Arabidopsis Histone Methyltransferase SET DOMAIN GROUP8 Mediates Induction of the Jasmonate/Ethylene Pathway Genes in Plant Defense Response to Necrotrophic Fungi

As sessile organisms, plants have to endure a wide variety of biotic and abiotic stresses, and accordingly they have evolved intricate and rapidly inducible defense strategies associated with the activation of a battery of genes. Among other mechanisms, changes in chromatin structure are thought to provide a ﬂexible, global, and stable means for the regulation of gene transcription. In support of this idea, we demonstrate here that the Arabidopsis ( Arabidopsis thaliana ) histone methyltransferase SET DOMAIN GROUP8 (SDG8) plays a crucial role in plant defense against fungal pathogens by regulating a subset of genes within the jasmonic acid (JA) and/or ethylene signaling pathway. We show that the loss-of-function mutant sdg8-1 displays reduced resistance to the necrotrophic fungal pathogens Alternaria brassicicola and Botrytis cinerea . While levels of JA, a primary phytohormone involved in plant defense, and camalexin, a major phytoalexin against fungal pathogens, remain unchanged or even above normal in sdg8-1 , induction of several defense genes within the JA/ethylene signaling pathway is severely compromised in response to fungal infection or JA treatment in mutant plants. Both downstream genes and, remarkably, also upstream mitogen-activated protein kinase kinase genes MKK3 and MKK5 are misregulated in sdg8-1 . Accordingly, chromatin immunoprecipitation analysis shows that sdg8-1 impairs dynamic changes of histone H3 lysine 36 methylation at defense marker genes as well as at MKK3 and MKK5 , which normally occurs upon infection with fungal pathogens or methyl JA treatment in wild-type plants. Our data indicate that SDG8-mediated histone H3 lysine 36 methylation may serve as a memory of permissive transcription for a subset of defense genes, allowing rapid establishment of transcriptional induction.

To compensate for their sessile nature, plants have evolved intricate and diverse strategies enabling them to survive and adapt to a broad range of biotic stresses, including insect, herbivore, and pathogen attacks. Plant resistance or tolerance to biotic stresses is mediated via preexisting physical and chemical barriers as well as rapidly inducible defense mechanisms. Induction of basal defenses is triggered by the perception of conserved pathogen-associated molecular patterns, whereas more specific microbial effectors elicit genefor-gene resistance responses (Jones and Dangl, 2006).
In both cases, induction of the antimicrobial arsenal (Ferreira et al., 2007) depends in part on phytohormone signaling, including mediation by salicylic acid (SA), jasmonic acid (JA), and ethylene (ET; Glazebrook, 2005). In general, activation of SA-dependent responses, including systemic acquired resistance, is most efficient against biotrophic pathogen attacks, whereas responses mediated by JA and ET are more prominent following infection with necrotrophic pathogens. Transcriptome analysis of Arabidopsis (Arabidopsis thaliana) has demonstrated that the characteristic signaling network of each plant-attacker combination orchestrates very complex and wide-ranging transcriptional reprogramming, ultimately leading to increased plant resistance (Glazebrook et al., 2003;De Vos et al., 2005). Among the mechanisms capable of such rapid and broad reprogramming of gene expression, the role that chromatin remodeling plays in response to changing environmental cues has received little attention until now.
Histone variants and posttranslational modification (e.g. acetylation, methylation, phosphorylation) of histone tails form the so-called histone code (Strahl and Allis, 2000;Turner, 2000), in which chromatin remodeling establishes a rapid and reversible differential pattern of gene expression across the genome. In euchromatin, methylation of histone H3 lysine 4 (H3K4) and H3K36 is associated with transcriptional activation, whereas trimethylation of H3K27 (H3K27me3) is associated with gene silencing. These active and repressive methylation markers are established by the evolutionarily conserved SET , E(z), and Trithorax] domain protein of the Trithorax Group (TrxG) and the Polycomb Group, respectively. The role of TrxG/Polycomb Group factors in the regulation of plant homeotic genes and developmental processes has been extensively studied (for review, see Pien and Grossniklaus, 2007;He, 2009;Shen and Xu, 2009). More recent studies have shown that changes in histone methylation also occur under stress conditions. In Arabidopsis, changes in histone modification patterns (e.g. enrichment in H3K4me3) were observed at several genes in response to drought stress (Kim et al., 2008). H3K27me3 was decreased in the chromatin of COLD-REGULATED15A and GALAC-TINOL SYNTHASE3 during cold exposure and remained at low levels after returning to normal growth conditions (Kwon et al., 2009). ATX1, a TrxG member involved in H3K4 trimethylation (Alvarez-Venegas and Avramova, 2005), was shown to be necessary for the induction of WRKY70, a transcription factor gene in the SA pathway involved in defense against bacterial pathogens (Alvarez-Venegas et al., 2007).
The SET DOMAIN GROUP8 (SDG8; also named EFS, ASHH2, and CCR1) gene is a yeast SET2 and Drosophila ASH1 homolog, and its mutations cause pleiotropic plant phenotypes, including early flowering (Kim et al., 2005;Zhao et al., 2005), reduced organ size and enhanced shoot branching Xu et al., 2008), altered carotenoid composition (Cazzonelli et al., 2009), and reduced fertility (Grini et al., 2009). In sdg8 mutants, H3K36me2 and H3K36me3 levels were reduced not only at specific loci but also in global histone extracts, suggesting that SDG8 may have additional uncharacterized functions. In this work, we provide evidence that SDG8 plays crucial roles in plant defense against necrotrophic fungal pathogens through H3K36me3-mediated activation of a subset of genes within the JA/ET signaling defense pathway.

SDG8 Promoter Activity Is Induced by Wounding and Fungal Pathogen Infection
Previous reverse transcription-PCR analysis has shown that SDG8 is ubiquitously expressed in different plant organs, and in situ hybridization revealed a higher abundance of SDG8 transcripts in actively dividing tissues, such as the ovule primordia and shoot apical meristem (Zhao et al., 2005). Here, we examined spatial expression patterns in transgenic plants using a GUS reporter gene driven by 2,655 bp of the SDG8 promoter (SDG8p::GUS). T2 progeny from three independent primary transformants were examined for GUS expression by histochemical staining. All three transgenic lines display a largely similar expres-sion pattern, and hereinafter data from one representative line (line A) are shown. In 8-d-old seedlings, GUS activity was detected in root tips and stipules (Fig. 1, A and 1B). In true leaves, strong GUS activity was observed in hydathodes at the leaf margin (Fig. 1C) and at the base of cauline leaves from both primary and secondary stems (Fig. 1D). In flowers, GUS staining was observed in anther vasculature (Fig.  1E), the transmitting tract of the upper part of the pistil, and at the base of pistil following abscission of sepals and petals (Fig. 1F). These tissue-specific expression patterns are relatively distinct from the previous report using a shorter version of the SDG8 promoter fused with GUS (Cazzonelli et al., 2010) but are more similar to those reported for a GUS fusion at the 3# end of the SDG8 gene (Kim et al., 2005), indicating that multiple cis-elements exist in regulating SDG8 expression.
In addition, SDG8p::GUS expression was inducible in response to stress. Upon mechanical wounding, strong GUS activity was observed along the cut edge of tissues (Fig. 1G). When leaves of transgenic plants were spot inoculated with Botrytis cinerea or Alternaria brassicicola, both necrotrophic fungal pathogens, strong GUS staining was observed in the tissue bordering inoculation sites and in tissues more distant from the inoculation site (Fig. 1, H and I). Similar results were obtained in three independent transgenic SDG8p::GUS lines, and negative controls did not show obvious induction of GUS expression (Supplemental Fig. S1). Quantitative reverse transcription-PCR analysis showed that SDG8 expression was significantly induced (P , 0.05 in two-sided t test; Supplemental Table S1) following fungal infection of wild-type Columbia (Col) plants (Fig. 1J), confirming the results observed in the GUS reporter lines. The finding that SDG8 expression is inducible in response to fungal pathogens, together with the fact that pathogen-responsive genes are overrepresented in the previously generated list of down-regulated genes in sdg8 mutant seedlings (Xu et al., 2008;Supplemental Fig. S2), prompted us to investigate the role of SDG8 in pathogen defense. We compared the sdg8-1 mutant with wild-type plants in order to study their respective resistance to infection with the fungi B. cinerea and A. brassicicola.
Rosette leaves of 6-week-old soil-grown plants were challenged with B. cinerea by inoculation of a fungal spore suspension onto small holes made using a needle. Lesions caused by B. cinerea infection were readily visible on wild-type Col and sdg8-1 and sdg8-2 leaves at 3 d post inoculation (dpi). Nevertheless, these lesions developed more extensively in mutant plants, eventually affecting the entire leaf blade ( Fig. 2A). Previous studies have established that sdg8-1 and sdg8-2 are allelic mutants and that loss of function of SDG8 causes the previously described mutant phenotypes (Kim et al., 2005;Zhao et al., 2005;Dong et al., 2008;Xu et al., 2008;Cazzonelli et al., 2009;Grini et al., 2009). Consistently, the sdg8-1 and sdg8-2 mutants show similar disease symptoms ( Fig.  2A). Hereinafter, we concentrated on sdg8-1 for detailed studies. The size of lesions caused by B. cinerea was approximately three times larger in the sdg8-1 mutant compared with those on Col leaves (Fig. 2B). Enhanced fungal growth was apparent in sdg8-1 (Fig. 2C), and quantitative PCR (qPCR) analysis further confirmed higher levels of fungal multiplication in sdg8-1 compared with Col ( Fig. 2D; Supplemental Fig. S3A), which correlates with the increased severity of disease symptoms.
A. brassicicola inoculation was performed by application of a fungal spore suspension directly onto intact leaf surfaces. In Col, disease symptoms were limited to tiny necrotic spots at the site of inoculation. In contrast, sdg8-1 leaves displayed severe necrotic symptoms that turned a purplish color (Fig. 2E), with a dramatic 5-fold increase in average lesion diameter compared with Col at 5 dpi (Fig. 2F). Enhanced fungal growth in sdg8-1 was also apparent from qPCR analysis of infected leaf tissue ( Fig. 2G; Supplemental Fig. S3B).
Taken together, these results indicate that SDG8 is required for resistance to the fungal pathogens B. cinerea and A. brassicicola in Arabidopsis.
The JA/ET Defense Pathway Is Compromised in the sdg8-1 Mutant We next investigated the molecular mechanisms associated with reduced resistance of sdg8-1 plants to fungal infection. It has been reported that resistance to necrotrophic fungal pathogens in plants is strongly dependent upon the production of camalexin, an indole-type phytoalexin (Thomma et al., 1999). Therefore, we compared camalexin accumulation in sdg8-1 and Col plants in response to A. brassicicola infection. Prior to inoculation, basal levels of camalexin were similarly low in sdg8-1 and Col plants (Fig. 3A). Upon infection, sdg8-1 plants retained a strong capacity to accumulate camalexin, and levels in mutant plants were approximately two times higher than in Col plants at 5 dpi (Fig.  3A). The increased accumulation of camalexin at later stages of infection may reflect more extensive tissue colonization by A. brassicicola in the sdg8-1 plants. In any case, we did not detect any reduction in camalexin synthesis that would explain the reduced resistance to fungal infection observed in the sdg8-1 mutant.
Activation of the JA/ET-dependent defense pathway, also required for resistance to necrotrophic pathogens (Thomma et al., 1998;Glazebrook et al., 2003), results in induction of the downstream defense genes PDF1.2a and VSP2 (Fig. 3B). We analyzed PDF1.2a and VSP2 expression in order to determine if the JA/ET defense pathway response is perturbed in the sdg8-1 mutant. Interestingly, sdg8-1 plants showed a reduction in basal expression levels of PDF1.2a compared with Col, while VSP2 levels remained relatively unchanged ( Fig. 3C; Supplemental Table S2). In response to A. brassicicola infection, PDF1.2a and VSP2 induction was severely impaired in the sdg8-1 mutant, which is in sharp contrast to their strong induction in Col plants ( Fig. 3C; Supplemental Table S2). Similar defects in the induction of PDF1.2a and VSP2 expression were also observed in sdg8-1 following inoculation with B. cinerea (Supple- cinerea at 3 dpi before (left) and after (right) GUS staining. I, Leaf inoculated with A. brassicicola at 5 dpi before (left) and after (right) GUS staining. J, Relative level of endogenous SDG8 expression after B. cinerea (gray bars) or mock (white bars) inoculation. Data represent means 6 SD of triplicate determinations. Asterisks indicate significant differences between Col and sdg8-1 at P , 0.05 (two-sided t test). Similar results were obtained in two independent experiments (Supplemental Table S1).  Table S3). These results suggest that the JA/ET defense signaling pathway is compromised in sdg8-1 plants, likely resulting in their reduced capacity to defend against fungal pathogens.
SDG8 Is an Essential Activator Required for Induction of a Subset of Genes of the JA/ET Defense Pathway To determine more precisely which aspects of the JA/ET signaling pathway are affected in sdg8-1, JA levels in methanolic extracts of Col and sdg8-1 plants were measured before and after A. brassicicola inoculation. Basal JA levels were comparable in Col and sdg8-1 at 0 dpi, and a significant and similar increase in JA accumulation occurred at 5 dpi (Fig. 4A). Thus, JA biosynthesis/accumulation in response to fungal attack is unaffected in sdg8-1. We next examined gene expression profiles in response to exogenously applied methyl jasmonate (MeJA) in sdg8-1 and Col plants. ERF1 and MYC2, encoding two transcription factors, Figure 2. Comparison of pathogen-responsive phenotypes between mutant sdg8 and wild-type Col plants. A, Plant disease symptoms observed at 3 dpi on leaves inoculated with B. cinerea. Mock inoculation is shown as a control. White arrows indicate small lesions caused by B. cinerea on wild-type Col leaves. Note that more severe lesions occur in both sdg8-1 and sdg8-2 mutants. B, Mean leaf lesion diameter at 3 dpi with B. cinerea (n $ 20; 6SE). C, Visualization of fungal growth and cell death in Col and sdg8-1 leaves after trypan blue staining at 3 dpi with B. cinerea. D, Quantification of in planta growth of B. cinerea. qPCR was used to analyze the relative genomic DNA level of B. cinerea CUTINASE A compared with Arabidopsis ACTIN2 (Bo cut-A/At ACT). Data represent means 6 SD of triplicate determinations. E, Plant disease symptoms observed at 5 dpi on leaves inoculated with A. brassicicola. White arrows indicate small lesions observed on Col leaves. F, Mean leaf lesion diameter at 2 and 5 dpi with A. brassicicola (n $ 20; 6SE). G, Quantification of in planta growth of A. brassicicola. qPCR was used to analyze the relative genomic DNA level of A. brassicicola CUTINASE compared with Arabidopsis ACTIN2 (Alt cutab1/At ACT). Data represent means 6 SD of triplicate determinations. Asterisks indicate significant differences between Col and sdg8-1 at P , 0.05 (two-sided t test).
are known to control defense gene regulation in separate branches of the JA/ET signaling pathway (Fig.  3B). MYC2 plays a pivotal role in the JA signaling pathway and positively regulates JA-and wound/ insect-responsive genes (e.g. VSP2 and LOX3; Lorenzo et al., 2004). ERF1 plays a key role in integrating JA and ET signals and positively regulates the pathogenresponsive genes LOX2, PR4, and PDF1.2a (Lorenzo et al., 2003). Additionally, MYC2 and ERF1 exert antagonistic effects on each other's target genes and thus provide the means to intricately regulate pathway dynamics in response to varying JA and/or ET input signals (Boter et al., 2004;Lorenzo et al., 2004). In Col plants, all examined genes (including ERF1, MYC2, PDF1.2a, VSP2, LOX3, LOX2, and PR4) were rapidly induced, to varying extents, upon exposure to MeJA  Table S4). Remarkably, the increase in ERF1 and MYC2 transcript levels was significantly lower in sdg8-1 compared with Col plants, and an even more dramatic reduction was observed in their respective downstream target genes, PDF1.2a and VSP2, after MeJA treatment ( Fig. 4B; Supplemental Table S4). It is possible that the observed differences in inducible expression of JA/ET pathway genes is due to aberrant development and/or size differences between 6-week-old Col and sdg8-1 plants; however, similar results were also obtained following MeJA treatment of 10-d-old Col and sdg8-1 seedlings of comparable size in two independent experiments (Supplemental Table S5). These results establish that SDG8 is necessary for transcriptional activation of genes in both the ERF1 and MYC2 branches of the JA/ ET signaling pathway. Data represent means 6 SE of triplicate determinations. B, Simplified model for the regulation of plant defense networks in response to necrotrophic pathogen infection. Genes (in italics) involved in the JA and ET signaling pathways, as well as MAPK kinase genes and their corresponding phosphorylation cascade (arrows with black heads), are defined in the text. Positive (arrows with white heads) and negative (bars) regulations are depicted. C, Relative expression levels of PDF1.2a and VSP2 in sdg8-1 and Col leaves in response to A. brassicicola inoculation. Gene expression values are presented relative to average wild-type levels at time point 0 (set as 1). Data represent means 6 SD of triplicate determinations. Asterisks indicate significant differences between Col and sdg8-1 at P , 0.05 (two-sided t test). Similar results were obtained in two independent experiments (Supplemental Table S2). Endogenous JA levels were measured by UPLC-MS using sdg8-1 and Col leaves from plants inoculated with A. brassicicola. Data represent means 6 SD of triplicate determinations. B, Relative expression levels of JA/ET pathway genes in sdg8-1 and Col leaves in response to treatment with exogenously applied MeJA. Gene expression values are presented relative to average wild-type levels at time point 0 (set as 1). Data represent means 6 SD of triplicate determinations. Asterisks indicate significant differences between Col and sdg8-1 at P , 0.05 (two-sided t test). Similar results were obtained in two independent experiments (Supplemental Table S4). We hypothesized that JA treatment or pathogen attack may engender a massive transcriptional reprogramming in Arabidopsis through a global change in H3K36 trimethylation catalyzed by SDG8. We addressed this assumption by western-blot analysis of changes in H3K36me1 and H3K36me3 levels in Col plants following A. brassicicola infection or exogenous MeJA treatment. H3K36me1 was included in our analysis because this marker was previously shown to accumulate in sdg8 mutants (Xu et al., 2008). Our results indicate that under the examined stresses, H3K36me1 and H3K36me3 levels remain relatively constant (Supplemental Fig. S6), indicating that JA treatment or fungal infection does not affect global levels of H3K36 methylation in Arabidopsis.
We next investigated gene locus-specific levels of H3K36me1 and H3K36me3 in order to better understand the molecular mechanisms underlying SDG8mediated transcription activation of JA/ET defense pathway genes. Chromatin immunoprecipitation (ChIP) coupled with qPCR analysis was performed for PDF1.2a and VSP2, analyzing two regions of each gene. We chose these defense marker genes because they act at the most downstream steps of the JA/ET signaling pathway (Fig. 3B) and thus provide interesting models to test whether their expression is directly or indirectly associated with SDG8-mediated H3K36 methylation. In unchallenged plants, levels of H3K36me3 at the chromatin regions of both PDF1.2a and VSP2 were lower in sdg8-1 compared with Col ( Fig. 5; Supplemental Table S6). Upon fungal inoculation, the H3K36me3 levels were significantly increased in Col but were entirely unaffected in sdg8-1. In contrast, basal H3K36me1 levels were similar at PDF1.2a or slightly higher at VSP2 in sdg8-1 compared with Col, and fungal challenge decreased H3K36me1 levels at both PDF1.2a and VSP2 in Col but not in sdg8-1. In comparison, H3K27me3 at PDF1.2a and VSP2 were found at very low levels in both sdg8-1 and Col and were unchanged following fungal infection ( Fig. 5; Supplemental Table S6). The presence of only low levels of H3K27me3 at the chromatin of genes involved in defense against necrotrophic pathogens is likely to be beneficial for the plant, as the likelihood of silencing is reduced. We also examined H3K36me3, H3K36me1, and H3K27me3 levels in response to MeJA treatment. Similar to A. brassicicola infection, PDF1.2a and VSP2 displayed increased H3K36me3, reduced H3K36me1, and unchanged H3K27me3 levels in response to MeJA treatment in Col, and changes in Figure 5. Analysis of histone methylation at defense marker genes in response to pathogen infection in mutant sdg8-1 and wildtype Col plants. ChIP analysis was used to determine the relative levels of H3K36me3, H3K36me1, and H3K27me3 before (white bars) and 2 d after (black bars) A. brassicicola inoculation of 6-week-old Col and sdg8-1 plants at the indicated regions of PDF1.2a and VSP2. Data represent means 6 SD of triplicate determinations. Similar results were obtained in two independent experiments (Supplemental Table S6). Amplified regions (named a-d) are indicated below each gene schematic, which is represented by a white box for the coding region and gray boxes for the 5# and 3# untranslated regions.
H3K36me3 and H3K36me1 were undetectable in sdg8-1 (Supplemental Fig. S7; Supplemental Table  S7). Taken together, these results indicate that dynamic changes in H3K36 methylation at PDF1.2a and VSP2 occur in response to fungal infection and MeJA treatment and show that SDG8-mediated H3K36me3 is directly associated with transcriptional induction of these defense genes. As previously proposed (Xu et al., 2008), H3K36me1 may be catalyzed by a different enzyme, and defects in converting monomethyl to dimethyl/trimethyl markers could result in elevated levels of H3K36me1 observed at these loci in sdg8-1. So far, our results have established that SDG8 is directly implicated in the activation of defense marker genes that exert functions at downstream steps in the JA/ET defense pathway. To further assess the molecular function of SDG8 in pathogen defense, we investigated early genes in the defense signaling pathway. Several mitogen-activated protein kinases (MAPKs) activate transcription factors involved in regulating defense gene expression (Pitzschke et al., 2009). The roles of the MAPK kinases MKK4 and MKK5 in defense pathways are best characterized, and they act as positive regulators of the ET pathway in response to pathogen infection, while MKK3 acts downstream of JA to repress MYC2 expression (Takahashi et al., 2007;Pitzschke et al., 2009;Fig. 3B). MKK5 expression was shown to be down-regulated in the sdg8-1 and sdg8-2 mutants in a previous microarray screen (Xu et al., 2008). We used qPCR to investigate the expression of MKK5, MKK4, and MKK3 as well as MKK2 and MKK1, which are involved in cold and drought stress responses, respectively (Teige et al., 2004;Xing et al., 2008). While basal expression levels of MKK1, MKK2, and MKK4 were similar in sdg8-1 and Col (Supplemental Fig. S8; Supplemental Table S2), a significantly reduced level of MKK5 and increased level of MKK3 were observed in sdg8-1 ( Fig. 6A; Supplemental Table S2). MKK3 sequences are not present on the CATMA chip, and consequently, MKK3 expression data were not available from the previous microarray screen (Xu et al., 2008). Interestingly, in response to A. brassicicola infection, expression of MKK3 and MKK5 increased over the measured time period in Col but remained unchanged in sdg8-1 ( Fig.  6A; Supplemental Table S2), indicating a similar requirement of SDG8-dependent transcriptional regulation. In sdg8-1 plants, the high basal level of MKK3 was comparable to that of A. brassicicola-infected Col plants, and MKK3 expression in sdg8-1 remained unchanged following fungal inoculation ( Fig. 6A; Supplemental Table S2). In contrast, MKK1 and MKK2 expression was not responsive to fungal challenge, and the observed inducibility of MKK4 expression occurred in an SDG8-independent manner in both Col and sdg8-1 plants (Supplemental Fig. S8A; Supplemental Table S2). MeJA treatment did not affect MKK1, MKK2, or MKK4 expression levels in either Col or sdg8-1 mutant plants (Supplemental Fig. S8B; Supplemental Table S4). Consistent with their respective positions in the JA/ET defense signaling pathway (Fig.  3B), MKK3 but not MKK5 expression was induced by MeJA treatment in Col ( Fig. 6B; Supplemental Table S4). In sdg8-1, MKK3 expression levels remained constantly high and MKK5 levels remained constantly low following MeJA treatment ( Fig. 6B; Supplemental Table S4).
To gain insight into the SDG8-dependent induction of MKK3 and MKK5 expression following A. brassicicola infection, we analyzed H3K36 methylation of these loci at three different regions by ChIP coupled with qPCR. In unchallenged plants, decreased levels of H3K36me3 accompanied by increased levels of H3K36me1 were detected at MKK5, particularly toward the 3# end of the gene, in the sdg8-1 mutant compared with Col plants (Fig. 6C; Supplemental Table S6). This result is consistent with the previously observed pattern of down-regulation of SDG8-dependent genes. In agreement with the basal up-regulation of MKK3 expression in sdg8-1, H3K36me3 levels were increased and H3K36me1 levels were decreased at MKK3 chromatin regions in sdg8-1 compared with Col ( Fig. 6C; Supplemental Table S6). We speculate that SDG8 may share a limiting factor with another H3K36methyltransferase complex more extensively involved in H3K36 trimethylation at the MKK3 locus; consequently, SDG8 knockdown results in enhanced MKK3 transcription. Remarkably, despite the differences in basal H3K36 methylation observed in unchallenged plants, both MKK5 and MKK3 loci displayed increased H3K36me3 and decreased H3K36me1 in Col but not in sdg8-1 mutant plants in response to fungal infection ( Fig. 6C; Supplemental Table S6). Taken together, these results indicate that SDG8-mediated H3K36 methylation is directly involved in inducible expression of both MKK3 and MKK5 at early steps of the defense signaling pathway against fungal pathogens.

DISCUSSION
Previous studies have established that SDG8 and H3K36 methylation play important roles in several developmental processes (Kim et al., 2005;Zhao et al., 2005;Dong et al., 2008;Xu et al., 2008;Cazzonelli et al., 2009;Grini et al., 2009). In this study, we have demonstrated that SDG8 and H3K36 methylation are involved in the establishment of a chromatin state required for inducible defense against necrotrophic fungal pathogens.
sdg8-1 mutant plants display reduced resistance to the necrotrophic fungal pathogens A. brassicicola and B. cinerea. sdg8-1 plants contain normal levels of camalexin, an important compound involved in antimicrobial defense against necrotrophic pathogens. sdg8-1 plants also contain normal levels of JA but are compromised in the inducible expression of a subset of genes, including ERF1, PDF1.2a, LOX2, PR4, MYC2, LOX3, and VSP2, involved in the JA/ET signaling pathway for plant defense against necrotrophic pathogens. To date, few other chromatin modifiers have been found to regulate the JA/ET signaling defense pathway. Ectopic overexpression of the histone deacetylase gene HDA19 was shown to increase ERF1 expression and enhance plant resistance to A. brassicicola (Zhou et al., 2005), and knockdown of HDA6 resulted in impaired basal and JA-inducible expression of PDF1.2a and VSP2 (Wu et al., 2008). However, histone acetylation levels at these defense genes were not investigated; thus, it is unclear whether HDA19 and HDA6 are directly or indirectly involved in regulation of the JA/ET pathway. More recently, it was reported that loss of function of HISTONE MONO-UBIQUITINATION1 (HUB1) increases plant susceptibility to necrotrophic fungal pathogens; however, the basal and inducible expression of PDF1.2a was unaffected in hub1 mutant plants, and it was suggested that HUB1 regulates plant defense through an alternative pathway (Dhawan et al., 2009). Moreover, H2B ubiquitylation catalyzed by HUB1 has not been investigated with regard to plant defense against pathogens. The ATP-dependent chromatin-remodeling factor gene SPLAYED has been shown to be required for basal and inducible expression of PDF1.2a and VSP2 in plant defense against necrotrophic fungal pathogens (Walley et al., 2008). Our study here demonstrates that H3K36 methylation at both PDF1.2a and VSP2 is dynamic in response to fungal pathogen infection and MeJA treatment.
In wild-type plants, we uncovered the presence of a basal level of H3K36me3 at PDF1.2a and VSP2. Upon fungal pathogen infection or application of MeJA, the level of H3K36me3 increased, which positively correlates with the induction of PDF1.2a and VSP2  Tables S2 and S4). C, Relative levels of H3K36me3 and H3K36me1 before (white bars) and 2 d after (black bars) A. brassicicola inoculation of 6-week-old Col and sdg8-1 plants at the indicated regions of MKK3 and MKK5. Data represent means 6 SD of triplicate determinations. Similar results were obtained in two independent experiments (Supplemental Table S6). Amplified regions (named e-j) are indicated below each gene, which is represented by a white box for the coding region and gray boxes for the 5# and 3# untranslated regions. expression. In sdg8-1 mutant plants, basal levels of H3K36me3 were low, and most interestingly, increases in H3K36me3 and induced expression of PDF1.2a and VSP2 failed to occur in response to fungal infection or MeJA treatment. Changes in H3K36me1 levels were directly opposed to modifications in H3K36me3 in most cases examined. This is consistent with previous observations concerning different degrees of H3K36 methylation at floral repressor genes and supports the idea that SDG8 is a histone methyltransferase responsible for converting H3K36me1 to H3K36me2 and then H3K36me3 (Xu et al., 2008). Our results further suggest that the enzyme catalyzing H3K36me1 deposition is not involved in the response to fungal pathogen infection or MeJA treatment. The particular importance of SDG8 in plant pathogen defense was also highlighted by the observation that SDG8 expression is induced by fungal pathogen infection. Nevertheless, global levels of H3K36me3 in wild-type plants remained unchanged upon fungal pathogen infection or following MeJA treatment. Future experiments will be required to better understand how SDG8 is activated and/or recruited to specific genes in response to fungal infection or MeJA treatment.
Our work provides novel insights into the transcriptional regulation of MKK genes. Through phosphorylation cascades, MKKs play important roles at early steps of the SA, JA, and ET defense signaling pathways (Asai et al., 2002;Takahashi et al., 2007;Zhang et al., 2007), and there is some evidence suggesting that they contribute to pathogen resistance (Asai et al., 2002). Knowledge of the mechanisms regulating MKK transcription remains scarce. We demonstrated that inducible expression of both MKK3 and MKK5 is dependent on SDG8-mediated H3K36me3. Epigenetic regulation of these early-acting defense genes might be advantageous, providing a more efficient way to trigger downstream events of the signaling pathways; however, further experiments will be needed to determine to what extent the MKK3 and MKK5 transcriptional differences in the sdg8 mutant contribute to the reduced resistance against necrotrophic fungal pathogens. Strikingly, while MKK5, like the genes we characterized previously, showed reduced levels of expression and H3K36me3, MKK3 conversely showed increased transcript and H3K36me3 levels in the unchallenged sdg8-1 mutant compared with wild-type plants. This finding implies that in the absence of SDG8, another enzyme of yet unknown identity is activated/recruited to catalyze H3K36me3 at MKK3. The TrxG family in Arabidopsis is composed of 12 SDG genes; so far, the functions of six of these SDG genes have been examined, and only SDG25, in addition to SDG8, was shown to be involved in H3K36 methylation (Berr et al., 2009). Future experiments will be required to examine whether SDG25 and/or other uncharacterized SDG genes are involved in H3K36 methylation-associated regulation of MKK gene transcription.
The evidence provided in this study supports a model whereby SDG8 targets at least two distinct levels in the JA/ET signaling pathway in plant defense against necrotrophic fungal pathogens. SDG8 catalyzes H3K36me3 for basal and inducible expression of defense genes, including MKK5, involved very early in the signaling pathway, and PDF1.2a and VSP2, at later stages in the defense response. The reduced expression in sdg8-1 of ERF1 and MYC2, two genes encoding central transcription factors that control a wide spectrum of downstream defense genes, is consistent with the accepted positive and negative control exerted by MKK5 and MKK3 over ERF1 and MYC2, respectively. The implication of H3K36 methylation in the regulation of a subset of genes, rather than a single target gene within the JA/ET defense pathway, is consistent with the necessity of multiple gene networks allowing for enhanced efficiency in plant defense. Based on our study, we propose that SDG8-mediated H3K36me3 establishes a favorable chromatin environment that may be considered as a memory marker of active transcription and is highly beneficial to plant survival.

Biological Materials and Growth Conditions
The sdg8-1 and sdg8-2 mutants in the Arabidopsis (Arabidopsis thaliana) Col ecotype were described previously (Zhao et al., 2005). For pathogen treatments, mutants and wild-type Col plants were grown on soil under a 12-hlight/12-h-dark photoperiod in a growth chamber. Alternaria brassicicola (strain MUCL20297) and Botrytis cinerea (strain IMI169558) were grown and maintained as described previously (La Camera et al., 2005).

SDG8p::GUS Construct and Plant Transformation
The -2,685 to -3 bp region relative to the ATG start codon of SDG8 was amplified from wild-type genomic DNA by PCR using primers 5#-AATGC-TACCTGATTCAAAGC-3# and 5#-AATGAGACGCTTCTTAAGC-3#. The PCR product was cloned into the pCRII-TOPO (Invitrogen) vector and sequenced to confirm the absence of errors. A 2,655-bp HindIII-XbaI fragment of the SDG8 promoter was subsequently cloned before the GUS coding region, replacing the HindIII-XbaI 35S promoter fragment in the binary vector pBI121 (Clontech). The resulting plasmid was introduced into Agrobacterium tumefaciens and used to transform Col plants using the floral dip method (Clough and Bent, 1998). More than 30 independent transgenic lines were obtained, and three representative lines were used for in-depth analysis in this study.

Histochemical Assays and Microscopy
For GUS staining, freshly harvested plant material was collected, immediately vacuum infiltrated for 15 min in GUS staining buffer (Jefferson et al., 1987), and incubated at 37°C for 3 to 15 h. Plant material was cleared in 70% ethanol and observed directly. The results presented here were reproducibly obtained from more than three independent transgenic lines.
For trypan blue staining, leaves were harvested, stained, boiled for 1 min in lactophenol-trypan blue solution (Koch and Slusarenko, 1990), and cleared overnight in chloral hydrate solution (75 g of chloral hydrate dissolved in 30 mL of distilled water). Seedlings or organs of wild-type and mutant plants were examined using a Leica MZ12 dissecting microscope. Higher magnification images and histology sections were acquired with a Nikon Eclipse 800 microscope. Images were processed with Adobe Photoshop 6.0.

Pathogen Inoculation and JA Treatment
Inoculation with A. brassicicola and B. cinerea and MeJA treatments were performed on 6-week-old mature vegetative plants. For A. brassicicola disease assays, two 5-mL droplets of spore suspension (5 3 10 5 spores mL 21 water) were placed directly on the upper surface of the leaf. Prior to B. cinerea inoculation, two small holes were made with a needle on either side of the midrib before a 5-mL droplet of spore suspension (5 3 10 5 spores mL 21 potato dextrose broth medium) was placed on each hole. Mock inoculations, without fungal spores, were also performed. Plants were then incubated in air-tight translucent boxes to achieve saturating humidity. Typically, two inoculation sites per leaf were performed on four to six leaves of each plant.
For JA treatments, potted wild-type and mutant plants were placed into an air-tight translucent container for 24 h to allow for acclimation before treatment with 10 mL of MeJA (95% purity; Sigma-Aldrich) applied to cotton tips placed in the container.

In Planta Quantification of Pathogen Growth
A. brassicicola and B. cinerea growth in infected plants was determined by relative quantification of fungal and plant DNA by means of qPCR analysis. Total fungal and plant DNA was extracted as described (Gachon and Saindrenan, 2004) from 8-mm-diameter leaf discs centered on the inoculation site at various time points before and after pathogen inoculation. Typically, leaf discs were pooled from at least six individual plants for each time-point analysis. The relative quantity of A. brassicicola and B. cinerea was calculated according to the abundance of the respective fungal CUTINASE gene (loci ABU03393 and Z69264, respectively), relative to the Arabidopsis-specific ACTIN2 DNA measured by qPCR. Analyses were performed in triplicate. Primer sequences are detailed in Supplemental Table S8.

Camalexin Measurement
For camalexin measurements, inoculated leaves were pooled from six individual plants at each time point after A. brassicicola inoculation, frozen in liquid nitrogen, ground to a fine powder with a mortar and pestle, and dissolved in 5 volumes of ice-cold 80% methanol. After centrifugation at 10,000g for 20 min, the supernatants were separated on a HPLC system equipped with a RP C18 column (Novapak, 4 mm, 4.6 3 250 mm; Waters) using an increasing gradient of acetonitrile. Gradient conditions at a flow rate of 1 mL min 21 were as follows: 0% to 25% solvent B for 10 min, 25% to 60% for 30 min, 100% for 5 min, followed by 100% solvent A for 8 min (solvent A, 10% acetonitrile in water; solvent B, 100% acetonitrile). Camalexin and fluorescent compounds were measured with a 474 detector (Waters) set at 315-nm excitation and 405-nm emission. Sinapoyl-malate abundance does not vary upon infection in leaves and was used as an internal standard for recovery. Camalexin amounts were calculated using a standard curve established with purified camalexin and an extinction coefficient of 14,000.

JA Measurement
JA measurements were performed using ultraperformance liquid chromatography coupled to tandem mass spectrometry (UPLC-MS/MS). Four volumes of ice-cold 90% methanol containing 330 mM methyl-dihydro-JA (Sigma-Aldrich), serving as an internal standard, was added to 100 to 150 mg of frozen leaf powder in a screw-capped tube containing glass beads. Material was ground twice for 30 s with a Precellys 24 tissue homogenizer (Bertin Technologies). Homogenates were cleared by two successive centrifugations at 20,000g, and supernatants were saved for UPLC-MS analysis. Endogenous JA values were corrected according to methyl-dihydro-JA recovery. All analyses were performed using an Acquity UPLC system (Waters) coupled to a Quattro Premier XE triple quadrupole mass spectrometer equipped with an electrospray ionization source. Chromatographic separation was achieved using an Acquity UPLC BEH C 18 column (100 3 2.1 mm, 1.7 mm) and Acquity UPLC BEH C 18 precolumn (2.1 3 5 mm, 1.7 mm). The mobile phase consisted of (A) water:acetonitrile:formic acid (94.9:5:0.1, v/v/v) and (B) acetonitrile: formic acid (99.9:0.1, v/v). The column was first equilibrated in 100% A, then a linear gradient was applied for 10 min to reach 100% B, followed by an isocratic run with 100% B for 2 min, before returning to initial conditions (100% A) in 1 min. The total run time was 17 min. The column was operated at 28°C with a flow rate of 0.45 mL min 21 (sample injection volume of 3 mL). Nitrogen was used as the drying and nebulizing gas. The nebulizer gas flow was set to approximately 50 L h 21 , and the desolvatation gas flow was 900 L h 21 . The interface temperature was set to 400°C, and the source temperature was 135°C. The capillary voltage was set to 3.4 kV, and the cone voltage and the ionization mode (positive and negative) were optimized for each molecule. The selected ion recording MS mode was used to determine parent mass transition of JA (mass-to-charge ratio 209 in negative mode) and methyldihydro-JA (mass-to-charge ratio 227 in positive mode) with a cone voltage of 25 V for all the components. JA fragmentation was performed by collisioninduced dissociation with argon at 1.0 3 10 24 mbar using multiple reaction monitoring. The collision energy was set to 20 V, and the 209/59 transition was used for JA as described previously (Pan et al., 2008). The combination of chromatographic retention time, parent mass, and unique fragment ion analysis was used to selectively monitor JA. Low mass and high mass resolution was 15 for both mass analyzers, ion energies 1 and 2 were 0.5 V, entrance and exit potential were 2 and 1 V, and detector (multiplier) gain was 650 V. Data acquisition and analysis were performed with the MassLynx software (version 4.1; Waters).

RNA Isolation and Reverse Transcription
Total RNA was extracted from plant tissue with Trizol reagent (Molecular Research Center). Three tissue types were used in this study. For fungal infection of 6-week-old plants, leaf discs (8 mm diameter) centered on the inoculation site were pooled from at least six individual plants for each sample. The use of leaf discs of uniform size aims to minimize any possible effects introduced by size variations among individual plants and between the wild-type Col and sdg8 mutant. For MeJA treatment of 6-week-old plants, aerial tissues from six individual plants were collected for each sample. For MeJA treatment of 10-d-old seedlings, at least 100 individual seedlings grown in four small pots (7 cm 3 7 cm) were pooled for each sample. For cDNA synthesis, 2 mg of total RNA was treated first with 2 units of DNase I (Promega France) and then reverse transcribed in a 40-mL total volume with 2 mM oligo (dT) 20 , 0.5 mM deoxyribonucleotide triphosphates, 5 mM dithiothreitol, and 200 units of SuperScript III reverse transcriptase (Invitrogen). The resulting cDNA was used for qPCR analysis.

Histone Methylation Analysis
Histone methylation was analyzed using western blot and ChIP. These methods require more starting material than the previously described experiments; thus, leaf discs were pooled from at least 12 individual plants for each sample. Specific antibodies used in this study were anti-H3 (05-499; Millipore), anti-monomethyl-H3K36 (ab9048; AbCam), anti-trimethyl-H3K36 (ab9050; AbCam), and anti-trimethyl-H3K27 (07-449; Millipore). Westernblot analysis was performed as described previously (Xu et al., 2008), and ChIP was performed with the following modifications. Chromatin was sheared with a BIORUPTOR (Cosmo Bio Co.) sonicator twice for 15 min with a 50% duty cycle and high power output, in order to obtain 200-to 1,000bp DNA fragments. Chromatin was immunoprecipitated with specific antibodies together with protein A magnetic beads (Millipore). Following elution with Proteinase K (Invitrogen), DNA was recovered using Magna ChIP spin filters (Millipore). ChIP experiments using protein A magnetic beads without the addition of antibody were carried out as negative controls. The resulting ChIP DNA was subjected to qPCR analysis.
Quantitative Real-Time PCR Analysis qPCR analyses were performed on cDNA or ChIP products with genespecific primers designed using the Light Cycler Probe Design 2 program (Roche) in a final volume of 10 mL of SYBR Green Master Mix using a Light Cycler 480 II instrument (Roche) according to the manufacturer's instructions.
For qPCR analysis of cDNA products, several conventionally used reference genes were evaluated for their respective stability under our experimental conditions using geNorm (Vandesompele et al., 2002) and Norm Finder (Andersen et al., 2004), and based on these results, the housekeeping genes EXP (At4g26410), GAPDH (At1g13440), and TIP41 (At4g34270) were selected for use as internal references. After qPCR amplification, melting curve analysis was performed to verify the amplification of a single PCR product. The target gene expression level was normalized against internal reference genes, averaged over triplicate determinations, and shown as a relative value.
Gene-specific primer sequences used for qPCR of cDNA or ChIP products are listed in Supplemental Table S8.

Supplemental Data
The following materials are available in the online version of this article.
Supplemental Figure S1. Stress-induced expression of SDG8p::GUS in independent transgenic lines.
Supplemental Figure S2. Gene ontology analysis of differentially expressed genes in sdg8-1 seedlings.
Supplemental Figure S3. Comparison of fungal pathogen growth in infected leaves of mutant sdg8-1 and wild-type Col plants.
Supplemental Figure S4. Comparison of defense gene expression in response to B. cinerea infection in mutant sdg8-1 and wild-type Col plants.
Supplemental Figure S5. Comparison of JA/ET pathway gene expression in response to MeJA treatment in mutant sdg8-1 and wild-type Col plants.
Supplemental Figure S6. Western-blot analysis of global H3K36 methylation in response to A. brassicicola inoculation or MeJA treatment.
Supplemental Figure S7. Histone methylation at JA/ET pathway defense genes in response to MeJA treatment in mutant sdg8-1 and wild-type Col plants.
Supplemental Figure S8. Expression of MKK genes in response to A. brassicicola inoculation and MeJA treatment in mutant sdg8-1 and wildtype Col plants.
Supplemental Table S1. Expression of SDG8 in 6-week-old Col plants upon B. cinerea infection.
Supplemental Table S2. Comparison of the expression and induction of different genes between 6-week-old Col and sdg8-1 plants upon A. brassicicola infection.
Supplemental Table S3. Comparison of the expression and induction of different genes between 6-week-old Col and sdg8-1 plants upon B. cinerea infection.
Supplemental Table S4. Comparison of the expression and induction of different genes between 6-week-old Col and sdg8-1 plants upon MeJA exposure.
Supplemental Table S5. Comparison of the expression and induction of different genes between 10-d-old Col and sdg8-1 plants upon MeJA exposure.
Supplemental Table S6. Comparison of histone methylation levels at different regions of selected genes between 6-week-old Col and sdg8-1 plants upon A. brassicicola infection.
Supplemental Table S7. Comparison of histone methylation levels between 6-week-old Col and sdg8-1 plants upon MeJA exposure.
Supplemental Table S8. List of primers used in this study.