.