Control of DEMETER DNA demethylase gene transcription in male and female gamete companion cells in Arabidopsis thaliana

Abstract

The DEMETER (DME) DNA glycosylase initiates active DNA demethylation via the base-excision repair pathway and is vital for reproduction in Arabidopsis thaliana DME-mediated DNA demethylation is preferentially targeted to small, AT-rich, and nucleosome-depleted euchromatic transposable elements, influencing expression of adjacent genes and leading to imprinting in the endosperm. In the female gametophyte, DME expression and subsequent genome-wide DNA demethylation are confined to the companion cell of the egg, the central cell. Here, we show that, in the male gametophyte, DME expression is limited to the companion cell of sperm, the vegetative cell, and to a narrow window of time: immediately after separation of the companion cell lineage from the germline. We define transcriptional regulatory elements of DME using reporter genes, showing that a small region, which surprisingly lies within the DME gene, controls its expression in male and female companion cells. DME expression from this minimal promoter is sufficient to rescue seed abortion and the aberrant DNA methylome associated with the null dme-2 mutation. Within this minimal promoter, we found short, conserved enhancer sequences necessary for the transcriptional activities of DME and combined predicted binding motifs with published transcription factor binding coordinates to produce a list of candidate upstream pathway members in the genetic circuitry controlling DNA demethylation in gamete companion cells. These data show how DNA demethylation is regulated to facilitate endosperm gene imprinting and potential transgenerational epigenetic regulation, without subjecting the germline to potentially deleterious transposable element demethylation.

Keywords: DNA demethylation; DNA enhancer elements; cell-specific transcription; central cell; vegetative cell.

Fig. 1.

DME is specifically expressed in the vegetative nucleus of late bicellular stage pollen. (A) Sequential development of flowers (Top) and corresponding pollen development in 2.3 kb DME::GUS transgenic plants with DAPI (Middle) and GUS staining (Bottom). (B) The 2.3 kb DME::GFP expression (Left) in microspore (Top), bicellular (Middle), and tricellular (Bottom) stage pollen grains stained with DAPI (Right). G, generative nucleus; N, microspore nucleus; S, sperm cell nucleus; V, vegetative nucleus. (Scale bars: 5 μm.) (C) qRT-PCR analysis of DME expression in WT pollen development after normalization with ACT1, ACT3, and ACT12 expression. The four different stages analyzed using qRT-PCR are indicated in A. Values are plotted relative to the expression of DME in stage 4 mature pollen, which was set at 1.0, and represent the average of triplicate measurements ± SD.

Fig. S1.

Diagram and expression of a 2.3pDME::GUS reference line and two complementing constructs. (A) Diagram (Top) and expression pattern (Middle and Bottom) of a 2.3pDME::GUS reference line. The number in black before the slash is based on a TSS that we defined using 5′ RACE (see Fig. S2). The number in red after the slash is based on the translation start site of At5g04560.2, the major splice variant. (Middle) Images of seedlings at 7 days after germination (DAG 7), showing shoot meristems on the Left and root tips on the Right. (Bottom) Pistils on the Left and an ovule at higher magnification on the Right. (B) Diagram of the two cDME complementation constructs with 2.3kb upstream sequences (2.3kb cDME) or deleted 5′ UTR (+46 cDME). Both constructs contain the first and second introns.

Fig. 2.

Diagram of the DME::GUS reporter constructs and expression of the T-DNA insertion lines in the DME region. (A) The name, staining intensity, and the coordinates for each construct are shown. CC, central cells; VC, vegetative cell of pollen; −, none; +, moderate; ++, strong. (B) GUS staining is shown in ovules and pollen. DAPI-stained pollen grains are shown in the Bottom row. Plants expressing transgenes 2.3kb to +395 displayed GUS expression in the central cell nucleus (arrow) and vegetative cell nucleus. No GUS expression was detected in +473 transgenic plants, and 2.3kb Pro. plants exhibited GUS expression only in the synergid cells (arrowhead). (Scale bars: ovule, 50 μm; pollen, 20 μm.) (C) dme T-DNA insertion alleles at 72 nt upstream (CS857766) and at 25 nt upstream (SALK-036171) of the TSS. Black box, translated exon; gray box, untranslated exon; first line, 5′ flanking sequences; other lines, intron. (D) qRT-PCR analysis of DME expression in homozygous dme mutant seedlings after normalization with ACT1, ACT3, and ACT12 expression. Values are plotted relative to the expression of DME in Ler WT, which was set at 1.0, and represent the average of triplicate measurements ± SD.

Fig. S2.

The 5′ RACE analysis of DME using inflorescence RNA. (A) Diagram of the two alternative forms of the DME 5′ region. Light gray box, 5′ UTR; dark gray box, translated exon; front line, 5′ flanking region; second line, first intron; black arrow, 5′ RACE primer; GSP, gene-specific primer; TSS, transcription start site. (B) Agarose gel electrophoresis of DME 5′ RACE products from total inflorescence RNA using the GSP2 primer. (C) qRT-PCR analysis to compare the relative expression levels of the two DME splice variants, At5g04560.1 and At5g04560.2, in aerial tissue and root tissue of 2-week-old seedling and in immature inflorescences. Expression levels are normalized to the ACT2 housekeeping gene and represent the average of triplicate measurements ± SD. (D) Relative distribution of DME TSS determined by 5′ RACE clones. Nucleotide positions relative to the start codon are indicated on the x axis. The graph summarizes the results of 24 RACE clones.

Fig. S3.

DME expression driven by the +46 minimal reproductive promoter transgene rescues dme-2-mediated seed abortion. (A) The +46 pDME:GFP transgene is expressed in the central cell before fertilization, but not in the endosperm after fertilization. (Scale bars: 50 μm.) (B) Siliques of dme-2 mutants containing 2.3kb cDME or +46 cDME complementing constructs. See also Fig. S1 and Table S2. (Scale bar: 1 mm.)

Fig. 3.

DME expression driven by the +46 transgene can correct the methylation phenotype of homozygous dme-2 mutant endosperm. (A) Snapshots of CG DNA methylation at selected imprinted loci. Each track represents a different genotype: crimson trace, WT subtracted from dme-2 homozygous endosperm expressing the +46 transgene; orange trace, WT subtracted from dme-2 heterozygous endosperm; green tracks are raw CG methylation data in the three genotypes compared. Differential methylation at both maternally expressed (FIS2, FWA) and paternally expressed (YUK10, PHE1) imprinted loci (i.e., maternal hypomethylation of imprinting control regions) is regained in dme-2 homozygous endosperm when the +46 transgene is expressed. Gray boxes show the imprinting control regions at each locus, and arrows show the direction of gene transcription. (B) Kernel density plots of CG methylation differences between the maternal alleles of dme-2 homozygous endosperm expressing the +46 transgene and WT (i, crimson trace) and dme-2 heterozygous endosperm and WT (ii, orange trace,). Hypermethylation of the dme-2 mutant endosperm is evident in the increased density at a fractional methylation difference of between 0.5 and 1 in ii and is corrected by the +46 transgene as seen by the loss of this density increase in i. (C) Kernel density plots of CG methylation differences between the paternal alleles of dme-2 homozygous endosperm expressing the +46 transgene and WT (i, blue trace) and dme-2 heterozygous endosperm and WT (ii, aquamarine trace). Methylation of the paternal (WT Ler) alleles is the same in each genotype, showing that the +46 transgene does not affect methylation postfertilization.

Fig. S4.

DME expression driven by the +46 minimal reproductive promoter transgene can correct the methylation phenotype of homozygous dme-2 mutant endosperm. (A) Kernel density plots of CG methylation differences between the maternal alleles of dme-2 mutant endosperm (3) and dme-2 mutant endosperm expressing the +46 transgene (i, red trace) and dme-2 heterozygous endosperm and WT (ii, orange trace). Similar hypermethylation of the dme-2 mutant endosperm relative to both the WT and the dme-2 mutant expressing the +46 transgene is evident in the increased density at a fractional methylation difference of between 0.5 and 1 in both plots. Compared with the dme-2 mutant endosperm, the same genomic sites are relatively hypomethylated in WT and the +46 transgene DNA. (B) Kernel density plots of CG methylation differences between the maternal alleles of dme-2 mutant endosperm and dme-2 mutant endosperm expressing the +46 transgene for (i) all genomic sites (ii) DME target sites only.

Fig. 4.

Diagram of the DME::GUS reporter constructs for fine mapping of cis-elements and their expression patterns. The TU (truncated 5′-UTR) series of constructs. (A) The name, staining intensity, and coordinates for each construct are shown. CC, central cells; VC, vegetative cell of pollen; −, none; +, moderate; ++, strong. (B) GUS staining is shown in ovules and pollen. DAPI-stained pollen grains are shown in the Bottom row. TU0, TU34, and TU45 transgenic plants exhibited GUS expression in the central cell and pollen. No GUS expression was detected in TU12 and TU23 plants. (Scale bars: 50 μm.)

Fig. S5.

Diagram of the DME::GUS reporter constructs for fine mapping of cis-elements. The TU (truncated 5′ UTR) series of constructs. The name, staining intensity, and the coordinates for each construct are shown. CC, central cells; SDL, seedling; −, none; +, moderate; ++, strong.

Fig. S6.

Catalog of the expression patterns of the TU DME:GUS construct series. TU12, TU13, and TU23 plants showed weak GUS signal only in sporophytic tissues. By contrast, TU34, TU35, and TU45 plants showed GUS expression only in central cells. (Scale bars: seedlings, 1,000 μm; shoot meristem and pistil, 200 μm; ovule, 50 μm.)

Fig. 5.

Internal deletion/substitution of cis-elements. (A) Summary of DME cis-regulatory elements. Dark gray box, translated exon; light gray box, 5′-UTR; line, first intron; red line, sporophytic element (SPE); blue line, central cell element (CCE); green line, pollen vegetative cell element (VCE) ; dotted line, quantitative regulatory element (QE). (B) Diagram of DME::GUS internal deletion and substitution constructs of the cis-elements. CC, central cells; VC, vegetative cell of pollen; −, none; (+), weak; ++, strong; Δ, deletions or substitutions. (C) GUS staining is shown in ovules and pollen. DAPI-stained pollen grains are shown in the bottom of each pollen. TU0_ΔSP, same GUS expression pattern as TU0; TU0_ΔPOL, central cell and pollen GUS disappeared; TU0_ΔCC1 and TU0_ΔCC2, only the pollen expression disappeared; TU0_ΔCC3, central cell and pollen GUS disappeared. TU0_ΔHB, central cell GUS was significantly reduced and pollen GUS disappeared. (Scale bars: ovule, 50 μm; pollen, 20 μm.)

Fig. S7.

Diagram of the DME::GUS reporter construct series and their expression patterns. CC, central cells; SDL, seedlings; −, none; +, moderate; ++, strong.

Fig. S8.

Catalog of the expression patterns of the DME::GUS deletion series. The 2.3 kb, 0.5 kb, +7 DME::GUS, and 2.3kb Δ5′ constructs showed GUS expression in both sporophytic tissues and central cells. The +20, +33, +46, and +396 were expressed only in central cells. No expression was detected in +473 DME::GUS plants. The 2.3 kb Pro. exhibited DME:GUS expression in sporophytic tissues, but not in the central cell nucleus. Ectopic expression was detected in the micropylar end of the embryo sac. (Scale bars: seedling, 1,000 μm; shoot meristem and pistil, 200 μm; ovule, 50 μm.)

Fig. S9.

Diagram of the gain-of-function VCE tandem repeat constructs with and without the 35S minimal promoter. A scheme is shown at left. The name of the construct is shown, and the presence of staining in seedlings (SDL) and central cells (CC) is indicated.

Fig. S10.

DME and ROS1 homolog comparisons in publically available Brassica family DNA sequences. Lines above the DNA sequence indicate the cis-regulatory elements found by these experiments. SPE in the DME CT-repeats (red box) and the 9 bp of sequence that are similar to the pseudopalindromic target sequence that is similar to Arabidopsis thaliana Homeobox 1 (Athb-1) (blue box) are well conserved in Brassica family. A.Ly, Arabidopsis lyrata; A.Th, Arabidopsis thaliana; B.Ra, Brassica rapa; C.Ru, Capsella rubella.