Paternal imprinting of dosage-effect defective1 contributes to seed weight xenia in maize

Abstract

Historically, xenia effects were hypothesized to be unique genetic contributions of pollen to seed phenotype, but most examples represent standard complementation of Mendelian traits. We identified the imprinted dosage-effect defective1 (ded1) locus in maize (Zea mays) as a paternal regulator of seed size and development. Hypomorphic alleles show a 5-10% seed weight reduction when ded1 is transmitted through the male, while homozygous mutants are defective with a 70-90% seed weight reduction. Ded1 encodes an R2R3-MYB transcription factor expressed specifically during early endosperm development with paternal allele bias. DED1 directly activates early endosperm genes and endosperm adjacent to scutellum cell layer genes, while directly repressing late grain-fill genes. These results demonstrate xenia as originally defined: Imprinting of Ded1 causes the paternal allele to set the pace of endosperm development thereby influencing grain set and size.

Fig. 1. Ded1 is a transcription factor required for kernel development.

a Self-pollinated ears segregating for ded1-ref in W22, B73, and Mo17 genetic backgrounds. Arrowheads indicate mutants. Scale bars are 1 cm. b Kernel phenotype and sagittal sections of the ded1-ref and normal sibling in W22. Scale bars are 5 mm. Red arrowheads indicate the embryo, and blue arrowhead indicates vitreous endosperm. c Normal and ded1-ref sibling plants. d Self-pollinated normal sibling ear at 19 DAP. e Homozygous ded1-ref ear at 19 DAP. f Schematic of the B73_v4 genome annotation for the Ded1 gene. Boxes are exons with coding sequences in black. Black lines are introns. Orange lines are the qRT-PCR products used in Figs. 1h, 2a, and 4a. The triangle indicates the ded1-ref retrotransposon insertion. The 5′ transposon junction sequence is highlighted in blue. g Schematic of DED1 protein domains showing R2 (blue) and R3 (red) MYB DNA binding domains, the nuclear localization signal (yellow), and C-terminal acidic domain (green). The triangle indicates the ded1-ref insertion. Protein sequences of the Cas9-induced frameshifts are highlighted in fuchsia. h Endosperm expression of the ded1 locus in W22 and homozygous ded1 mutant alleles at 12 DAP. Relative qRT-PCR used 18 S rRNA as the control. Data are mean ± SD (n = 3 replicate PCR experiments from pooled mutant endosperm samples from a single ear). Letters denote significant differences (P < 0.05) from Tukey’s HSD test. i Average kernel weight based on 50 homozygous mutant seeds from heterozygous ded1-ref (both W22 and B73 background), ded1-3 (B73), and ded1-4 (W22) self-pollinations and reciprocal crosses. Female parent is listed first. Mutant seeds from three ears were weighed for each genetic combination. Data shown as mean ± SD (n = 3 biologically independent ears from different plants with 50 mutant kernels sampled per ear). Letters denote significant differences (P < 0.05) from Tukey’s HSD test.

Fig. 2. Endosperm expression and paternal imprinting of Ded1 affects kernel weight.

a Expression of Ded1 in dissected W22 endosperm and embryo tissues. Transcript levels were normalized to 18 S rRNA. Data shown as mean ± SD (n = 3 replicate PCR experiments of pooled tissue samples from individual ears). b Allele-specific expression of Ded1 in 12 DAP endosperm tissue comparing inbred self-pollinations and three biological replicates of reciprocal crosses between W22 and Mo17. The female parent is listed first with the endosperm genetic dosage below. Proportional mixes of cDNA derived from W22 and Mo17 inbred are indicated with fractional mix ratios. RT-PCR products were digested with AluI to digest W22 products. c, d Box plots of homozygous normal and ded1 heterozygous sibling kernels for maternal (c) and paternal (d) transmission of ded1. Boxes are the interquartile range with the median denoted by a horizontal line. Whiskers show the 1.5× interquartile range. Points show individual kernel weights with outlier kernels denoted with an asterisk. For each cross, the female parent is listed first. Homozygous normal (DDD) kernels are plotted in blue. Heterozygous kernels are plotted in orange. The x-axis also indicates the number of kernels plotted and p values for two-sided Student’s t tests without correction for multiple tests. Individual kernel weights and descriptive statistics are in the Source Data file.

Fig. 3. Identification of DED1 target genes.

a Number of DED1 DAP-seq peaks per 100 bp bin relative to annotated transcriptional start sites. Orange line shows the bound promoter region used to identify direct target genes. b Sequence logo of top DED1 binding motifs identified by meme-chip v4.12.0 using the 5000 highest scoring DAP-seq peaks. c Venn diagram showing genes bound by DED1 from −1 to +0.1 kb of the TSS that were tested for differential expression (orange) as well as DEGs from 12 DAP ded1-ref and normal sibling endosperm (q < 0.05, FC > 2, and TPM > 1). Overlapping genes (white text) are inferred to be direct targets of DED1, including 258 DED1-activated and 180 DED1-repressed genes. d Genome browser visualization of DAP-seq reads at fl3. e EMSA with DED1-GST purified protein and labelled fl3 promoter. Normal (Nor) or mutant (Mut) competitor probes were added at 100-fold greater concentration than the labelled fl3 probe. This experiment was completed twice with similar results. f Public maize transcriptome data for Ded1 and DED1 targets with known functions in endosperm development^,,. The color scale indicates the relative expression level of each gene. Expression of the se1 locus was not detected in the 8 DAP dissected tissues.

Fig. 4. Dosage-dependent regulation of downstream DED1 targets.

a Expression of the normal allele of Ded1 and DED1 downstream genes by qRT-PCR in four dosage states of Ded1 (D) and ded1-ref (d). Genes analyzed included the fl3 direct target, de18 potential direct target, sweet4c activated DEG, tcrr1 imprinted DEG, az22z5 α-zein repressed DEG, and a2 repressed DEG. Data points are averages of three technical replicates. Bars and error bars are the mean ± SD of three biological replicates from independent ears (n = 3). Letters denote significant differences (P < 0.05) from Tukey’s HSD test. b Schematic of DED1 target and downstream loci with documented roles in endosperm development. DED1 activates loci that act early in endosperm development and in nutrient transfer tissues. DED1 represses loci acting later during grain fill.