**4. Discussion**

The study of changes in the level of genome methylation mostly accompanied plant development and plant response to biotic and abiotic stressors. In the case of environmental stresses, they can affect methylation in two ways, causing hyper- or hypomethylation. For example, di fferences in plant methylomes have been found during maize leaf development [24] and precede transcriptional activation of genes leading to cell division and meristem growth in potatoes [25]. Also, endosperm and embryo development [26], vernalization, and fruit ripening are a ffected by DNA methylation [27]. For example, rice blight pathogen resistance is related to genomic hypermethylation and hypomethylation of resistance-related genes [28]. Also, Infection of tobacco with tobacco mosaic virus (TMV)revealed an increase in genomic methylation and hypomethylation of resistance-related leucine-rich-repeat (LRR)-containing loci [29]. The data thus sugges<sup>t</sup> that DNA methylation a ffects the plant genome not only at the global level but also at very specific sites, such as individual genes [30]. The identification of di fferentially methylated CCGG sites within pathogenesis-related (PR) genes: β*-1,3-glucanase* and *chitinase*, the time course of their modification, and the correlation with their expression profile in plants remain largely unexplored. Here we analyzed the methylation state of all CCGG sites of three flax genes: Two β*-1,3-glucanase* (*1* and *2*) and *chitinase* genes. It was found that three CCGG sites of the β*-1,3-glucanase 1* gene (located in promoter, exon, and intron regions) and *chitinase* were changed upon infection and these were further analyzed. The tetramer methylation and expression of β*-1,3-glucanase 2* change slightly upon infection, and served for comparison. The β*-1,3-glucanase 1* and *chitinase* appeared to be strongly activated upon pathogen infection. The analysis revealed that the level of unmethylated sites CCGG ranged from 0 to 19%, CCmGG methylation occurred in 75% on average, and the CmCmGG methylation level was around 15% of all investigated CCGG sites. In contrast to the developing maize leaf where *HpaII*/*MspI* restriction analysis of the CCGG site revealed that almost half (48.1%) of the loci were unmethylated, full CG methylation was represented by 36.7% and CHG hemimethylation by 15.2% of all investigated CCGG sites [31].

As mentioned, two types of CCGG sites in β*-1,3-glucanases* have been detected: Those whose methylation status is unchanged and those with a highly modified cytosine state. The profile of the latter was changing upon plant growth and progression of the infection. It is in agreemen<sup>t</sup> with several observations that plants as sessile organisms have developed prompt response mechanisms to react to rapid environmental changes. It is interesting that the CCGG site methylation level induced by both pathogenic and non-pathogenic *Fusarium* strains is time dependent. After artificial inoculation with both *Fusarium* strains, the methylation pattern of the pathogen-responsive β*-1,3-glucanase* and *chitinase* genes undergoes rapid (6–24 h after inoculation) and dynamic changes in tetramer methylation. It was found that both strains induced a similar time-dependent profile of tetramer methylation but they di ffer significantly at the beginning and at the end of the profiles. Far earlier pulse demethylation of the β*-1,3-glucanase 1* gene (promoter) occurs and it is maintained longer upon non-pathogenic strain treatment than in the case of the pathogenic one. This suggests that the same CCGG site within the gene might be di fferentially methylated. Similar to this was the observation that in two cases, two di fferent CCGG sites within the same gene were found to be di fferentially methylated. In the first one, a conserved C-terminal domain CTC-interacting domain-encoding gene (human ATAXIN2 orthologue) was hypermethylated in two exons, and in the second case a gene encoding the protein kinase PK/UbiC was di fferentially methylated at two consecutive CCGG sites in an exon and an intron [31].

The *chitinase* gene's tetramers located at the beginning and end of the coding sequences (exons 1, 7, 8) are also pulse demethylated. Notably, in both genes the exon peak demethylation is reached earlier upon pathogen infection. These findings agree with observations that in maize the majority of the di fferentially methylated CCGG sites that map to a gene lie within an exon and they are not distributed equally throughout the gene body. The highest number of di fferentially methylated sites was in the first 10% and the last 20% of the gene body. In addition, di fferential methylation of the promoter and 5' part of the gene anticorrelated with the gene expression while di fferential methylation of the central and 3' part of the gene body or sequences downstream of the gene was unrelated to the gene expression. [31].

In *Arabidopsis thaliana*, promoter-specific methylation occurs in less than 5% of genes, most of which are under tissue-specific control. A surprising result of genome-wide methylation profiling revealed that about one-third of all genes contain CG-specific genic or body methylation patterns within their transcribed regions which are highly expressed [32]. However, it was indicated that in *A. thaliana,* genes with the highest and the lowest transcription level were the least methylated, while moderately transcribed genes were the most frequently methylated [33]. The correlation between expression levels and DNA methylation levels in the gene body region was also observed in *Populus trichocarpa* where the hypermethylation of genes caused the reduced level of their expression while the hypomethylation of genes led to the increased level of their expression [34]. Due to the lack of unambiguous literature data, there are still many questions about the gene body methylation function and its correlation with gene expression.

The levels of β*-1,3-glucanase 1* and *chitinase* gene expression are also significantly altered, while in the case of β*-1,3-glucanase 2* only slight changes were noted. This is consistent with the hypothesis that pathogen infection changes the methylation level of plant genomic DNA and leads to alterations in gene transcription. The peak of demethylation might be a signal for the alteration in gene expression induced by the non-pathogenic strain. In the case of pathogen infection the expression peak lags behind the gene demethylation: Maximal transcript accumulation was detected 24–36 h after the onset of peak methylation.

The result might sugges<sup>t</sup> that flax treatment with the non-pathogenic strain prepares plants for reaction after contact with pathogen. Thus, following non-pathogenic pretreatment, plants were infected with the pathogenic strain, and gene expression was analyzed. Indeed, upon infection, the highest gene expression was reached far earlier in the case of plants pretreated with the non-pathogenic strain. The interesting fact is that the level of gene induction by the pathogen is 2-fold lower in the case of pretreated plants, but even so, they are efficiently protected against infection. While plants infected with the pathogen started to die (at 48 h), the pretreated plants grew normally.

We measured the expression of genes involved in this process and found that both *AGO4* and *RDR2* are activated upon non-pathogenic *Fusarium* treatment while they are suppressed by the pathogenic strain. The gene induction and thus potent target methylation appeared later (at 48 h) than β*-1,3-glucanase* and *chitinase* gene demethylation and their expression increased. We deduced that non-pathogenic strain infection induced at first demethylation and gene up-regulation and at the end, after reaching a certain level of transcript accumulation, activated the re-methylation pathway and thus gene silencing. Temporary demethylation/remethylation serves as memorized experience and helps the plant to overcome pathogenic infection by a mechanism that is similar to immunization.

It was detected that the plant-specific methyltransferase expression profile showed significant suppression upon infection with both *Fusarium* strains. However, the pattern of DNA methylation is the result of co-operative or competing interactions of the methyltransferases and the silencing pathways which involves repressor of silencing (ROS1) and demeter (DME), which encode two closely related DNA glycosylase domain proteins [21]. In *Arabidopsis thaliana*, ROS1 and DME are required for release of transcriptional silencing of a hypermethylated transgene and activate the maternal expression of two genes silenced by methylation [35]. The expression profile of these genes in infected flax differs depending on *Fusarium* strain used and growth time. However, the ratio between the level of methyltransferase and repressor of silencing pathway gene expression promote CCGG sites' demethylation.

In summary, two contrasting effects on the levels of methylation in flax were detected upon infection of both *Fusarium* strains: Genome-wide hypermethylation and hypomethylation of two resistance-related genes, which resulted in an increase of their expression. Plants pretreated with the non-pathogenic strain memorized the hypomethylation pattern and then reacted more efficiently upon pathogenesis (Figure 5). However, the changes in methylation profile and thus the regulation of PR-related gene expression were not the only mechanisms of plant response to stressors. For example, it is known that the argonaute family proteins are core constituents of RNA interference pathways in eukaryotes and mediate gene expression. Guided by sRNAs, AGO proteins bind target sequences through base pairing and following recruitment of cofactors mediate post-transcriptional or transcriptional gene silencing. AGO proteins are also involved in epigenetic modifications of chromatin. In *Arabidopsis thaliana*, AGO4 guided by 24-nt small interfering RNAs (siRNA) recruits DNA methyltransferase for de novo DNA methylation at target loci [36]. Likewise, AGO1 with siRNAs facilitates H3K9 methylation by recruiting H3K9 methyltransferase [37].

**Figure 5.** Hypothetical model of priming induced by non-pathogenic *Fusarium* strain. Flax treatment with non-pathogenic *Fusarium* strain (Fo47) results in changes in DNA methylation pattern of β*-1,3-glucanase* and *chitinase* genes which leads to their earlier expression during infection by pathogenic *Fusarium* strain (Foln3).

Very recently, it was reported that direct binding of AGO1, guided by a 21-nt siRNA, to the chromatin of active genes promotes their transcription. Various stimuli, including plant hormones and stresses, specifically trigger AGO1 guided by siRNAs to bind stimulus-responsive genes [38]. Here, we found strong up-regulation of *AGO1* expression in *Fusarium*-treated plants. It thus suggests that in addition to the transcriptional regulation of gene expression, microRNAs and siRNAs, epigenetics occurs as an important mechanism involved in transcriptional regulation of the plant stress response. These phenomena may contribute to the adaptation of plants to the environment and stress situations.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2076-2607/7/12/589/s1, Figure S1: A simplified scheme of the treatment of flax plants with the pathogenic and non-pathogenic strain of *F. oxysporum* and the treatment of flax plants with pathogenic strain after sensitizing with a non-pathogenic strain; Figure S2: Schematic diagram of the position of all CCGG sites and the relative position of the primers in β*-1,3-glucanase* 1, β*-1,3-glucanase 2* and *chitinase* genes that were used to perform qPCR to estimate methylation levels; Figure S3: Changes in DNA methylation pattern of β*-glucanase 1* and *chitinase* genes in flax seedlings treated with non-pathogenic or pathogenic strains of *Fusarium oxysporum* and after sensitizing effects of the non-pathogenic strain treated with pathogenic Foln3 at 6, 12, 24, 36 and 48 h after inoculation. The analysis of the third position in the exon, the first in the intron and the third in the promoter of β*-1,3-glucanase 1*, and the first, seventh and eight positions in the exon of *chitinase* was performed by the digestion of genomic DNA by the restriction enzymes *HpaII* and *MspI* and then the real-time PCR reaction. C, control; N-P, non-pathogen; P, pathogen; CCGG, sequence where cytosines are not methylated; CCmGG, sequence where internal cytosine is methylated; CmCmGG, sequence where two cytosines are methylated. CCGG positions in DNA are counted from ATG. The data represent the mean from three independent experiments. Asterisks mark statistically significant differences (*p* < 0.05) between the treated samples and their own controls; Figure S4: Phenotypic changes in flax after priming of the non-pathogenic strain of *Fusarium oxysporum* treated with pathogenic Foln3. Flax plants from in vitro culture were grown for 2 days on PDA medium or on PDA medium with a non-pathogenic (Fo47) or pathogenic (Foln3), and then were transferred to control PDA medium or PDA medium with Foln3 or Fo47 for 6 days. Analyzed combinations: control to control, control to Foln3, Fo47 to Foln3, Fo47 to Fo47, Foln3 to Foln3; Table S1: Primer sequences designed for real-time PCR: (A) PR genes (β*-1,3-GLU1*, β*-1,3-glucanase 1*; β*-1,3-GLU2*, β*-1,3-glucanase 2*; *CHIT*, *chitinase*), genes involved in chromatin modification (*CMT1*, *chromomethylase 1*; *CMT3*, *chromomethylase 3*; *ROS1*, *repressor of silencing 1*; *DME*, *demeter*; *AGO1*, *argonaute 1*; *AGO4*, *argonaute 4*; *RdRp2*, *RNA-dependent RNA polymerase 2*) and *actin* gene. (B) DNA methylation. E3-G1 (the third position in the exon of β*-1,3-glucanase 1*); I1-G1 (the first position in the intron of β*-1,3-glucanase 1*); P1-G1 (the third position in the promoter of β*-1,3-glucanase 1*); E1-G2 (the first position in the exon of β*-1,3-glucanase 2*); E11-G2 (the eleventh position in the exon of β*-1,3-glucanase 2*); I1-G2 (the first position in the intron of β*-1,3-glucanase 2*); E1-CH (the first position in the exon of *chitinase*); E7-CH (the second position in the exon of *chitinase*); E8-CH (the eighth position in the exon of *chitinase*).

**Author Contributions:** W.W. performed most of the experiments and wrote the manuscript. A.B. performed the analysis of the expression levels of genes involved in DNA methylation. K.K. performed statistical analyses and graphics. A.K., M.P., and J.S. participated in study design, coordination, and in writing the manuscript. All of the authors read and approved of the final version of the manuscript.

**Funding:** This work was supported by a gran<sup>t</sup> from the National Science Centre (NCN) [grant number 2013/11/N/NZ1/02378 and 2012/06/A/NZ1/00006].

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