The Polycomb-Group Repressor MEDEA Attenuates Pathogen Defense

Pathogen inoculation in Arabidopsis thaliana activates the expression of the imprinted gene MEDEA, a component of the PRC2 complex, which hinders defense against pathogens.

Abstract

Plants recruit positive and negative regulators for fine tuning the balance between growth and development. Negative regulators of pathogen defense generally modulate defense hormone biosynthesis and signaling. Here, we report a mechanism for attenuation of the defense response in Arabidopsis (Arabidopsis thaliana), which is mediated by the polycomb-group repressor MEDEA (MEA). Our results showed that pathogen inoculation or exogenous application of salicylic acid, methyl jasmonate, or the bacterial 22-amino acid domain of flagellin peptide induces the expression of MEA. MEA expression was higher when plants were inoculated with the avirulent strain of Pseudomonas syringae pv. tomato (Pst) carrying the AvrRpt2 effector (Pst-AvrRpt2) compared to the virulent Pst strain. MEA remains suppressed during the vegetative phase via DNA and histone (H3K27) methylation, and only the maternal copy is expressed in the female gametophyte and endosperm via histone and DNA demethylation. In contrast, Pst-AvrRpt2 induces high levels of MEA expression via hyper-accumulation of H3K4me3 at the MEA locus. MEA-overexpressing transgenic plants are susceptible to the fungal pathogen Botrytis cinerea and bacterial pathogens Pst and Pst-AvrRpt2, whereas mea mutant plants are more resistant to bacterial pathogens. AvrRpt2-mediated immunity requires the function of RESISTANCE TO P. SYRINGAE2 (RPS2) in Arabidopsis. Using transcriptional analysis and chromatin immunoprecipitation, we established that MEA directly targets RPS2 and suppresses its transcription. We screened an Arabidopsis cDNA library using MEA as the bait in a yeast two-hybrid assay and identified DROUGHT-INDUCED19, a transcription factor that interacts with MEA and recruits it at the RPS2 promoter. The results identified a previously unknown mechanism of defense response attenuation in plants.

Plants are capable of defending themselves from pathogen attack with the help of well-elaborated immune machinery. Plants possess both constitutive and inducible immune systems (Spoel and Dong, 2012). By virtue of its cellular content, plants impose a constitutive barrier to many pathogens. The inducible immune system is activated upon the recognition of pathogens. Recognition of conserved pathogen-/microbe-associated molecular patterns activates pattern-triggered immunity (PTI). Pathogen-derived molecules such as bacterial flagellin, elongation factor-TU (EF-TU) lipo-oligosaccharides, fungal cell wall chitin, glucan, and glycoproteins of oomycetes are sources of patterns for activating PTI (Zhang and Zhou, 2010). Successful pathogens release effector molecules to suppress PTI. During the coevolution of plants and microbes, plants often evolved recognition systems for certain effector molecules to activate strong immune response (effector-triggered immunity; ETI) that renders the pathogen incompatible with the host (Jones and Dangl, 2006; Spoel and Dong, 2012). For ETI activation, the effectors (avirulent factors) function in combination with resistance (R) genes of the host. In the absence of cognate R genes, the avirulent factors contribute to effector-triggered susceptibility (Jones and Dangl, 2006; Kim et al., 2009; Deslandes and Rivas, 2012). Both PTI and ETI involve the activation of MAPK signaling, the accumulation of reactive oxygen species and hormones, and the biosynthesis of antimicrobial compounds such as phytoalexins and peptides (Zhang and Zhou, 2010). ETI is a magnified form of PTI, which results in the activation of defense responses to a much higher level (Jones and Dangl, 2006). Additionally, ETI is often associated with the hypersensitive response (HR), a rapid programmed cell death at the pathogen invasion site (Morel and Dangl, 1997). HR helps in restricting the growth of pathogens and signaling for systemic acquired resistance. Plant hormones such as salicylic acid (SA), ethylene (ET), and jasmonic acid (JA) play central roles in activating both PTI and ETI (Robert-Seilaniantz et al., 2011; Pieterse et al., 2012).

Plants activate defense at the cost of growth and development (Heil and Baldwin, 2002; Tian et al., 2003; Huot et al., 2014). With limited resources, plants must balance the trade-off between growth and defense. Signaling cross talk among plant hormones plays fundamental roles in maintaining this balance. Hormones like auxin, gibberellins, cytokinins, brassinosteroids, and abscisic acid promote growth while limiting the defense output (Denancé et al., 2013; Huot et al., 2014). These hormones are implicated in functioning as antagonists to defense signaling activated by SA or ET/JA pathways. In addition, SA- or ET/JA-mediated defense responses are also regulated by proteins that limit the biosynthesis of hormones and signaling pathways (Frye et al., 2001; Jirage et al., 2001; Shah et al., 2001; Journot-Catalino et al., 2006; Zhang et al., 2006a; Swain et al., 2011; Giri et al., 2017). Here, we report a mechanism of attenuation of the immune response mediated by a polycomb-group (PcG) repressor protein MEDEA (MEA). PcG proteins regulate gene expression by chromatin modification. These modifications lead to stable transcription silencing, which can be inherited through many mitotic cell divisions (Margueron and Reinberg, 2011; Derkacheva and Hennig, 2014; Grossniklaus and Paro, 2014). PcG proteins function as large protein complexes. Plants contain two major PcG protein complexes, Polycomb Repressive Complex1 (PRC1) and PRC2. Both PRC1 and PRC2 complexes work together for stable transcriptional silencing of target genes. PRC2 methylates H3 at Lys-27 to induce epigenetic silencing, whereas PRC1 identifies and binds to these modifications to induce structural changes in chromatin (Köhler and Hennig, 2010; Kalb et al., 2014). MEA belongs to PRC2 and has a SET domain for methyltransferase activity (Grossniklaus et al., 1998). Our results show that pathogen inoculation enhances MEA expression, and enhanced MEA expression limits the growth of pathogens.

This observation is important because the expression of MEA is tightly controlled by developmental cues. MEA is an imprinted gene, for which only the maternal copy expresses in the female gametophyte and endosperm (Grossniklaus et al., 1998; Kinoshita et al., 1999). MEA remains repressed throughout the vegetative stage and in floral buds; its mRNA starts appearing in unpollinated siliques having female gametophytes and continues to express until seed maturation. Our results identified a mechanism of MEA activation during pathogenesis, i.e. accumulation of H3K4me3 at the MEA locus, which is distinct from its developmental activation.

RESULTS

MEA Expression Is Derepressed upon Activation of the Defense Response

Though MEA is epigenetically silenced throughout the vegetative phase, we observed its transcript abundance while analyzing pathogen-induced transcriptome profiles generated in our laboratory and by others (Lewis et al., 2015). To experimentally validate this observation, we treated wild-type Arabidopsis (Arabidopsis thaliana; Col-0) leaves with the virulent pathogen Pseudomonas syringae pv. tomato DC3000 (Pst) and an avirulent strain of Pst that carried the AvrRpt2 effector (Pst-AvrRpt2) and monitored MEA transcript accumulation by reverse transcription quantitative PCR (RT-qPCR). We detected a high level of MEA transcripts within 6 h of Pst-AvrRpt2 inoculation, which further increased until 12 to 24 h postinoculation (hpi; Fig. 1A). The virulent pathogen also enhanced MEA expression but to a lower level than the avirulent pathogen in the early hours (Fig. 1A). However, MEA expression induced by the virulent pathogen was enhanced in the late hours. The results suggested an association of MEA expression with defense response activation. The observation was further validated by the MEA promoter activity and by analyzing MEA expression after induction of the defense response by chemicals. MEA promoter-driven GUS reporter expression (MEA:GUS), which was barely detectable in mock-inoculated leaves of Arabidopsis, was significantly enhanced after inoculating with Pst-AvrRpt2 (Fig. 1B). Induction of defense by salicylic acid (Fig. 1C), flg22 (Supplemental Fig. S1) or methyl jasmonate (MeJA; Fig. 1D) also induced MEA expression in Arabidopsis leaves. The other members of the transcriptional repressor complex in which MEA belongs include MULTICOPY SUPPRESSOR OF IRA1 (MSI1) and FERTILIZATION-INDEPENDENT ENDOSPERM (FIE; Köhler et al., 2003). However, pathogen (Pst-AvrRpt2) inoculation failed to induce MSI1 or FIE significantly (Supplemental Fig. S2, A and B). Thus, defense activation specifically induced MEA of the PcG core complex. These results demonstrated that activation of the defense response overrides silencing of MEA in the vegetative tissue.

Figure 1.

MEA expression after pathogen, SA, and MeJA treatment. A, Relative abundance of MEA mRNA in wild-type (WT; Col-0) plants after mock, Pst, and Pst-AvrRpt2 (1 × 10⁶ CFU/mL) inoculation. B, 35S:GUS or MEA:GUS activity in Arabidopsis leaves after Pst-AvrRpt2 (1 × 10⁶ CFU/mL) or MgCl₂ infiltration. C, MEA mRNA accumulation after SA treatment (0.5 mm spray). D, MEA expression after 5µm MeJA treatment. Each bar represents mean ± sd (n = 3). *P < 0.05 and **P < 0.001 indicate the mean values that are significantly different from 0-h samples or respective mock samples as determined by Student’s t test. Experiments were repeated at least two times with similar results.

Homozygous mea-6 Mutant Plants Are Resistant to Virulent and Avirulent Bacterial Pathogens

Loss-of-function mutants of MEA are embryonic lethal (Grossniklaus et al., 1998) and thus, it is difficult to obtain homozygous mea mutants by conventional diploid genetic methods. However, the female gametophyte lethal mea mutant allele can normally be transmitted through the male gamete/gametophyte (pollen). Therefore, by exploiting a haploid genetics approach it is possible to obtain mea haploid progeny by producing paternal haploids (Ravi et al., 2014). To generate the mea-6 homozygous mutant (C24 ecotype), we crossed heterozygous MEA/mea-6 plants that harbor a point mutation at the MEA locus (Guitton et al., 2004) to the haploid inducer GFP-tailswap line as the female parent (Ravi et al., 2014). MEA functions in seed development by regulating endosperm development. In the case of seeds with a haploid embryo carrying the mutant allele of MEA, the endosperm receives two functional copies of MEA via the haploid inducer female parent. A fraction of the resultant F1 seeds are viable and develop into haploid plants and subsequently into homozygous mea-6 diploids (doubled haploids). A majority of seeds from the doubled haploid mea-6/mea-6 plants were dead, but a fraction (213/676) were viable and among the viable seeds around 30% (78/213 seeds) were late germinating (up to a week delay) in contrast to wild-type seedlings (Supplemental Fig. S3, A and B). These late-germinating seedlings show aberrant phenotypes during early vegetative growth until 2 to 3 weeks postgermination, as shown in Supplemental Figure S3, C and D. However, these plants recover later and regain normal morphology prior to bolting. Only wild-type-looking mea-6/mea-6 plants as shown in Supplemental Figure S3E were used for pathogen inoculation experiments. To investigate the possible role of MEA in disease defense we studied pathogen growth and defense responses in the mea-6 homozygous mutant and wild type (wild-type) C24 plants. Compared to wild-type plants, mea-6 mutants showed a higher level of resistance against the virulent strain of Pst (Fig. 2A) and the avirulent strain of Pst carrying the AvrRpt2 effector (Pst-AvrRpt2; Fig. 2B). Lower bacterial loads in the mea-6 mutant also resulted in a reduced level of disease symptoms in these plants. Hyper-resistance of the mea-6 mutant was supported by enhanced expression of PATHOGENESIS-RELATED GENE1 (PR1) in these plants compared to the corresponding wild-type plants (Fig. 2, C and D). We also observed higher SA accumulation in both Pst- and Pst-AvrRpt2-inoculated leaves of the mea-6 mutant than in wild-type plants (Fig. 2E). Also, mea-6 mutants showed enhanced HR-associated cell death (Fig. 2F) and H₂O₂ accumulation (Fig. 2G) after Pst-AvrRpt2 inoculation. The HR as measured by ion leakage was significantly more in mea-6 mutants compared to wild-type plants upon Pst-AvrRpt2 inoculation (Fig. 2H). The results suggested that MEA function may be associated with the susceptibility toward pathogens.

Figure 2.

Bacterial numbers and defense responses in mea-6 and C24 (wild-type) plants. A, Pst numbers and disease symptoms at 3 days post-inoculation (dpi). B, Pst-AvrRpt2 numbers and disease symptoms at 3 dpi. C, PR1 expression in wild-type (WT) and mea-6 plants after Pst inoculation. D, PR1 expression in wild-type and mea-6 plants after Pst-AvrRpt2 inoculation. E, Total SA (free SA + SA-glucoside) content in leaves of wild-type and mea-6 plants after 36 h of mock, Pst, or Pst-AvrRpt2 inoculation. F, HR-induced cell death after Pst-AvrRpt2 inoculation. Samples were harvested at 15 hpi for staining with trypan blue. G, DAB staining for H₂O₂ accumulation at 15 hpi with Pst-AvrRpt2. H, HR-induced ion leakage in wild-type and mea-6 plants after Pst-AvrRpt2 inoculation (10⁷ CFU/mL). Pathogens were inoculated at 5 × 10⁵ CFU/mL. Each bar represents mean ± sd (n = 3). *P < 0.05 and **P < 0.001 indicate the mean values of mea-6 plants that are significantly different from respective wild-type samples as determined by Student’s t test. Experiments were repeated at least two times with similar results.

Enhanced MEA Expression Supports Growth of Pathogens

To further investigate the consequence of enhanced MEA expression upon pathogenesis (Fig. 1), we generated multiple independent transgenic Arabidopsis lines (in the Col-0 background) constitutively expressing MEA (Supplemental Fig. S4) under the Cauliflower mosaic virus 35S promoter (35S:MEA). The 35S:MEA transgenic plants were morphologically normal like wild-type plants (Supplemental Fig. S4C). To examine whether constitutive MEA expression affects embryo development, we observed the developing 35S:MEA embryos. We did not find any defective embryos in 40 siliques randomly taken from three different 35S:MEA lines (Supplemental Fig. S5). We used MEA/mea (CS876294; ABRC) plants as a control for this experiment and as expected (Grossniklaus et al., 1998) found half of the embryos were aborted in these plants (Supplemental Fig. S5). Regarding defense against pathogens, we observed enhanced bacterial and fungal growth in MEA overexpression plants. The MEA overexpression plants supported higher Pst growth than wild-type plants (Fig. 3A), suggesting reduced immunity in these plants. This observation was further supported by the reduced pathogen-induced PR1 expression (Fig. 3B) and SA accumulation (Fig. 3, C and D) in the 35S:MEA plants compared to wild-type plants. To investigate the possible consequence of elevated MEA expression on ETI, we monitored the growth of Pst-AvrRpt2 in 35S:MEA and wild-type plants. As a control we used nonexpresser of PR genes1 (npr1-1), a susceptible mutant of Arabidopsis (Cao et al., 1994). We observed higher bacterial load in npr1-1 and 35S:MEA plants than wild-type plants (Fig. 3E; Supplemental Fig. S6). The reduced resistance in 35S:MEA plants was also associated with reduced PR1 transcript accumulation (Fig. 3F). Similarly, the HR as measured by ion leakage was significantly reduced in the 35S:MEA plants compared to wild-type plants upon Pst-AvrRpt2 inoculation (Fig. 3G). The reduced level of HR in the MEA overexpressing plants was in agreement with the reduced resistance against Pst-AvrRpt2. In addition, the MEA overexpressing plants showed a much higher level of disease symptoms than wild-type plants when inoculated with the necrotrophic pathogen Botrytis cinerea (Fig. 3H). Resistance to necrotrophic pathogens is associated with ET/JA signaling. In agreement with the loss-of-resistance phenotype, the MEA overexpression plants also showed reduced PLANT DEFENSIN1.2 (PDF1.2) expression compared to wild-type plants upon exogenous application of MeJA (Fig. 3I).

Figure 3.

Defense response in MEA overexpression and wild-type (Col-0) plants. A, Numbers of Pst in wild-type (WT) and 35S:MEA plants at 3 dpi (5 × 10⁵ CFU/mL). B, PR1 expression in wild-type and 35S:MEA plants after Pst inoculation. C, Total SA (free SA + SA-glucoside) content in leaves of wild-type and 35S:MEA plants after 36 h of Pst or mock inoculation. D, Free SA content in leaves of wild-type and 35S:MEA plants after 36 h of Pst inoculation. E, Numbers of Pst-AvrRpt2 in wild-type, 35S:MEA, and npr1-1 plants at 3 dpi (5 × 10⁵ CFU/mL). F, PR1 expression in wild-type and 35S:MEA plants after Pst-AvrRpt2 inoculation. G, HR-induced ion leakage in wild-type and 35S:MEA plants after Pst-AvrRpt2 inoculation (10⁷ CFU/mL). H, Disease symptoms in wild-type and 35S:MEA plants after 4 d of Botrytis cinerea inoculation (5 × 10⁵ spores/mL). I, Expression of PDF1.2 in wild-type and 35S:MEA plants after MeJA treatment (5 µm). Inset shows PDF1.2 expression after water treatment only. In A and E, error bars represent mean ± sd (n = 5). Different letters above the bars indicate mean values that are significantly different (P < 0.05) as analyzed by one-way ANOVA (post-hoc Holm-Sidak method). In B, F, and I, relative abundance of transcripts was determined by RT-qPCR. Error bars represent mean ± sd (n = 3). In C and D, error bars represent mean ± sd (n = 5). In G, each point represents mean ± sd (n = 3), and each sample contained eight leaf discs of 7 mm diameter. *P < 0.05 and **P < 0.001 indicate the mean values that are significantly different from mock-treated or respective wild-type samples as determined by Student’s t test. Experiments were repeated at least two times with similar results.

The results described above showed that activation of the defense response enhances MEA expression and enhanced MEA expression attenuates the defense response. Thus, MEA may function as a negative feedback regulator of defense in Arabidopsis. Since MEA expression was induced to a very high level upon Pst-AvrRpt2 inoculation (Fig. 1A), we investigated the mechanism and consequence of MEA expression using this pathogen.

Pathogenesis-Induced MEA Expression Is Associated with Altered Methylated Histone Occupancy at the MEA Locus

Silencing of MEA is mediated by methylation of DNA and di- and trimethylation of histone 3 at Lys-27 (H3K27me2/H3K27me3; Baubec and Mittelsten Scheid, 2006; Gehring et al., 2006; Jullien et al., 2006). Two distinct mechanisms are in place for activation of the maternal MEA allele in the female gametophyte and endosperm, and repression of the paternal allele in the sperm cell and endosperm (Baubec and Mittelsten Scheid, 2006; Gehring et al., 2006). The PcG repressor complex, involving MEA, is responsible for H3K27 methylation and suppression of MEA in vegetative tissue and suppression of the paternal allele in endosperm (Gehring et al., 2006). Activation of the maternal MEA allele in the endosperm is mediated by DEMETER that removes cytosine (C) methylation (Gehring et al., 2006). HpaII cleaves at unmethylated CCGG contexts in DNA but not when the central C is methylated (Yaish et al., 2014). We designed a pair of primers for amplifying 260 bp of the MEA promoter that covers the potential methylation sites (Fig. 4A). An HpaII-digested genomic DNA template would amplify that 260-bp region only when not cleaved (i.e. methylated). Contrary to our expectation, we did not observe any reduction in the level of methylation at the MEA locus upon pathogen inoculation (Fig. 4B). We used another restriction enzyme, McrBc, for which methylation at any two C residues preceded by a purine (A or G; Pu^mC) within 40 to 3,000 bp of the recognition site generates a restriction site (Gast et al., 1997; Stewart et al., 2000). Thus, the McrBc enzyme can detect methylation over a larger region of the DNA. We also did not observe any significant difference in the level of PCR amplicons between pathogen- and mock-inoculated samples (Fig. 4C), suggesting no change in DNA methylation at the MEA promoter after Pst-AvrRpt2 inoculation. The results ruled out the possibility of MEA activation through DNA demethylation.

Figure 4.

DNA and histone methylation at the MEA locus. A, Schematic diagram of the MEA locus. Black bars, coding sequences; gray bars, UTR; line, intron/promoter; thick lines above the structure, region used for histone modification study; blue line below promoter, region used for DNA methylation study. Numbers indicate nucleotide position with respect to the transcription start site (TSS). B Relative amount of HpaII-digested DNA in mock or Pst-AvrRpt2-treated leaves of wild-type (Col-0) plants. C, Relative amount of McrBc-digested DNA in mock- or Pst-AvrRpt2-treated leaves of wild-type (Col-0) plants. In B and C, samples were collected at 24 hpi with Pst-AvrRpt2 (1 × 10⁶ CFU/mL). Error bars represent mean ± sd (n = 3). D to F, Fold enrichment of H3K27me3 containing nucleosomes at the MEA locus. G to J, Fold enrichment of H3K4me3 containing nucleosomes at the MEA locus. In D to J, samples were collected at 24 hpi of Pst-AvrRpt2 (1 × 10⁶ CFU/mL). In D to J, gray and black bars indicate specific antibody and no antibody control, respectively. Error bars represent mean ± sd (n = 3). *P < 0.05 and **P < 0.001 indicate the mean values of pathogen-inoculated antibody precipitated sample that are significantly different from mock-inoculated antibody precipitated sample as determined by Student’s t test. Experiments were repeated at least two times with similar results.

To investigate the possible change in the occupancy of methylated H3K27 at the MEA locus, we performed a chromatin immunoprecipitation (ChIP) assay by using chromatins from mock, Pst-, and Pst-AvrRpt2-inoculated samples with an anti-H3K27me3 antibody. We found a significant reduction in H3K27me3 occupancy at three different regions of the MEA locus (Fig. 4A) upon Pst or Pst-AvrRpt2 inoculation (Fig. 4, D–F). In addition to removal of H3K27 methylation, enrichment of H3K4me3 also activates transcription. Thus, we also performed a ChIP assay by using the anti-H3K4me3 antibody. In agreement with the increased expression of MEA, we observed enhanced occupancy of H3K4me3 in regions of the MEA locus, especially after Pst-AvrRpt2 inoculation (Fig. 4, G–J). Compared to mock-treated samples, Pst-inoculated samples showed enhanced occupancy of H3K4me3 in area 3, which is about 1 kb downstream of the transcription start site (Fig. 4I). However, Pst-AvrRpt2 inoculation resulted in enhanced occupancy of H3K4me3 in all the regions tested. As positive controls of our ChIP experiment, we monitored the occupancy of H3K27me3 and H3K4me3 in FLOWERING LOCUS T (FT) and ACTIN2 (ACT2) loci respectively, which were known to accumulate these modified histones (Saleh et al., 2008). Both FT and ACT2 loci showed predicted enrichment of modified histones (Supplemental Fig. S7). The results suggested that pathogen-induced MEA expression is associated with the decrease of H3K27me3 and increase of H3K4me3 at the MEA locus, especially after inoculation with the avirulent pathogen Pst-AvrRpt2.

High-Level MEA Induction by Pst-AvrRpt2 Requires RPS2 Function

The results described above (Fig. 1A) showed that inoculation with Pst-AvrRpt2 results in high-level expression of MEA, compared to inoculation with the virulent pathogen. Pst-AvrRpt2 inoculation activates ETI in a RPS2-dependent manner (Kunkel et al., 1993; Bent et al., 1994; Guttman and Greenberg, 2001; Mackey et al., 2003; Belkhadir et al., 2004; Lim and Kunkel, 2004, 2005; Chen et al., 2007). To investigate whether RESISTANCE TO P. SYRINGAE2 (RPS2) function is required for MEA activation, we inoculated wild-type and rps2 mutants with Pst-AvrRpt2 and monitored MEA transcript accumulation. We observed Pst-AvrRpt2-induced high-level MEA expression in the wild-type background but not in rps2 mutant plants (Fig. 5). The level of MEA induction in the rps2 mutant was comparable to that of Pst-induced expression in wild-type plants. Thus, RPS2 function, which is required for AvrRpt2 effector-mediated ETI, is also required for Pst-AvrRpt2-mediated induction of MEA expression.

Figure 5.

Relative abundance of MEA mRNA in wild type (Col-0) and rps2 mutant after Pst-AvrRpt2 inoculation. Five-week-old plants were inoculated with Pst-AvrRpt2 (1 × 10⁶ CFU/mL), and leaf samples were harvested at the indicated time. Inset shows MEA expression in the rps2 mutant background. Relative abundance was measured by RT-qPCR. Error bars represent mean ± sd (n = 3). *P < 0.05 and **P < 0.001 indicate the mean values that are significantly different from respective 0-h samples as determined by Student’s t test. Experiments were repeated at least two times with similar results.

RPS2 Is a Target of MEA for Transcriptional Suppression

Our result showed that Pst-AvrRpt2 inoculation enhances MEA expression in a RPS2-dependenet manner (Fig. 5), and MEA expression suppressed RPS2-medited ETI (Fig. 3, E–G). Thus, there is a feedback inhibition of the defense response mediated by MEA. To investigate whether MEA expression modulates RPS2 expression, we monitored its mRNA accumulation in wild-type and 35S:MEA plants. Basal, as well as Pst-AvrRpt2-induced RPS2 expression was suppressed in 35S:MEA plants (Fig. 6, A and B), suggesting that MEA negatively regulates RPS2 expression. This was further validated in mea-6 mutants, in which both constitutive and pathogen-induced RPS2 expression was higher than in wild-type plants (Fig. 6C). We further demonstrated that coexpression of MEA suppressed RPS2:GUS, but not 35S:GUS expression (Fig. 6D), in Nicotiana benthamiana leaves in a transient expression system. Thus, RPS2 appeared to be a transcriptional target of MEA-mediated suppression.

Figure 6.

Influence of MEA on RPS2 expression. A, Relative abundance of RPS2 mRNA in wild-type (WT) and 35S:MEA plants without pathogen inoculation. B, Relative abundance of RPS2 mRNA in wild-type and 35S:MEA plants after 10 h of Pst-AvrRpt2 inoculation. C, Relative abundance of RPS2 mRNA in wild-type and mea-6 plants after 10 h of mock or Pst-AvrRpt2 inoculation. In A to C, error bars represent mean ± sd (n = 3). *P < 0.05 and **P < 0.001 indicate the mean values that are significantly different from respective wild-type samples as determined by Student’s t test. D, RPS2:GUS expression in N. benthamiana with or without coexpression of MEA. E, Schematic diagram of the RPS2 locus showing the transcription start site (TSS) and the areas used for the ChIP experiment. F, Fold enrichment of MEA-HA at the RPS2 locus. Error bars represent mean ± sd (n = 3). *P < 0.05 and **P < 0.001 indicate the mean values of anti-HA antibody precipitate that are significantly different from respective no antibody treated samples as determined by Student’s t test. G, Fold enrichment of H3K27me3 containing nucleosomes at the RPS2 locus in wild-type and 35S:MEA plants. H, Fold enrichment of H3K27me3 containing nucleosomes at the RPS2 locus in wild-type and mea-6 plants. Samples were harvested at 12 h post-inoculation with Pst-AvrRpt2 (1 × 10⁶ CFU/mL). In F through H, error bars represent mean ± sd (n = 3). *P < 0.05 and **P < 0.001 indicate the mean values of antibody-precipitated sample that are significantly different from wild-type antibody-precipitated samples as determined by Student’s t test. Experiments were repeated at least two times with similar results.

The PRC2 repressor complex physically associates with the target loci. The transgenic 35S:MEA plants contained a hemagglutinin (HA) tag as translational fusion with MEA. We examined the physical association of MEA-HA with the RPS2 locus by ChIP using an anti-HA tag antibody. As a negative control, we used ACT2, the expression of which was not affected by MEA overexpression. And as positive control, we used PHERES1 (PHE1), a known target of MEA. MEA is recruited at the PHE1 promoter, which results in enrichment of nucleosomes with H3K27me3 and thereby suppresses its expression (Makarevich et al., 2006). As expected, we found a MEA-HA association with the PHE1 promoter, but not with the ACT2 promoter (Supplemental Fig. S8). Three different regions of the RPS2 locus were used for the ChIP study, one at the promoter and two in the coding areas (Fig. 6E). We observed constitutive MEA-HA accumulation throughout the RPS2 locus in 35S:MEA plants (Fig. 6F). Upon pathogen inoculation, MEA-HA occupancy further increases at areas 1 and 3 of the RPS2 locus (Fig. 6F). Interestingly, MEA accumulation at area 2 of RPS2 reduces after Pst-AvrRpt2 inoculation (Fig. 6F). It is possible that the other transcription factors that are involved for RPS2 expression partly displaced MEA-HA during pathogenesis. Being a part of a polycomb-group repressor, MEA contributes to transcriptional suppression of target loci by histone methylation (Makarevich et al., 2006). Since MEA physically associates with the RPS2 locus, we speculated a similar mechanism for AvrRpt2-mediated suppression of RPS2. To test this hypothesis, we monitored H3K27me3 occupancy at the RPS2 locus by ChIP. Wild-type and 35S:MEA plants were inoculated with Pst-AvrRpt2 and chromatin fragments were precipitated by anti-H3K27me3 antibody. Relative abundance of histone methylation at the RPS2 locus was determined by qPCR. We observed significantly high levels of H3K27me3 occupancy in 35S:MEA plants compared to the wild-type at the RPS2 locus (Fig. 6G). In agreement with this observation, we found reduced enrichment of H3K27me3 in mea-6 mutants than in wild-type plants at all tested regions (Fig. 6H). The FT that accumulates high levels of H3K27me3 served as the required control (Saleh et al., 2008; Supplemental Figs. S9 and S10). The results demonstrated that RPS2 is a direct target for MEA-mediated transcriptional repression.

Di19 Interacts with and Recruits MEA at the RPS2 Promoter

The interaction of SET-domain-containing proteins of the PRC-2 complex with target DNA is indirect, mediated by other DNA-binding proteins (Margueron and Reinberg, 2011). We could not detect any direct association of MEA with the RPS2 promoter in a gel-shift assay. By screening an Arabidopsis cDNA library, using MEA as bait, we identified DROUGHT-INDUCED19 (Di19/AtDi19-1) as an interacting factor involved in RPS2 suppression and promotion of bacterial growth. Di19 is a member of the AtDi19 gene family, which has seven members (AtDi19-1 to AtDi19-7), each containing two hydrophilic Cys-2/His-2 zinc-finger-like domains (Milla et al., 2006). These Cys-2/His-2 zinc-finger motifs are evolutionarily conserved among monocots and dicots, suggesting a conserved biological function. Most of the family members express ubiquitously in all organs and have similar subcellular localization, i.e. the nucleus (Milla et al., 2006), indicating possible functional redundancy among family members. To identify the essential domains of Di19 and MEA for their physical interaction, we performed a yeast two-hybrid assay with full-length and deletion proteins. MEA contains an acidic region and a Cys-rich domain in the N-terminal half and a nuclear localization signal (NLS), a CXC domain, and the SET domain in the C-terminal half (Yadegari et al., 2000; Fig. 7A). Di19 contains two zinc-finger domains and one NLS (Milla et al., 2006; Liu et al., 2013; Fig. 7A). Interaction studies in yeast with deleted domains of MEA and Di19 suggested that the N-terminal part containing the acidic region of MEA was sufficient, whereas the zinc-finger domains and the C-terminal region, including the NLS of Di19, were required for the interaction (Fig. 7B). An in planta interaction of MEA with Di19 was confirmed in onion (Allium cepa) epidermal cells by a bimolecular fluorescence complementation assay (Fig. 7C). Di19 codes for a DNA-binding transcription factor-like protein. The RPS2 promoter contains one predicted Di19-binding sequence [DiBS; TACA(A/G)T; Liu et al., 2013] at −422 bp from the transcription start site. A gel-electrophoresis mobility shift assay (EMSA) revealed that Di19 binds with the RPS2 promoter (Fig. 7D). To further establish the possible role of Di19 in defense, we used the di19 mutant (Salk_088814; Supplemental Fig. S11, A and B) and the Di19 overexpressing line (Supplemental Fig. S11B; Liu et al., 2013). In agreement with the predicted function of the repressor complex, the di19 mutant showed enhanced RPS2 expression (Fig. 7E). Moreover, similar to MEA overexpression, Di19 overexpression supported higher Pst-AvrRpt2 growth (Fig. 7F), whereas its mutant showed resistance (Fig. 7G). A modest difference in bacterial growth between wild-type and Di19 overexpression or mutant plants indicates that MEA-mediated susceptibility may involve other proteins in addition to Di19. Functional redundancy among Di19 family members may also be the possible reason for the difference. Nevertheless, our results suggest that MEA and Di19 forms a functional PRC2-like complex, which associates at the RPS2 promoter and suppresses it expression.

Figure 7.

Di19 interacts with MEA and influences disease resistance. A Schematic diagram of MEA and Di19, full and truncated proteins used in the interaction study. B, Yeast two-hybrid interaction. Transformed yeast cells were grown on Leu⁻, Trp⁻, His⁻, and Ade⁻ (−LTHA) medium, which allows only interacting clones to grow. p53 and T-antigen were used as positive controls, and empty vectors were used as negative controls. C, Bimolecular fluorescence complementation (BiFC) in transiently expressed onion epidermal cells. D, EMSA for confirmation of Di19 with theRPS2 promoter. Each reaction contained 50 ng of radiolabeled oligonucleotides containing DiBS from the RPS2 promoter and either MBP or MBP-Di19 (0.4–2 µg). For the competitive binding assay, 50× and 100× of nonradiolabeled oligonucleotide was used with 1 µg of protein. *nonspecific bindings. E, Relative abundance of RPS2 mRNA in wild-type and di19 mutant plants after Pst-AvrRpt2 inoculation. Error bars represent mean ± sd (n = 3). *P < 0.05 and **P < 0.001 indicate the mean values that are significantly different from respective time point wild-type samples as determined by Student’s t test. F, Numbers of Pst-AvrRpt2 in wild-type and Di19 overexpression plants at 3 d post-Pst-AvrRpt2 inoculation. G, Numbers of Pst-AvrRpt2 in wild-type and di19 mutant plants at 3 d post-Pst-AvrRpt2 inoculation. In F and G, error bars represent mean ± sd (n = 5). Experiments were repeated at least two times with similar results. *P < 0.05 indicates the mean values that are significantly different from wild-type samples as determined by Student’s t test.

DISCUSSION

MEA Functions as a Feedback Inhibitor of Defense

Activation of the immune response takes place at the cost of metabolic energy. Thus, plants possess factors that do not allow spontaneous activation of the immune response and checks that limit the defense responses once activated to an optimal level. Genetic screens identified many negative regulators, mutations in which activate spontaneous defense. For example, mutants of CONSTITUTIVE FOR PR1 (CPR1), CPR5, SUPPRESSOR OF SA INSENSITIVE1, SA INSENSITIVE2, CONSTITUTIVE EXPRESSION OF VSP1, and SUPPRESSOR OF NPR1-1 CONSTITUTIVE1 activate constitutive defense, suggesting a negative regulatory role of these genes in defense (Bowling et al., 1994, 1997; Shah et al., 1999, 2001; Ellis and Turner, 2001; Jirage et al., 2001; Li et al., 2001). The mutants of these genes spontaneously activate SA or ET/JA signaling and thereby activate cell death and other defense responses. In addition, plants also recruit factors such as LESION SIMULATING DISEASE1 (lsd1) that regulate excessive cell death upon pathogen inoculation (Dietrich et al., 1994). lsd1 negatively regulates basal defense independent of SA but regulates cell death downstream of SA accumulation in a NPR1-dependent manner (Aviv et al., 2002). Our results identified a very different mechanism of defense response regulation mediated by MEA, a known epigenetic modulator and transcriptional repressor. MEA expression, which remains suppressed in the vegetative tissue, is induced upon pathogenesis (Fig. 1), and artificial expression of MEA negatively regulates defense (Fig. 3). The results prompted us to hypothesize that MEA functions as a feedback inhibitor of defense (Fig. 8A). Since MEA is induced by both SA- and JA-pathway activation, and MEA expression negatively regulates both biotrophic and necrotrophic pathogens, MEA is likely to control multiple aspects of plant immune response. Via the Arabidopsis and Pst-AvrRpt2 interaction, we showed that MEA suppresses RPS2 expression (Fig. 6). RPS2 is the R gene that functions in combination with the AvrRpt2 effector for activating ETI (Kunkel et al., 1993; Bent et al., 1994; Guttman and Greenberg, 2001; Mackey et al., 2003; Belkhadir et al., 2004; Lim and Kunkel, 2004, 2005; Chen et al., 2007). Interestingly, only partial suppression of RPS2 expression in MEA-overexpressing plants was sufficient to significantly suppress Pst-AvrRpt2-mediated ETI. However, the results are in agreement with earlier observations in CPR1 overexpressing plants, which showed partial accumulation of RPS2 with a dramatic reduction in the resistance against Pst-AvrRpt2 (Cheng et al., 2011). Interestingly, RPS2 function is also required for Pst-AvrRpt2-induced high-level expression of MEA (Fig. 5). This result further supports the feedback inhibitory role of MEA in defense (Fig. 8A). However, target genes of the MEA-PRC2 complex involved in basal defense remain unidentified.

Figure 8.

Models depicting the involvement of MEA in defense. A, General role of MEA in defense attenuation. Pathogen infection leads to induction of defense responses and also activates MEA expression. MEA negatively regulates the defense output. B, Involvement of MEA in attenuating AvrRpt2-induced ETI. Under normal conditions, MEA expression is suppressed by histone and DNA methylation. AvrRpt2 activates RPS2-mediated ETI and MEA expression. Pst-AvrRpt2-induced high-level accumulation of H3K4me3 overrides MEA suppression. MEA along with Di19 binds to the promoter of RPS2 to suppress its expression and thereby to attenuate RPS2-mediated ETI.

Mechanism of MEA Activation and RPS2 Suppression

Chromatin modification by the PRC is a common strategy of gene silencing in higher eukaryotes (Simon and Kingston, 2013). Two groups of PRCs exist. PRC2 contributes to methylation at H3K27, whereas PRC1 recognizes such modification and brings structural changes in chromatin (Köhler and Hennig, 2010; Kalb et al., 2014). MEA, which belongs to the PRC2 group, contributes to its own suppression in vegetative tissues (Baubec and Mittelsten Scheid, 2006; Jullien et al., 2006). Suppression of MEA in vegetative tissues and the paternal allele in the embryo is associated with H3K27 methylation (Jullien et al., 2006). In the female gametophyte, embryonic tissues, and central cells, demethylation of H3K27me3 takes place through a yet-unidentified histone demethylase that allows its expression. In addition to H3K27 methylation, MEA is also suppressed by DNA methylation in vegetative tissues and in the paternal allele in endosperm (Gehring et al., 2006). Our results demonstrated that, contrary to the pathogen-induced expression, the epigenetic repressor marks (DNA methylation) are not removed from the MEA locus (Fig. 4, B and C). However, pathogenesis overrides the suppression through a different epigenetic mechanism, which is indicated by the enhancement of H3K4me3 with reduction of H3K27me3 marks at the MEA locus. Though the role of histone methylations in gene expression is not fully established (Henikoff and Shilatifard, 2011), H3K4me3 is often associated with actively transcribing genes, whereas H3K27me3 is associated with the transcriptionally silenced genes (Schones and Zhao, 2008). Chromatin modification at the MEA locus is in agreement with the pathogenesis-induced transcript accumulation of MEA. Our results clearly showed that MEA and its interactor Di19 negatively regulate RPS2 expression (Figs. 6, A–C, and 7E). PcG repressors associate physically with the target loci and induce chromatin modification (Simon and Kingston, 2013; Entrevan et al., 2016). Through ChIP and EMSA, we showed that MEA and Di19 associate with the RPS2 locus (Figs. 6F and 7D). The most important function of PRC2 is to methylate H3 at K27. Similar to the reported MEA target PHE1, the RPS2 locus also accumulates H3K27me3 in MEA-overexpressing plants (Fig. 6G), whereas the mea-6 mutant has lower occupancy of H3K27me3 (Fig. 6H). All together, our results demonstrated that RPS2 is a direct target of MEA for transcriptional suppression, and Di19 takes part in recruiting MEA at the RSP2 promoter (Fig. 8B).

In summary, the imprinted PcG suppressor MEA remains transcriptionally silent in vegetative tissues. Pathogen inoculation activates MEA transcription, which in turn limits the induction of excessive immune response. Thus, MEA functions as a feedback inhibitor of defense and plays roles in the growth-defense tradeoff.

MATERIALS AND METHODS

Plant Growth Conditions and Pathogen Inoculation

Arabidopsis (Arabidopsis thaliana) plants were grown as described previously (Swain et al., 2011; Singh et al., 2013), in a growth room at 21°C and 65% relative humidity with an alternate light (80 µE m⁻¹s⁻¹)/dark period of 12 h each. Bacterial cultures were grown overnight and harvested and resuspended in 10 mm MgCl₂ and diluted as required before infiltrating abaxial sides of leaves with a needleless syringe. Bacterial loads in the leaves were determined as described previously (Singh et al., 2013). In brief, we inoculated about 12 to 16 plants per line and 1 to 2 leaves per plant with the bacteria suspended in 10 mm MgCl₂. While collecting samples, we randomly selected 20 leaves. Leaves were pooled into groups of five each having four leaves. Using a cork-borer, a disc of 5 mm diameter was taken from each leaf, homogenized and serially diluted in 10 mm MgCl₂ before plating for counting colony-forming units (CFU). For Botrytis cinerea inoculation, spores were suspended in potato dextrose broth (5 × 10⁵ spores/mL) and sprayed on plants. Inoculated plants were covered with a transparent plastic dome and kept in low light for 4 d. Symptoms were observed after 4 d of inoculation.

Generation of mea-6 Homozygous Lines

Heterozygous MEA/mea-6 (CS6996) plants were obtained from the Arabidopsis Stock Center. To generate mea-6 homozygous mutants, MEA/mea-6 plants were crossed as the male parent to haploid inducer GFP-tailswap plants as the female parent (Ravi et al., 2014). The resultant F1 seeds were germinated on Murashige and Skoog agar plates and later transferred to soil for further growth. Haploids were identified both phenotypically and cytologically as described earlier (Ravi and Bondada, 2016). All the haploids were PCR genotyped using a derived cleaved amplified polymorphic sequence analysis. Both MEA and mea-6 alleles generate a 152-bp amplicon with MR260 and MR261 primers (Supplemental Table S1). The PCR product from the mea-6 allele cleaved into two fragments of 128 bp and 24 bp upon digestion with XbaI enzyme, whereas the wild-type MEA allele remains uncut (Supplemental Fig. S12). The mea-6 haploid plants were grown to full maturity, and spontaneous seeds arising either due to mitotic and/or meiotic chromosome doubling were collected, and viable seeds were sown further to generate doubled haploid (mea-6/mea-6) plants, which were again confirmed by derived cleaved amplified polymorphic sequence genotyping for homozygosity (Supplemental Fig. S12). The seeds obtained from the doubled haploid mea-6/mea-6 plants were used for the experiments described here.

Chemical Treatment

For expression analysis, SA (500 µm in water) was thoroughly sprayed on plants, whereas flg22 (1 µm in water) was pressure infiltrated through the abaxial leaf surface (Swain et al., 2015). After the treatment, plants were transferred to a growth chamber and covered with a plastic dome overnight for maintaining humidity. Detached leaves of 5-week-old plants were floated in 5 µm MeJA dissolved in 0.1% ethanol. Control samples were floated in 0.1% ethanol. Samples were collected at the indicated time intervals. Expression was determined by RT-qPCR.

Salicylic Acid Estimation

Estimation of SA was done by HPLC (Agilent 1220 LC) exactly as described previously (Singh et al., 2013).

RNA Isolation, cDNA Synthesis, and Expression Analysis

RNA was isolated from leaf samples. Total RNA was extracted by the guanidiniumthiocyanate-phenol-chloroform method (Chomczynski and Sacchi, 1987). For RT-PCR, 1.0 µg of RNA was treated with DNase I (Thermo Scientific) for 30 min at 37°C and was used for first-strand cDNA synthesis using a kit (iscriptcDNA synthesis kit, Bio-Rad; catalog no. 170-8891). Semiquantitative RT-PCR and RT-qPCR were carried out by gene-specific primers listed in Supplemental Table S1. Quantitative PCR was carried out by a 7500 Fast Real-Time PCR machine (Applied Biosystems) using 2× Power SYBR Green master mix (Applied Biosystems; catalog no. 4367659). Typically, each experiment contained three biological samples with two technical replicates. The average of the two technical replicates was taken as the reading for that biological sample ACT2 (At3g18780) and TUBULIN2 (At5g62690) were used for normalization. The data represented here are normalized with ACT2 only. Nontemplate controls were included in each RT-qPCR reaction. The melting curve generated by the software was used to ensure the presence of a single PCR product in each lane, which was verified by agarose gel electrophoresis. Further, we sequenced the PCR products from each set of primers to confirm the specific product.

Ion Leakage Experiment

Leaves of 5-week-old plants were infiltrated through the abaxial surface with a suspension of Pst-AvrRpt2 at 1 × 10⁷ CFU/mL in 10 mm MgCl₂. Only 10 mm MgCl₂ was used as the mock treatment. Infiltrated plants were covered with a plastic dome and dark incubated for 7 h in a growth room. After that leaf discs (0.7 cm diameter) were punched out with a cork borer and washed for 45 min in distilled water with gentle shaking. Then the leaf discs were floated in distilled water in a 6-well plate. Usually, every sample contained eight leaf discs in 8 mL of water and each experimental set contained three biological replicates. The conductivity of water in terms of µS/cm²/s using a conductivity meter (HI2300; Hanna) was measured from 8 to 22 h after inoculation. At the end, the leaf discs along with water were autoclaved to achieve a maximum release of ionic content. Ion leakage was plotted as a percentage of maximum conductivity.

Generation of MEA Overexpression and GFP-Tagged Lines

The full-length MEA coding sequence (CDS) was amplified from an Arabidopsis cDNA pool prepared from a pathogen-inoculated leaf sample, using a proofreading capable DNA polymerase Pfu (New England Biolabs). For the 35S:MEA construct, the MEA CDS was amplified using Pfu DNA polymerase with end primers, and an A-overhang was generated by Taq DNA polymerase. The pCXSN vector (Chen et al., 2009) was digested with XcmI to generate a T-overhang and ligated with the PCR amplified MEA CDS. For expression as a GFP-tagged protein (MEA-GFP), the MEA CDS was cloned into the pCXDG vector (Chen et al., 2009) as described for 35S:MEA. For generation of transgenic plants, wild-type Arabidopsis (Col-0) was transformed by the Agrobacterium tumefaciens-mediated floral-dip transformation method (Zhang et al., 2006b). Transformed seeds were screened on MS media supplemented with hygromycin (25 mg/L). Antibiotic resistant plants were later confirmed for the presence of the antibiotic resistance gene by PCR and expression analysis.

MEA:GUS Vector Construction and Transient Assay

The DNA fragment spanning the 1,085-bp upstream region from the transcription start site of MEA was amplified using specific primers (Supplemental Table S1). The 35S CaMV promoter of the pBI 121 vector was excised by HindIII and BamHI, and the MEA promoter was ligated to generate the MEA:GUS construct. A. tumefaciens strain C58 was transformed with either 35S:GUS (pBI121) or MEA:GUS. Transient GUS activity in Arabidopsis leaves was observed as described in (Lee and Yang, 2006). In brief, 4-week-old Arabidopsis leaves were infiltrated with A. tumefaciens containing either 35S:GUS or MEA:GUS (0.4 OD). After 48 h, leaves were infiltrated with Pst-AvrRpt2 (10⁶ CFU/mL suspended in 10 mm MgCl₂) or only 10 mm MgCl₂ as the mock. At 12 h after pathogen inoculation, leaves were stained overnight at 37°C in GUS staining solution (1 mm EDTA [pH 8], 5 mm potassium ferricyanide, 5 mm potassium ferrocyanide, 100 mm sodium phosphate [pH 7.0], 1% Triton X-100, and 1 mg/mL X-Gluc). Stained leaves were kept in ethanol for removal of chlorophyll.

RPS2:GUS Vector Construction and Transient Expression in Nicotiana benthamiana

The DNA fragment of 995 bp upstream of RPS2 was amplified and cloned in the binary vector pBI121 between PstI and XbaI after replacing the 35S CaMV promoter. N. benthamiana leaves were coinfiltrated with A. tumefaciens carrying binary vectors expressing RPS2:GUS with either MEA under the CaMV 35S promoter in the pCXDG vector or the empty vector (0.4:0.4 OD). After 2 d, the leaves were either infiltrated with Pst-AvrRpt2 (10⁶ CFU/mL suspended in 10 mm MgCl₂) or only 10 mm MgCl₂ as the mock. At 12 h after pathogen inoculation, leaves were stained for GUS expression as described above.

ChIP

ChIP was performed as described previously (Saleh et al., 2008; Singh et al., 2014). In brief, Arabidopsis leaves were inoculated with the bacterial pathogen (1 × 10⁶ CFU/mL) suspended in 10 mm MgCl₂ or only 10 mm MgCl₂ as the mock treatment. Each sample consisted of 4.0 g of freshly harvested leaves. Immunoprecipitation was done with either anti-H3K4me3, anti-HA (Abcam), or anti-H3K27me3 (Millipore) antibody. Fold enrichment of immunoprecipitated chromatin for each target gene was plotted according to the ΔΔCT method by RT-qPCR. The CT value for the antibody sample (+AB) and for no antibody control (−AB) was independently subtracted from the CT value of the corresponding input to find ΔCT. Then ΔCT^−AB was subtracted from ΔCT^+AB to get the ΔΔCT for each sample and 2^−ΔΔCT was plotted (Mukhopadhyay et al., 2008; Han et al., 2016). Primers used in this study are mentioned in Supplemental Table S1.

Embryo Microscopy

Siliques of different maturity levels were taken for the study. Developing embryos were taken out and cleared in Hoyer’s solution overnight to remove chlorophyll. The morphology of developing embryos was observed under microscope.

DNA Methylation Analysis

Five-week old wild-type Arabidopsis leaves were inoculated with either P. syringae pv. tomato carrying AvrRpt2 (10⁷ CFU/mL) suspended in 10 mm MgCl₂ or only 10 mm MgCl₂ as mock treatment. Leaf samples were harvested at 24 hpi and genomic DNA was extracted using plant DNA extraction kit (Thermo Scientific). To 1 µg of gDNA, methylation sensitive enzymes HpaII 2.0 U (Thermo Scientific) or McrBc 2.0U (New England Biolabs) was added in a 50 µL reaction and incubated for 8 h. Relative content of digested DNA was determined by qPCR.

Bimolecular Fluorescence Complementation

Bimolecular fluorescence complementation constructs were generated by cloning the desired gene CDS in either pSPYNE(R)173 or pSPYCE(M) as described previously (Waadt et al., 2008). Clones were transformed in A. tumefaciens C58 strain. Fleshy onion scales were fully immersed in transformed C58 strain of A. tumefaciens suspension (0.8 OD) and kept at 28°C for 12 to 24 h. Scales were then transferred on half-strength Murashige and Skoog media and incubated for 2 to 3 d. Cocultivated scales were thoroughly washed with sterile water and the epidermal layer was peeled off and mounted on a slide for observation. Samples were observed under a confocal microscope and analyzed with the Olympus FV1000 viewer software.

EMSA

For construction of the MBP-Di19 recombinant fusion protein, Di19 sequences were cloned in the pMAL-p2X vector (New England Biolabs) at the C-terminal end of MBP between EcoRI and BamHI restriction sites. MBP and MBP-Di19 were expressed in the Escherichia coli BL21 (DE3) strain and purified using amylose resin (New England Biolabs). For the binding assay, 400 ng, 1 µg, 1.6 µg and 2 µg of MBP or MBP Di19 was used. Oligonucleotides (40 bp) surrounding DiBs were used and radiolabeled with γP32 ATP at the 5′ end by T4 polynucleotide kinase (New England Biolabs). For competitive binding, 50× and 100× nonradiolabeled RPS2 DNA was used along with 1 µg of MBP or MBP-Di19. EMSA was performed as described previously (Hellman and Fried, 2007) using a 10% polyacrylamide gel.

Yeast Two-Hybrid Library Screening

The MEDEA CDS was cloned into the pBGKT7 vector between EcoRI and BamHI restriction sites to fuse with the GAL4 transcription factor DNA-binding domain. The Di19 CDS was cloned into the pGADT7 vector between NdeI and EcoRI, to fuse with the activation domain. Confirmation of interactors was done by activation of reporter genes and survival on quadruple dropout (-leu, -trp, -his, -ade) media. Yeast growth, transformation, and depleted synthetic media preparation was done according to the manufacturer’s protocol (Clontech).

Accession Numbers

The gene accession numbers that were used in this study are as follows: At1g02580 (MEDEA), At1g56280 (AtDi19), At3g18780 (ACTIN2), At4g26090 (RPS2), At5g62690 (TUBULIN2), At1g65480 (FT), At5g44420 (PDF1.2), At2g14610 (PR1), At1g65330 (PHE1), At5g58230 (MSI1), and At3g20740 (FIE).

Supplemental Data

The following supplemental materials are available.

• Supplemental Figure S1. MEA transcript accumulation in wild-type plants after flg22 treatment.

• Supplemental Figure S2. FIE and MSI1 expression in wild-type plants after pathogen inoculation.

• Supplemental Figure S3. Morphological phenotype of wild-type (C24) and mea-6 plants.

• Supplemental Figure S4. 35S:MEA transgenic plants.

• Supplemental Figure S5. Morphology of developing embryo of wild-type, 35S:MEA, and mea/+ plants.

• Supplemental Figure S6. Defense against the bacterial pathogen in 35S:MEA plants.

• Supplemental Figure S7. H3K27me3 and H3K4me3 occupancy on FT and ACT2.

• Supplemental Figure S8. Fold enrichment of MEA-HA at PHE1 and ACT2 loci.

• Supplemental Figure S9. H3K27me3 occupancy on the FT locus in Col-0 plants.

• Supplemental Figure S10. H3H27me3 occupancy on the FT locus in C24 wild-type plants.

• Supplemental Figure S11. Confirmation of the di19 mutant and overexpression lines.

• Supplemental Figure S12. Genotyping of plants for mea-6 and MEA alleles.

• Supplemental Table S1. Primer sequences.

Dive Curated Terms

The following phenotypic, genotypic, and functional terms are of significance to the work described in this paper:

• ATP CHEBI: CHEBI:15422

• AtDi19 Gramene: AT1G56280

• AtDi19 Araport: AT1G56280

• CPR5 Gramene: AT5G64930

• CPR5 Araport: AT5G64930

• ENDOSPERM AmiGo: PO:0009089

• FIE Gramene: AT3G20740

• FIE Araport: AT3G20740

• prc1 Gramene: AT5G64740

• prc1 Araport: AT5G64740

• VSP1 Gramene: AT5G24780

• VSP1 Araport: AT5G24780

• chemical treatment Gramene: EO:0007189

• chemical treatment Araport: EO:0007189

• dna Gramene: EO:0007532

• dna Araport: EO:0007532

• h3k27me3 CHEBI: CHEBI:85045

• ion CHEBI: CHEBI:24870

• leaf AmiGo: PO:0025034

• MEA Gramene: AT1G02580

• MEA Araport: AT1G02580

• plant hormone CHEBI: CHEBI:37848

• salicylic acid CHEBI: CHEBI:16914