SC23 and SC50 showed an average growth of 5.4 cm.

### *3.2. Subculturing Affects Conidial Production but Not an In Vitro Growth Rate*

Mycelium growth rates from subcultures SC1, SC23, and SC50 were measured on PDA plates. Despite the three subcultures having different adaptation times on PDA (1, 23, or 50 weeks), no significant differences in growth rate were observed (*p* = 0.42) after 5 days (Figure 3A). In detail, SC1 showed an average growth of 5.1 cm along the axes, whereas

**Figure 3.** (**A**) Subcultures development (cm) along the diameters of Petri dishes containing PDA. Columns represent the average of three replicates (±SE). Values with the same letters are not significantly different (*p* > 0.05) based on Duncan's multiple comparison tests. (**B**) Average conidia production by SC1, SC23, SC50, and SC50×3 developed on PDA under near-UV light for 28 days. Columns represent the average of three replicates (±SE). Values with different letters are significantly different based on Duncan's multiple comparison tests (*p* < 0.05).

> Conversely, significantly different conidia productions by the varying subcultures were detected (Figure 3B). In detail, conidiation followed the significant (*p* < 0.05) gradient: SC50×3 > SC1 > SC23 > SC50. In addition, after *in vivo* passages of the last subculture (SC50) for three times onto wheat head tissues, this strain produced a high number of conidia, which resulted in an even higher number (*p* < 0.05) than that observed in the first subculture (SC1).

### *3.3. Subculturing Does Not Affect In Vitro Secondary Metabolite Production*

The accumulation of the main secondary metabolites detected in SC1, SC23, SC50, and SC50×3 subcultures are reported in Table S1.

In general, no significant differences in secondary metabolite biosynthesis between the four subcultures were detected. In detail, SC1 produced 19,900 μg kg−<sup>1</sup> of 15-ADON in addition to 36,600 μg kg−<sup>1</sup> of DON. A very low production of 3-ADON (700 μg kg<sup>−</sup>1), NIV (114 μg kg<sup>−</sup>1), and sambucinol (412 μg kg<sup>−</sup>1) was also detected. In addition, SC1 biosynthesised 56,000 μg kg−<sup>1</sup> of zearalenone (ZEN) and some ZEN derivatives, such as 1700 μg kg−<sup>1</sup> of alpha-zearalenol and 12,000 μg kg−<sup>1</sup> of beta-zearalenol. Other metabolites such as culmorin (3400 μg kg<sup>−</sup>1), 15-hydroxyculmorin (4200 μg kg<sup>−</sup>1), and fusarin C (54,000 μg kg<sup>−</sup>1) completed the SC1 *in vitro* mycotoxigenic profile. All the reported values are the average of three replicates. As mentioned before, the subculturing process performed on PDA as well as the following passages on the host head tissues did not significantly alter the ability of the fungus to produce secondary metabolites. For example, the production of DON showed an increase (even if without significant differences, *p* > 0.05) going from SC1 to SC23 (60,000 μg kg−1) and to SC50 (74,000 μg kg−1). A partial, non-significant decrease was detected for SC50×3 (53,000 <sup>μ</sup>g kg<sup>−</sup>1).

The 15-ADON biosynthesis levels did not show significant changes between subcultures (*p* = 0.91). A very similar amount was detected in all subcultures (SC1 = 19,900 μg kg<sup>−</sup>1, SC23 = 16,600 <sup>μ</sup>g kg<sup>−</sup>1, SC50 = 18,100 <sup>μ</sup>g kg<sup>−</sup>1, SC50×3 = 19,400 <sup>μ</sup>g kg<sup>−</sup>1).

Similarly, 3-ADON biosynthesis was nearly the same in all subcultures (*p* = 0.37). In detail, 1100 μg kg−<sup>1</sup> of 3-ADON were produced by SC23 and 1400 μg kg−<sup>1</sup> by SC50 and SC50×3. NIV and sambucinol were detected with similar levels for SC1, SC23, SC50, and SC50×3 (*p* = 0.81 and *p* = 0.70, respectively). Culmorin levels showed a decreasing trend that was followed by a restoration after host re-infection, whereas 15-hydroxiculmorin increased from SC1 to SC50 and then decreased again on SC50×3, but both without significant differences (*p* = 0.96 for culmorin and *p* = 0.57 for 15-hydroxiculmorin).

Subculturing induced a very small increase of ZEN production while plant re-infection brought it back to the SC1 levels, even if no significant differences were detected (*p* = 0.32). In addition, alpha-zearelenol and beta-zearalenol did not show any significant differences (*p* = 0.87 and *p* = 0.57). Among other *Fusarium* metabolites, fusarin C showed a decrease in the total amount produced following the three passages but, again, without significant differences (*p* = 0.34).

Finally, no significant variations were detected for butenolid (*p* = 0.59), gibepyron D (*p* = 0.22), aurofusarin (*p* = 0.60), and rubrofusarin (*p* = 0.24) biosynthesis among the different subcultures.

### *3.4. DNA Methylation Analysis*

3.4.1. Identification of Differentially Methylated Positions and Differentially Methylated Regions

MCSeEd (Methylation Context Sensitive Enzyme ddRAD) [37,38] was used to investigate DNA methylation changes induced by subculturing. To this end, next-generation sequencing (NGS) libraries from genomic DNA purified from PDA plates (SC50) and wheat heads (SC50×3) were constructed. Therefore, a total of 12 libraries were produced by double restriction ligations with each using *Mse*I in combination with one of the three methylation-sensitive enzymes *Aci*I, *Sex*AI, and *Eco*T22I, for the CG, CHG, and CHH contexts, respectively (Supplementary Table S2).

After quality control, a mean of 822,679 thousand 150-bp-long reads from each library were obtained and aligned to the *F. graminearum* PH-1 reference genome [41]. Only reads mapped at unique genomic positions were retained. Thus, considering the three different contexts, a total of 2,899,673, 4,755,848, and 1,516,029 reads were mapped uniquely on the reference genome (92.6% of the total reads, with a minimum of 86.8% for *Aci*I, and a maximum of 96.6% for *Sex*AI) and were classified as MCSeEd loci (Supplementary Table S3).

Therefore, a total of 138,119 loci containing cytosines (120,439 in symmetric, and 17,680 in asymmetric contexts) (Supplementary Table S4) were identified.

The mapping location of each MCSeEd locus was investigated to determine whether it fell within a gene window that included the region within 0.5 kb upstream of the transcription start site (TSS), the transcribed region (i.e., the gene body), and the region within 0.5 kb downstream of the transcription termination site (TTS). Furthermore, 92% (*Aci*I), 91% (*Sex*AI), and 87% (*Eco*T22I) of the identified MCSeEd loci were included within these gene windows (Supplementary Figure S1).

After normalization of the MCSeEd loci, the sites covered by a total number of reads <4 or showing excessive read-count variation among the replicates (standard deviation of 5 for CG and 10 for GHG and CHH) were discarded. The remaining sites were used to estimate a total of 13,899 DMPs, out of the 138,119 MCSeEd loci, with significantly altered methylation levels between the SC50×3 and SC50 samples (false discovery rate, ≤0.05). Of these, 12,326 belonged to symmetric contexts, and 1573 belonged to asymmetric contexts (Supplementary Table S4).

Principal component analysis was used to graphically portray the samples based on the DMPs' methylation levels (Supplementary Figure S2).

The first latent component (PC1) accounted for 71.6%, 78.4%, and 71.8% of the total variance for the CG, CHG, and CHH, contexts, respectively, and clearly discriminated between SC50×3 and SC50, indicating that the head-to-head transfer of mycelium induced genome-wide methylation changes. Accordingly, complete linkage clustering of the methylation levels at DMPs clearly separated the SC50×3 and SC50 (Supplementary Figure S2).

Considering all of the methylation changes being induced by host colonization in the replicates, we observed 1.4-fold (CG) to 1.15-fold (CHH) more methylation decreases than increases in response to healthy heads' infection whereas, for CHG, the proportion of methylation changes was 1.12 in the other direction.

Genomic regions with co-regulated methylation changes upon subculturing, known as DMRs, were identified. In total, 932 DMRs were scored for CG (874), CHG (19), and CHH (39) contexts (Supplementary Table S5).

The estimated relative methylation level of the DMPs belonging to each DMR were hierarchically clustered and, as expected, clustered according to the treatment, as SC50×3 or SC50 (Figure 4 and Supplementary Figure S3). In particular, for all three contexts, the number of DMRs with higher methylation levels in the SC50×3 samples (relative to the SC50 samples) was lower than the number of DMRs that showed a lower level in the SC50×3 samples (Figure 4 and Supplementary Figure S3).

**Figure 4.** Relative methylation frequencies of differentially methylated regions as identified from the comparison between SC50 and SC50×3 subcultures. Relative methylation frequencies of the differentially methylated positions contained in each differentially methylated region (**A**–**C**) for CG, CHG, and CHH, respectively) were averaged and used in complete linkage clustering analysis of samples derived from SC50 and SC50×3 based on differentially methylated regions. (**D**) Circos plot. From outer inward: number of genes in adjacent genomic chromosome regions of 20 kb. Ratio of the number of *F. graminearum-F. verticillioides* collinear genes and total number of *F. graminearum* genes in each region. DMPs (CG context) and DMRs (CG context).

Next, we analysed the distribution of DMPs and DMRs along *F. graminearum* chromosomes. Several studies proposed that *F. graminearum* genome can be partitioned in a core portion enriched for housekeeping genes and a dispensable portion with a high

frequency of pathogenesis and virulence-related genes [42]. The dispensable genomic portions can be identified as regions of low gene collinearity (hereafter, referred to as not conserved (NC) regions) between *F. graminearum* and *F. verticillioides* or *F. oxysporum*. The distributions of both DMPs and DMRs along chromosomes were visualized as the total number of these features for each adjacent genomic window of 20 kb. The black histograms of Figure 3 indicate the location of NC regions in four *F. graminearum* chromosomes. For all the analysed contexts, no preferential accumulation of either DMPs or DMRs between NC regions and other chromosome regions were observed (*p* > 0.05 of a non-parametric permutation test).

### 3.4.2. Differentially Methylated Genes

DMP and DMR distributions were analysed in relation to the coding and regulatory genomic sequences. In particular, we compared the distribution of DMPs and DMRs in transcribed genic regions extended by 0.5 kb at both ends (extended gene bodies, EGBs) (Supplementary Figure S1) and found that DMRs mapped preferentially to EGBs.

In addition, we plotted the distribution of significant DMPs along the EGBs for CG (Figure 5), CHG, and CHH (Supplementary Figure S4) contexts. The main differences in CG relative methylation levels between SC50 and SC50×3 were observed in the regulative regions and in proximity of TSS and TTS sites. In particular, a decreased, relative, methylation level of the samples grown on plants as compared to artificial substrate was found. Conversely, along the gene body, no changes were highlighted. In the other two contexts, the low number of significant DMPs (Supplementary Table S4) were not able to properly reveal differences along EGB's genomic region (Supplementary Figure S4).

**Figure 5.** Plotted DMPs along the EGBs (coding region, in yellow, with the regions 500 bp upstream and downstream) for the CG context.

In particular 930, 13, and 40 EGBs were overlapped at least once by 932 DMRs in the 0.5-kb windows upstream of TSS, within the gene body, or in the 0.5-kb windows downstream of TTS, respectively. The genes belonging to these EGBs were defined as differentially methylated genes (DMGs, Supplementary Table S6).

Analysis of Gene Ontologies' enrichment demonstrated that DMGs are enriched for GO terms in relation to transcriptional regulation (GO\_0006355) and chitin metabolism (GO:0006032). More DMGs than expected by chance were involved in zinc ion (GO:0008270) and DNA (0003677) binding. Next, we assigned DMGs to gene families based on the presence of PFAM domains within the encoded proteins. The PFAM domains are significantly more abundant than expected by chance in the DMGs' dataset are reported in Table S7 (Fisher exact test *p* < 0.05). Several of these domains have been identified in genes known to be linked directly or indirectly to virulence.

The top five PFAM domains enriched in the DMGs were associated with isoprenes and carbohydrate metabolism (PF08544.13, PF00180.20), endonuclease and exonuclease activity (PF003372), vacuolar 14 fab1-binding (PF12755.7), and ureo-hydrolase (PF00491.21).

### **4. Discussion**

Fungal pathogens are the predominant causal agents of plant diseases, causing yield and quality losses [43–45]. To successfully infect plants, fungal pathogens use different strategies to exert their virulence during the infection process, such as when adjusting the activity of various molecules, which may be effectors or extracellular factors [46,47], by regulating their transcription levels [20]. Some virulence factors are upregulated to facilitate host colonization and infection, whereas others are downregulated to mitigate host responses [48,49]. Furthermore, the genome plasticity of fungal plant pathogens allows the adaptation of the metabolism and of the reproductive strategies to variable environmental conditions, such as light cycle, temperature, substrate type, and the presence/absence of hosts [50,51]. DNA methylation is a basic modification of genomic DNA in eukaryotes with significant effects on gene expression, genomic imprinting, and transposon silencing involving gene promoter regions, transposable elements, repeat sequences, and transcribed regions of genes [52–56]. In filamentous fungi, transcriptome and methylome studies have shown that DNA methylation is linked to gene expression and to the silencing of transposable elements [52,57]. For these reasons, the present study was based on the DNA methylation approach to reveal hypothetical genes involved in the virulence of *F. graminearum*, which is the most important FHB pathogen of wheat worldwide. In fact, DNA methylation studies can be considered useful tools to identify novel genes associated with the aggressiveness of fungal pathogens subject to environmental modifications [58] and/or stresses, such as adapting to an artificial substrate or to host tissues. Changes of DNA methylation patterns in response to environmental stresses were observed in plants, during their growth and development [53,59,60].

In the present work, a high virulent strain of *F. graminearum* (FG8) was preliminarily stressed when performing subculturing (50 times for 50 weeks) on an artificial substrate (PDA) to verify if the adaptation to a nutrient-rich medium induced an attenuation of virulence. To assess the effect of the *in vitro* subculturing process on fungal virulence, the aggressiveness of three selected subcultures (SC1–SC23–SC50) was examined. Stem base crown rot virulence assays showed a progressive, aggressive decline related to subculturing time toward this bread wheat tissue. A similar result was also observed on bread wheat heads. Furthermore, the mycelium from the subculture developed for 50 weeks on PDA (SC50) was used to inoculate for three consecutive head-to-head passages (SC50×3) bread wheat heads with the objective of restoring the native virulence of the FG8 strain. In fact, SC50×3 exhibited a strong aggressiveness on heads, that is even higher than SC1, likely due to the repeated inoculation of healthy heads in a short time window (6 weeks).

These results are in line with previous studies that showed a virulence decline of different entomopathogenic fungal species or isolates (e.g., *Metarhizium anisopliae*, *Beauveria bassiana*, *Beauveria densa*, *Nomuraea rileyi*, *Paecilomyces farinosus*, *Verticillium lecanii*), caused by artificial *in vitro* subculturing on nutrient-reach media and long-term routine maintenance [61–64], even if this decline is not always reported, such as in some strains of *Paecilomyces fumosoroseus*, *P. farinosus*, and *B. bassiana* [65–67]. As observed in this study, other researchers reported that virulence can be restored when a pathogen passes from an artificial media to a suitable host [68–71]. In addition to the subculturing process, the simple *in vitro* growth on different artificial media can influence conidial germination, growth, and virulence of fungal pathogens [72–74]. The *in vitro* growth on artificial media could also induce genetic modifications, as already demonstrated for a *F. verticillioides* strain subject to a subculturing process, in which the accumulation of about 14 genetic variants was observed [75]. The present work also shows that consecutive *in vitro* subcultures of *F. graminearum* caused a considerable decline of conidiation passing from SC1 to SC50, confirming previous studies on *B. bassiana*, *M. anisopliae*, and *Metarhizium brunneum*

strains [76,77]. With the passage on a healthy host, a significant increase of conidia production from SC50 to SC50×3 was observed. Even if some phenotypic changes, such as the reduction in the growth rate, may be typically associated with *in vitro* degeneration [69]. No differences were observed comparing the measures of the diameters of the subcultures, as also found in *B. bassiana* [76].

Finally, a very large spectrum of *F. graminearum* secondary metabolites biosynthesis following prolonged subculturing was investigated. Even if the lower biosynthesis of some secondary metabolites is known to be linked to subculturing processes in different fungal genera, such as *Periconia* sp., *Fusarium* spp., *Galactomyces* sp., and *Phomopsis* sp. [78–80], the exact mechanisms that cause this attenuation are still unclear. They may be attributed to the absence of a host stimulus or to gene silencing occurring in axenic cultures [81,82]. However, this phenomenon was not observed in this study.

DNA methylation and transposon activity have already been investigated to be at the base of virulence loss and conidiation ability as a consequence of serial *in vitro* subcultures as well as at the origin of virulence restoration following host re-inoculation [69]. In the present study, we detected a lower level of relative methylation of SC50×3 when compared to SC50 in all contexts, suggesting that this genomic strategy was employed by *F. graminearum* in order to restore an efficient virulence. No significant differences in accumulating methylation changes were observed between genomic compartments. This finding suggests that, to achieve the phenotypic changes described in this study, both genomic regions hosting genes involved in basal metabolism and those regulating virulence are equally important. At a lower scale, the regulatory regions of genes were mainly affected by the previously mentioned methylation changes, especially for the CG context. Moreover, the different number of significant DMPs (Supplementary Table S4) revealed that CG was the methylation context mainly affected by the subculturing process. Differentially methylated region distributions in relation to coding and regulatory genomic sequences identified a total of 1024 genes, which are putatively regulated by DNA methylation.

Some of these genes have already been investigated for their crucial role in *F. graminearum* aggressiveness. Hereafter, some examples of genes that showed different methylation levels between SC50 and SC50×3 with our analyses and that have been previously described in other studies for their role in *F. graminearum* virulence are discussed. The genes FGSG\_06675 (FgLeu2A) and FGSG\_10671 (FgLeu2B) are known to be involved in the leucine metabolic pathway [83] of *F. graminearum*, and their importance in pathogenicity (in particular of FGSG\_06675) and DON production (both FGSG\_06675 and FGSG\_10671) is well-known [84]. In the present experiment, other genes with different methylation levels during wheat infection are involved in the DON biosynthetic pathway: FGSG\_05912 (mevalonate kinase) and FGSG\_09764 (5 -phosphomevalonate kinase). These enzymes are responsible for the transformation of mevalonate in 5 -phosphomevalonate and, subsequently, in 5 -pyrphosphomevlonate during the chemical conversion of the acetyl-CoA in farnesyl pyrophosphate (FPP), which is the main substrate for DON biosynthesis [85]. Considering that DON is a well-known virulence factor of *F. graminearum* [86,87], likely FGSG\_05912 and FGSG\_09764 could be indirectly involved in fungal virulence due to their crucial role in DON production. Another interesting gene differentially methylated between SC50 and 50SCx3 is FGSG\_07896. Even if not directly implicated in fungal virulence, this gene encodes for a trichothecene 3-O-acetyltransferase (TRI101) involved in the DON self-protection mechanism of *F. graminearum* [88,89]. In addition, the target of rapamycin (TOR) kinase gene (FGSG\_08133) showed different methylation between the two samples. The protein FgTOR encoded by this gene is a key component of the TOR complex. The TOR signaling pathway of *F. graminearum* plays critical roles in regulating vegetative differentiation and virulence [90]. Other differentially methylated genes are FGSG\_07067 and FGSG\_06944, which encode for two transcription factors. *F. graminearum* mutants with the deletion of these genes showed a significant loss of virulence toward wheat heads [91]. Varied methylation levels of FGSG\_07593, encoding a glycoside hydrolase, were observed between SC50×3 and SC50. This gene is usually upregulated at the beginning of the

host colonization process [92]. Another gene differentially methylated comparing the two samples is FGSG\_11955. The gene was previously identified like the velvet gene (FgVe1 or FgVeA) of *F. graminearum*, a domain conserved in various genera of filamentous fungi. The velvet gene regulates the trichothecene biosynthesis and pathogenicity against wheat heads and affects fungal development [93,94]. In addition to this gene, FGSG\_01362 and FGSG\_06774 belonged to the velvet gene domain, and, in this study, have shown different methylation levels. FGSG\_01973, FGSG\_09917, and FGSG\_06175 encode for phospholipid hydrolases (phospholipase D, PLD) of *F. graminearum* (FgPLD1, FgPLD2, and FgPLD3). FgPLD1 is involved in the virulence toward flowering wheat heads and the mutant lacking this gene also showed reduced DON production, whereas FgPLD2 and FgPLD3 are not primarily involved in plant infection [95]. The differentially methylated FGSG\_05902 gene between SC50×3 and SC50, which is almost identical with FGL15 cloned in previous research, encodes for a lipase known to be an important virulence factor [96]. Again, gene FGSG\_04580, encoding a pleiotropic drug resistance class ABC transporter, is known to play a role in *F. graminearum* virulence toward different wheat tissues [97]. Furthermore, the ATP-binding cassette transporter Abc1, encoded by this gene, may be involved in DON release [98]. Several other ABC-G family transporters are highly expressed during host infection, such as FGSG\_08309, which showed a different methylation between the two subcultures investigated [97]. The gene FGSG\_09329, encoding an ABC-2 family transporter protein, showed a high expression during barley heads and wheat coleoptile colonization [97]. Recently, another ATP-binding cassette transporter (FgArb1) encoded by FGSG\_04181, differentially methylated between SC50 and SC50×3, proved to have a function in pathogenesis and DON production [99]. In general, the ABC transporters family has a crucial role in *F. graminearum* pathogenicity [97,100]. Comparing SC50×3 to SC50, FGSG\_01964 (GzCHS5), which encodes for a chitin synthase, is indispensable for perithecia formation and pathogenicity as well as for normal sept formation and hyphal growth [101]. Another chitin synthase gene, which is known to be involved both in DON synthesis and pathogenicity [102], is FGSG\_06550 that showed different methylation levels between the two explored subcultures, such as FGSG\_03538 (transcription factor Tri10) that is essential for DON production, and regulates the expression of the entire Tri-cluster [103,104]. FGSG\_00352, differentially methylated between SC50 and SC50×3, is the orthologous protein of Hap2, which is one of the three subunits composing the heme activator protein (HAP), also known as a nuclear factor Y (NF-Y) or CCAAT-binding factor (CBF). *F. graminearum* has eight different genes encoding for CCAAT-binding factors. The deletion of FGSG\_00352 did not significantly affect fungal mycelium growth, sexual development, mycotoxin production, and virulence [91], but other CCAAT-binding factors (FGSG\_01182 and FGSG\_05304) are involved in trichothecene production and virulence [105]. Thus, further studies on all the genes of the CCAAT-binding complex are necessary to reveal the relationship among them during host colonization. Another aggressiveness-associated gene (FGSG\_08010), that is, a regulatory virulence [106], and is usually reported to be up-regulated during infection [107], showed a different level of methylation in the present work. A different methylation between the two analyzed subcultures was observed in FGSG\_00332, encoding for a beta transducing-like (WD-40 repeat) protein, that has been demonstrated to be essential for pathogenicity in wheat [108], and in FGSG\_01665 (FSR1) that regulates *F. graminearum* virulence by acting as a scaffold for a signal transduction pathway [109]. Comparing SC50 to SC50×3, a different methylation level was observed in FGSG\_06798. Recently, this gene has been identified to encode for an acetyltransferase (FgHAT2) involved in regulating vegetative growth, conidiation, DNA damage repair, DON production, and virulence in the pathogen [110]. Another highlighted gene previously proven to be involved in pathogenicity was FGSG\_00416, belonging to a major facilitator superfamily (MFS) [111]. Furthermore, the deletion of FGSG\_03716 (Famfs1), which belongs to the MFS gene family, affected fungal development and virulence [112]. FGSG\_03541 (Tri12), with different methylations between the two subcultures, is required for DON production and fungal virulence [113].

In addition, it has been demonstrated that FGSG\_03168 has 90% similarity to FST1 of *F. verticillioides* (putative hexose transporter gene), which is functional in pathogenesis during the colonization of living maize kernels [114].

In the present study, we demonstrated that *F. graminearum* exhibits reduced virulence on bread wheat stem bases and heads after a prolonged subculturing process. However, the virulence on head tissue of bread wheat can be restored with the in planta transfer. Additionally, an innovative approach, based on the relative methylation level analysis, was used to explore novel putative virulence genes, comparing the pathogen after three generations of mycelium growth on bread wheat heads to the same fungus after approximately one-year of an *in vitro* subculturing process. Some of the genes that showed different methylation levels have been previously studied and were revealed to be related to *Fusarium* aggressiveness toward hosts. This suggests that the approach of the present study could be a promising tool in the study of *F. graminearum* genes associated with virulence on bread wheat tissues. In the future, it will be interesting to verify the function and possible involvement in virulence of all the other genes that have shown different methylation levels in this study (listed in Supplementary Table S7) but for which evidence about their implication in *F. graminearum* aggressiveness are not currently available. Finally, the present approach may be an important tool to use for other fungal pathogens to explore the pool of genes that could be involved in their virulence toward host species.

**Supplementary Materials:** The following materials are available online at https://www.mdpi.com/ article/10.3390/cells10051192/s1. Table S1: *In vitro* biosynthesis of secondary metabolites (μg kg<sup>−</sup>1) produced by subcultures of *F. graminearum* strain FG8 (SC1–SC23–SC50–SC50×3) as detected by LC-MS/MS analysis. Per each mycotoxin, the standard error (SE) is reported. Statistical differences were detected (*p* > 0.05) by one-way ANOVA. Table S2: Experimental design. Table S3: Sequencing data summary. Table S4: Methylation sequencing statistics. Uniquely mapped loci (MCSeEd loci) were normalized, filtered, and then analysed with the methyl kit to infer the number of differentially methylated positions (DMPs). Figure S1: Abundance of MCSeEd loci, DMPs, and DMRs in genomic regions (CDS, Introns, Regulatory Intergenic). Figure S2: Principal component analysis and clustering for the relative methylation levels at the differentially methylated positions obtained across all of the replicates in each sequence context: CG, CHG, and CHH. Number in brackets indicate the fraction of overall variance explained by the respective component (Dim1, Dim2). Table S5: List of significant DMRs identified in each context. Figure S3: Principal component analysis and boxplot for the relative methylation levels at the differentially methylated positions obtained across all of the replicates in each sequence context: CG, CHG, and CHH. Numbers in brackets indicate the fraction of overall variance explained by the respective component (Dim1, Dim2). Figure S4: Plotted DMPs along the EGBs (coding region, in yellow, with the regions 500 bp upstream and downstream) for CHG and CHH context. Table S6: Differentially methylated genes (DMGs) identified by intersecting DMRs. Table S7: Enrichment tests of PFAM domains within DMGs and *F. graminearum* loci. First column reports the PFAM domain, the second column reports the odds ratio, and the third column shows the *p*-value calculated for the Fisher exact test for an odd ratio equal to 1 (alternative hypothesis odds ratio >1).

**Author Contributions:** Conceived and designed the experiments, L.C., F.T., E.A., and G.M.; performed the experiments, F.T., G.B., G.M., and M.S.; statistical analysis, F.T.; data analysis, F.T., G.M., and A.P.; resources, L.C. and E.A.; wrote the original draft of the paper, F.T. and G.M.; critical revision of the manuscript, L.C., G.B., G.M., E.A., and D.M.G.; supervised the experiments, L.C. and G.M. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was supported by the "University of Perugia fund for basic research 2014", project title "DNA methylation analysis on the adaptation of fungi (yeasts and phytopathogenic fungi) grown outside their preferred habitat".

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Acknowledgments:** The authors wish to thank M.V. Consalvi, L. Ceccarelli, L. Bonciarelli, and M. Orfei for technical assistance.

**Conflicts of Interest:** The authors declare no conflict of interest.

### **References**

