Hypomethylated pollen bypasses the interploidy hybridization barrier in Arabidopsis

Abstract

Plants of different ploidy levels are separated by a strong postzygotic hybridization barrier that is established in the endosperm. Deregulated parent-of-origin specific genes cause the response to interploidy hybridizations, revealing an epigenetic basis of this phenomenon. In this study, we present evidence that paternal hypomethylation can bypass the interploidy hybridization barrier by alleviating the requirement for the Polycomb Repressive Complex 2 (PRC2) in the endosperm. PRC2 epigenetically regulates gene expression by applying methylation marks on histone H3. Bypass of the barrier is mediated by suppressed expression of imprinted genes. We show that the hypomethylated pollen genome causes de novo CHG methylation directed to FIS-PRC2 target genes, suggesting that different epigenetic modifications can functionally substitute for each other. Our work presents a method for the generation of viable triploids, providing an impressive example of the potential of epigenome manipulations for plant breeding.

Figure 1.

Bypass of Seed Abortion by Hypomethylation of the Paternal Genome. (A) Percentage of collapsed and partially collapsed seeds in interploidy hybridizations. n, total number of seeds tested. (B) Phenotype of mature seeds. Genotypes are indicated above. (C) Percentage of germinating seeds in hybridizations with different paternal genotypes. Pictures correspond to columns in the graph below; n, total number of seeds tested. (D) Expression of ADM in seeds 6 DAP. MNE, mean normalized expression. ACTIN11 was used as reference. Error bars indicate se. [See online article for color version of this figure.]

Figure 2.

Deregulated Genes in Interploidy Hybridizations Are Normally Expressed in Hypomethylated Paternal Excess Hybridizations. (A) Left panel: Venn diagram showing overlap of genes being upregulated in seeds derived from wild type × osd1 crosses (signal log ratio [SLR] > 1, P < 0.05) with genes that are downregulated in wild type × met1 (SLR < −1, P < 0.05) and genes that are downregulated in wild type × osd1 met1 compared with wild type × osd1 (SLR < −1, P < 0.05). Right panel: Venn diagram showing overlap of genes being downregulated in seeds derived from wild type × osd1 crosses with genes that are upregulated in wild type × met1 and genes that are upregulated in wild type × osd1 met1 compared with wild type × osd1 (SLR > 1, P < 0.05). (B) Expression fold change of maternal (M; red bars) and paternal (P; blue bars) alleles of genes being upregulated in crosses wild type × osd1 versus wild type × wild type; genes being downregulated in crosses wild type × osd1 met1 versus wild type × osd1; genes being downregulated in crosses wild type × osd1 met1 versus wild type × wild type; and genes being downregulated in crosses wild type × met1 versus wild type × wild type. SLRs and P values of crosses are as indicated in (A). (C) Expression of MEGs (red bars) and PEGs (blue bars) in crosses of wild type × osd1 versus wild type × wild type; wild type × osd1 met1 versus wild type × wild type; and wild type × met1 versus wild type × wild type. [See online article for color version of this figure.]

Figure 3.

DNA Methylation Profile of 1n and 2n Sperm Nuclei. Genes (left panels) and TEs (right panels) were aligned at the 5′ and 3′ ends (dashed lines), and average methylation levels in CG (top panels), CHG (middle panels), and CHH context (bottom panels) for each 100-bp interval were plotted.

Figure 4.

DNA Methylation Changes in Embryo and Endosperm Derived from Interploidy Hybridizations and Effects on Transcript Levels. (A) Genes (top panels) and TEs (lower panels) were aligned at the 5′ and 3′ ends (dashed lines), and average methylation levels in CG (left panels), CHG (middle panels), and CHH context (right panels) for each 100-bp interval were plotted. (B) Box plots of expression changes in triploid seeds. The orange box shows expression changes of all genes with increased expression in triploid seeds, while the yellow box shows expression changes of those genes that have increased expression in triploid seeds but lose CHH methylation.

Figure 5.

Allele-Specific DNA Methylation Profiles of Endosperm Derived from Interploidy Hybridizations. Genes (top panel) and TEs (lower panel) were aligned at the 5′ and 3′ ends (dashed lines), and average methylation levels in CG (left panels), CHG (middle panels), and CHH context (right panels) for maternal and paternal alleles for each 100-bp interval were plotted.

Figure 6.

DNA Methylation Differences Are Prominent on Long Heterochromatic TEs. (A) Box plots showing absolute fractional CHH demethylation of 50-bp windows within transposons; numbers on x axis represent length of TEs in kilobases. (B) Box plots showing absolute fractional CHG (left panel) and CHH (right panel) demethylation of 50-bp windows within euchromatic and heterochromatic TEs. TEs with H3K9m2 log2 scores lower than −1 and higher than 1 are considered euchromatic and heterochromatic, respectively. (C) CHH methylation levels are plotted as in Figure 4 for TEs belonging to four distinct families.

Figure 7.

Rescue of 3n Seeds by Hypomethylated Pollen Does Not Require a Functional FIS-PRC2. (A) H3K27m3 levels in the endosperm (data from Weinhofer et al., 2010) of genes upregulated (green line) and downregulated (blue line) in wild type × osd1 crosses compared with all genes (red line). Average methylation level is shown for 100-bp windows; numbers on x axis represent the length of genes in kilobases. (B) Percentage of collapsed and partially collapsed seeds in interploidy hybridizations with lack of FIS-PRC2; numbers on top of bars represent total number of seeds per genotype. (C) Expression analyses of PEGs in whole siliques 6 DAP derived from indicated crosses. MNE, mean normalized expression; ACTIN11 was used as reference. Error bars indicate se.

Figure 8.

Loss of CG Methylation Causes Redistribution of CHG and CHH Methylation. (A) DNA methylation profiles of genes in the endosperm (top panels) and embryo (lower panels) derived from seeds of the following crosses: wild type × wild type, wild type × osd1, wild type × osd1 met1, and wild type × met1. Genes were aligned at the 5′ and 3′ends (dashed lines), and average methylation levels in CG (left panels), CHG (middle panels), and CHH context (right panels) for maternal and paternal alleles for each 100-bp interval were plotted. (B) Venn diagram showing overlap of genes with decreased CHH methylation in wild type × osd1 crosses compared with the wild type with genes that have increased CHG methylation levels in wild type × met1 and wild type × osd1 met1 compared with wild type and wild type × osd1, respectively. (C) Box plots of expression changes in diploid seeds upon pollination with met1 pollen compared with wild-type pollinations. The yellow box shows expression changes of all genes with increased expression in triploid seeds, while the blue box shows expression changes of genes that gain CHG methylation upon met1 pollination. (D) Box plots of expression changes in triploid seeds upon pollination with osd1 met1 pollen compared with osd1 pollinations. The yellow box shows expression changes of all genes with increased expression in triploid seeds, while the green box shows expression changes of genes that gain CHG methylation upon osd1 met1 pollination.