A role for CHROMOMETHYLASE3 in mediating transposon and euchromatin silencing during egg cell reprogramming in Arabidopsis

Abstract

During embryogenesis there is a major switch from dependence upon maternally-deposited products to reliance on products of the zygotic genome. In animals, this so-called maternal-to-zygotic transition occurs following a period of transcriptional quiescence. Recently, we have shown that the early embryo in Arabidopsis is also quiescent, a state inherited from the female gamete and linked to specific patterns of H3K9 dimethylation and TERMINAL FLOWER2 (TFL2) localization. We also demonstrated that CHROMOMETHYLASE 3 (CMT3) is required for H3K9 dimethylation in the egg cell and for normal embryogenesis during the first few divisions of the zygote. Subsequent analysis of CMT3 mutants points to a key role in egg cell reprogramming by controlling silencing in both transposon and euchromatic regions. A speculative model of the CMT3-induced egg cell silencing is presented here, based on these results and current data from the literature suggesting the potential involvement of small RNAs targeted to the egg cell, a process conceptually similar to the division of labor described in the male gametophyte for which we show that H3K9 modifications and TFL2 localization are reminiscent of the female gametophyte.

In sexually reproducing organisms, the fusion of two highly specialized cells, the gametes, gives rise to a totipotent embryo. In animals, both gametes are transcriptionally silent and the early embryo remains in a relative quiescent status. This period of transcriptional quiescence is absolutely crucial for reprogramming the genome and conferring totipotency to the embryo.^1,2 Unlike animals, plant germ cells arise from somatic cells in the adult and fertilization involves two pairs of gametes: two sperm cells fuse with the egg and central cells to produce the embryo and the endosperm nurturing tissue, respectively. We recently showed³ that in Arabidopsis, as in animals, the early embryo was transcriptionally quiescent and that this transcriptional quiescence was first established in the egg cell. We further demonstrated that this transcriptional pattern correlated with both H3K9 dimethylation (H3K9me2) and TERMINAL FLOWER2 (TFL2) localization, respectively associated with silencing in heterochromatic and euchromatic regions. Finally, we found that a CHROMOMETHYLASE 3 (CMT3) mutation resulted in a dramatic loss of H3K9me2 in the egg cell and abnormal divisions of the zygote following fertilization, revealing a potential role of CMT3 in the reprogramming of the young plant embryo. Here, we show that CMT3 loss-of-function induces transcriptional reactivation of transposable elements and alters TFL2 localization in the female gametophyte. We also propose a model in which the CMT3 egg cell-specific silencing effect could rely on small interfering RNAs (siRNAs).

CMT3 Plays a Gametophyte-Specific Role in Transposon Silencing

CMT3 is known for its role in maintenance of methylation at transposon-related sequences⁴ and is necessary for immobilization of transposons.^5,6 To better understand its potential implication in the reprogramming of the young embryo, we asked whether it could have a role in transgenerational genome integrity by protecting the germline against transposon transcription. We first found by RT-PCR that Athila2 and Atlantys2 were transcriptionally reactivated in both homozygous and heterozygous cmt3-7 mature ovules (Fig. 1A). Since the mutation is recessive,⁷ heterozygous ovules were analyzed to eliminate any sporophytic effect and therefore indicate a gametophytic role for CMT3. To confirm this effect, we took advantage of an enhancer-trap line monitoring transcription of an Atlantys transposable element (ET10306,⁸) and looked for a potential reactivation of this marker in the egg cells of the F1 progeny of cmt3-7 × ET10306 crosses. A strong staining in the egg apparatus was observed in 20% of these ovules (which approximately corresponds to the ¼ proportion expected, n = 229, Fig. 1B), strengthening the idea of a gametophyte-specific role of CMT3 in repressing transcription of transposable elements. Release of transposon silencing in the egg apparatus was also recently described in mutant plants for the AGO9 protein acting in a 24nt siRNA silencing pathway,⁹ which might interact with CMT3 to control egg silencing.

Figure 1.

Transposon reactivation in cmt3-7 mature ovules. (A) Athila2 and Atlantys2 transcripts detection by RT-PCR in wild type (WT), heterozygous (cmt3/+) and homozygous (cmt3/cmt3) cmt3-7 mature ovules. (B) Histochemical detection of GUS reporter expression in mature ovules of the F1 progeny of cmt3-7 × ET10306 (enhancer-trap line monitoring Atlantys transcription) crosses. Bar: 10 µm.

CMT3 is Involved in Euchromatin Silencing in the Egg Cell

While CMT3 is known for its role in maintenance of methylation at non-CG sites, generally related to repeats or transposon sequences,¹⁰ it is not clear whether it could also be functionally associated with gene (i.e., euchromatic region) silencing. We previously showed that the two female gametes were highly dimorphic for the distribution of a LIKE HETEROCHROMATIN PROTEIN1/TERMINAL FLOWER2 (LHP1/TFL2) GFP translational fusion (pTFL2:TFL2-GFP¹¹), with fluorescence signals higher in the egg cell than in the central cell.³ This dimorphism, corresponding to a quantitatively more silenced state of euchromatic domains in the egg cell, correlated with different patterns of H3K9me2, this chromatin modification being specifically driven by CMT3 in the egg cell.³ To test the role of CMT3 in euchromatin silencing in the female gametes, we further examined ovules in the F1 progeny of cmt3-7 × pTFL2:TFL2-GFP crosses. Interestingly, the TFL2 dimorphism was inverted in cmt3-7 gametophytes, with a relative intensity EC/CC = 0.5 +/− 0.3 (n = 20) instead of 2.0 +/− 0.4 (n = 30) (Fig. 2) and a dramatic loss of TFL2 in the egg cell, suggesting that CMT3 is also involved in the silencing of euchromatic regions in this cell type.

Figure 2.

Effect of CMT3 loss-of-function on TFL2 localization in the mature ovule. The F1 progeny of both WT (Ler) × pTFL2:TFL2-GFP [Col(0)] and cmt3-7 (Ler) × pTFL2:TFL2-GFP [Col(0)] crosses was observed and segregated as expected for TFL2 signal in the gametophyte and cmt3 phenotype. (A) Wild type (WT) and cmt3-7 mature gametophytes expressing a pTFL2:TFL2-GFP transgene monitoring LHP1/TFL2 distribution; projection of consecutive optical sections showing the nuclei of the embryo sac and the ovule integuments; ec, egg cell; cc, central cell; sy, synergids; Bar, 10 µm. (B) 3-D reconstruction of WT and cmt3-7 synergids, central cell and egg cell nuclei with statistic coloration corresponding to the intensity mean of each cell type; ec:, egg cell; cc, central cell; sy, synergids. (C) Quantification of the relative TFL2 fluorescence intensity (yellow bars) between the egg cell and central cell (EC/CC) in WT and cmt3-7 gametophytes. The relative intensity is expressed as the ratio EC/CC of the mean intensity per voxel. The error bars represent standard deviation. EC, egg cell; CC, central cell.

A Model for RdDM-Dependent Egg Cell-Specific Silencing

Together, these findings reveal that CMT3 activity is essential for silencing in the egg cell, at both transposon and gene-related sequences. Since CMT3 acts downstream of siRNAs in somatic cells in the so-called RdDM (RNA-directed DNA methylation) pathway,¹² one possibility regarding the nature of this egg cell-specific silencing could be the involvement of siRNAs targeted to the egg cell. In the male gametophyte, transposon silencing in the sperm cells relies on siRNAs generated in the companion vegetative cell.⁸ Interestingly, H3K9 modifications and TFL2 localization in the sperm and vegetative cells¹³ (Fig. 3) are strikingly reminiscent of that observed in the egg and central cells.³ Furthermore, nuclei of both cell types appeared acetylated on the same lysine residue, H3K9ac. This mark is usually linked to permissive euchromatic state but it has been recently shown that when coupled with H3K9me2, it is associated with downregulation of gene expression.¹⁴ Thus, male and female gametophytes present a very similar epigenetic dimorphism between their germ and companion cells, fueling the hypothesis of a common silencing mechanism. Moreover, the discovery of repeat-related siRNAs of maternal gametophytic origin in the endosperm suggests that siRNAs might be produced in the companion central cell to generate transposon silencing in the egg cell and the subsequent developing embryo.^10,15,16 Because germ and companion cells of both male and female gametophytes also show a TFL2 dimorphism, inverted in cmt3 embryo sacs, this process could be extended to gene-related sequences as presented in Figure 4. The first requirement for the proposed model is siRNAs production. Since the egg cell is relatively quiescent, these siRNAs must be produced elsewhere in transcriptionally active cells, which could be either companion gametophytic cells (i.e., the central cell and/or the antipodal cells) or somatic cells surrounding the embryo sac. Whether siRNAs are translocated specifically into the egg cell or are present in other gametophytic cell types remains to be determined. In this regard, we showed that both female gametes are likely targeted similarly since depletion of DEMETER-LIKE (DML) enzymes in the central cell induces an egg cell-like H3K9me2 pattern.³ In this view, the central cell remains protected and transcriptionally active, possibly owing directly to DML activity. The second critical requirement is the recruitment of CMT3 for non-CG methylation, further triggering H3K9 dimethylation, as already shown in somatic cells,¹⁷ and leading to transposon and gene silencing. By establishing a global quiescent state immediately before fertilization, CMT3 may facilitate or initiate reprogramming of the young plant embryo. The siRNAs produced pre-fertilization might further maintain this transcriptional quiescence post-fertilization, by being transmitted to the zygote along with the maternal transcripts sustaining the first few cell divisions.

Figure 3.

H3K9 acetylation/dimethylation and TFL2 localization in wild-type pollen. (A) Wild-type pollen grain stained with DAPI (marking chromatin, white) and H3K9ac (green). (B) Wild-type pollen grain stained with DAPI (white) and H3K9me2 (green). (C) Wild-type pollen grain expressing a pTFL2:TFL2-GFP transgene monitoring LHP1/TFL2 distribution. vcn, vegetative cell nucleus; gcn, generative cell nucleus; scn, sperm cells nuclei; Bar, 10 µm.

Figure 4.

Speculative model of the CMT3-induced egg cell silencing in the female gametophyte. The small interfering RNAs (siRNAs) are produced in transcriptionally active cells (central cell, antipodal cells or somatic cells), then targeted to the egg cell, the other cells being eventually protected by enzymes such as DEMETER-LIKE (DML) in the central cell. Finally, CMT3 is recruited for non-CG methylation, further triggering H3K9 dimethylation and leading to transposon and gene silencing.

Methods

Plant materials.

Arabidopsis thaliana accessions Columbia-0 (Col(0)) and Landsberg erecta (Ler) were used as wild-type controls. The cmt3-7 mutant line was ordered from the Arabidopsis Stock Center (Ler, CS6365,⁷). The ET10306 and pTFL2:TFL2-GFP lines were generously provided by K. Slotkin and K. Goto/C. Baroux, respectively.^3,8,11

RT-PCR.

Mature ovules from emasculated pistils of wild type, homozygous and heterozygous cmt3-7 mutant plants (where the mutation was maternally transmitted) were dissected before RNA extraction. Reverse transcription was performed with random hexamers and PCR as described^5,8 for the Athila2/Atlantys2 transposable elements and the GapC gene, respectively.

GUS staining.

GUS staining of mature ovules (two-day staining, two days after emasculation) was performed as described.¹⁸

LHP1/TFL2-GFP fluorescence observation and quantification.

Pistils or stamens were placed in a drop of 1 M Glycine 1 M (pH = 9.4) on a microscope slide and dissected to isolate the ovules and pollen grains. Images were collected using a laser scanning confocal microscope (LSCM, Leica SP2, Leica, Germany) with an excitation line at 488 nm and emission window at 500–530 nm. The intensity mean per voxel of the TFL2-GFP fluorescence signal in the different cell types of the embryo sac was collected in 3-dimensional reconstructions using Imaris software (Bitplane, CH).

Pollen immunolocalization.

Immunolocalization of chromatin marks in pollen was performed as described,³ except for sample preparation that were digested in an enzymatic solution composed of 1% driselase, 1% cellulase, 0.3% pectolyase, 1% glucanase, all from Sigma. The primary antibodies were used at the following dilutions: 1:200 for H3K9ac (ab10812, ABCAM, UK) and 1:400 for H3K9me2 (ab1220, ABCAM, UK).