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

The study investigates how paternal genomic imprinting influences seed development and size in maize, focusing on the dosage-effect defective1 (ded1) locus. In angiosperm seeds, endosperm development has a predominant maternal genomic influence, but classical observations of xenia suggested paternal contributions to seed phenotypes. Most xenia cases reflect Mendelian dominance rather than true parent-of-origin effects. Genomic imprinting provides a mechanism for parent-specific gene expression to affect seed traits. While over 100 imprinted genes are known in maize, functional evidence for kernel development has been largely limited to maternally expressed genes (MEGs) such as meg1 and floury3 (fl3), and no paternally expressed genes (PEGs) had been demonstrated to have direct roles in kernel development. The authors hypothesize that ded1 is a quantitative PEG that regulates endosperm developmental progression and seed size through dosage-sensitive transcriptional control, thereby providing a mechanistic example of xenia.

Background literature establishes: (1) Xenia, historically viewed as paternal influence on seed phenotype, often reflects Mendelian effects rather than imprinting-based parent-of-origin expression. (2) The first imprinted gene discovered in maize was R′, with maternal bias affecting anthocyanin accumulation. (3) Allele-specific expression studies have identified more than 100 imprinted genes in maize; however, only a few MEGs (e.g., meg1 and fl3) have demonstrated qualitative seed phenotypes, and prior to this study, no PEGs were shown to directly regulate kernel development, though paternal-effect mutants exist. (4) Public transcriptome datasets indicate Ded1 expression peaks in early endosperm (5–13 DAP, peak ~6 DAP) and shows paternal-biased expression in reciprocal hybrid analyses. This body of work motivates testing Ded1 as a quantitative PEG influencing seed development and seed weight.

- Genetic screening and mapping: From 1068 UniformMu segregating ears, normal kernels were weighed and analyzed by near-infrared spectroscopy to detect dosage-effect mutants. Bulked segregant analysis and fine mapping localized ded1 to a 470 kbp interval on chromosome 1. Genomic sequencing identified a copia-like retrotransposon insertion in ZmMyb73 (Ded1). PCR and cDNA sequencing confirmed the gene model (B73_v4 Zm00001d033265) and characterized ded1-ref transcript truncation. - Gene editing and complementation: CRISPR-Cas9 generated four indel alleles (ded1-1 to ded1-4) causing premature stop codons, all failing to complement ded1-ref and producing kernel defects. qRT-PCR showed reduced ded1 transcript in homozygous mutants, consistent with nonsense-mediated decay. - Expression and imprinting assays: qRT-PCR across seed compartments and time confirmed early endosperm-specific expression. Allele-specific CAPS RT-PCR using an AluI site distinguished W22 and Mo17 alleles in 12 DAP endosperm from reciprocal crosses to quantify paternal bias. Public RNA-seq datasets (B73×Mo17) were reanalyzed for parental bias at 10 and 14 DAP. - Kernel weight dosage analysis: Reciprocal crosses using heterozygous ded1/+ with inbreds (W22, B73, Mo17) produced ears segregating for dosage states (DD/D, dd/D, DD/d, dd/d). For each ear, 96 kernels from the middle were weighed and genotyped to correlate genotype with individual kernel weight. - Protein localization and activity: Subcellular localization was tested by transient expression of DED1-GFP in maize protoplasts alongside an NLS-RFP marker (confocal microscopy). Yeast GAL4-BD autoactivation assays with full-length and truncated DED1 constructs mapped a C-terminal acidic activation domain; ded1-ref partial acidic domain retained activation; truncations mimicking ded1-3/-4 lacked activation. - Transcriptomics (RNA-seq) and differential expression: Strand-specific libraries from 12 DAP endosperm (ded1-ref homozygous mutants vs normal siblings) were sequenced, aligned (GSNAP), and analyzed with DESeq2 (TPM>1; adjusted p<0.05; |log2FC|>2). GO enrichment was performed with agriGO. - DAP-seq to map DED1 binding: HALO-DED1 was used to affinity-purify bound genomic fragments; 75 bp single-end NextSeq reads were aligned (bowtie2). Peaks were called and annotated relative to TSS. Motif enrichment identified MYB-like binding motifs. Direct targets were defined as genes with DAP-seq peaks within -1 kbp to +100 bp of TSS and differential expression. - EMSA validation: Recombinant DED1-GST was tested for binding to promoter fragments at selected targets (fl3, sus1, c1, vp1). Competition with unlabeled normal and motif-mutant probes assessed specificity. - Dosage-response qRT-PCR: Controlled crosses generated a Ded1 dosage series in 12 DAP endosperm (DDD, ddD, DDd, ddd). qRT-PCR assayed wild-type Ded1 (spanning the ded1-ref insertion junction) and several downstream targets (direct and indirect) to quantify dosage-dependent transcriptional changes. - Histology and developmental analyses: Embryo and endosperm histology at 12 DAP assessed developmental progression, including BETL formation and embryo stages. Public compartment-specific RNA-seq datasets (6–28 DAP; microdissections at 8 and 13 DAP) were reanalyzed to map temporal and spatial expression of Ded1 and its direct targets (e.g., EAS, ESR, aleurone, embryo).

- Identification of Ded1 as ZmMyb73 (R2R3-MYB TF) with a retrotransposon insertion in ded1-ref causing a truncated protein lacking the C-terminal acidic activation domain. CRISPR-induced loss-of-function alleles phenocopied ded1-ref and did not complement it. - Paternal imprinting and dosage effect: Ded1 is a quantitative PEG with strong paternal bias in endosperm. Paternal allele accounts for ~75–77% of total Ded1 transcript at 10 DAP and ~53–68% at 14 DAP in reciprocal B73×Mo17 crosses. CAPS RT-PCR in W22×Mo17 hybrids showed ~2/3 of transcript from the paternal allele at 12 DAP. - Seed weight/xenia: Paternal transmission of ded1 reduced kernel weight by 13–20 mg in DD/d kernels compared with DDD siblings, depending on background and allele. When ded1/+ served as the female, dd/D and DD/D had similar weights in 8/9 biological replicates, indicating paternal allele dosage drives the phenotype. Hypomorphic ded1 alleles caused a 5–10% seed weight reduction when transmitted through pollen; homozygous mutants had a 70–90% reduction. - Transcriptional network: RNA-seq at 12 DAP identified 2072 DEGs (1022 down, 1050 up in ded1). DAP-seq revealed 43,225 DED1-enriched peaks near 15,367 genes, with 61.6% within 1 kbp of TSS/stop; peaks concentrated within 100 bp upstream of TSS. Motif resembled Arabidopsis MYB119. Among genes with promoter binding and endosperm expression, 438 were direct DED1 targets: 258 activated and 180 repressed (significant enrichment, cumulative hypergeometric p=2.2×10^-5). EMSA validated binding to fl3, sus1, c1, and vp1. - Temporal-spatial regulation: Ded1 and activated targets peak at 6–8 DAP, with 13 DAP restriction to EAS. DED1-repressed targets peak after 10 DAP and are enriched in embryo, pericarp, and scutellar aleurone at 13 DAP. Ded1 expression is endosperm-specific at 8 DAP (minor in aleurone) and EAS-specific at 13 DAP. - Developmental consequences: Homozygous ded1 mutants fail to support embryo development (embryo arrest at pre-transition) and show incomplete BETL. Activated targets include regulators of endosperm proliferation and differentiation (e.g., fl3, cyc8), epigenetic regulation (ZmJmj11), and auxin biosynthesis (tar1; de18 as potential direct target). Repressed targets include anthocyanin regulators (c1, pl1) and late grain-fill genes (e.g., se1). - Dosage-dependent transcription: In dosage series qRT-PCR at 12 DAP, paternal ddD had ~50% Ded1 transcript relative to DDD, while maternal DDd had ~15%. Direct targets showed thresholds: fl3 increased 133-fold in ddD vs ddd but only ~20-fold in DDd; de18 increased 149-fold (ddD) vs 32-fold (DDd). Indirect targets reflected activation/repression patterns (e.g., sweet4c reduced ~3-fold in ddd; tcrr1 decreased progressively with ded1 dosage; az22z5 and a2 increased in ddd). - Imprinting among targets: Several MEGs and PEGs are direct DED1 targets; 8/9 PEGs are activated by DED1, including tar1; de18 is a PEG potentially directly activated by DED1.

The findings demonstrate that Ded1 is a paternally biased, dosage-sensitive transcription factor that sets the pace of endosperm development, providing a mechanistic instance of xenia as originally conceived. Paternal expression of a single functional Ded1 allele is sufficient for near-normal seed size and embryo support, whereas maternal-only expression yields smaller seeds, aligning with kinship theory predicting PEGs enhance resource acquisition for the developing seed. DED1 orchestrates early endosperm programs and nutrient-transfer interfaces (EAS/ESR) while delaying late grain-fill gene expression, consistent with promoting early growth and embryo nourishment. The study positions ded1 as the first clear PEG in maize with a direct role in kernel development, contrasting with the paucity of PEG functional evidence in other angiosperms. The overlap between DED1 target classes (auxin biosynthesis, epigenetic regulators, transcription factors, transporters) and those implicated by kinship theory underscores the evolutionary logic of paternal imprinting in cereals. Species differences in imprinting conservation suggest DED1-like xenia effects may be restricted to cereals, though analogs (e.g., Arabidopsis PHE1-regulated pathways) hint at broader principles.

Ded1 (ZmMyb73) is identified as a major, paternally biased transcriptional regulator of maize seed development that quantitatively controls endosperm proliferation, differentiation, and nutrient transfer, thereby influencing embryo development and final seed weight. Through extensive binding and expression analyses, DED1 directly activates early endosperm and EAS genes and represses late grain-fill genes, with paternal dosage dictating kernel weight outcomes. The work provides strong support for kinship theory by demonstrating a PEG that enhances resource allocation to the seed. Future research should: (1) define DED1’s dynamic interactome and cofactors across endosperm stages; (2) dissect cell type–specific regulatory circuits (EAS, ESR, BETL) driven by DED1; (3) resolve causal regulatory variants underlying natural variation in DED1 dosage effects; (4) test conservation and functional analogs in other cereals and evaluate agronomic manipulation of DED1 pathways to optimize yield components.

Differential expression analyses were conducted on 12 DAP endosperm comparing homozygous ded1 mutants to pooled normal siblings containing mixed Ded1 dosages (three doses of the normal allele), potentially reducing statistical power to detect DEGs attributable specifically to Ded1 dosage. Sampling occurred after the 6 DAP window when Ded1 expression peaks, possibly missing earlier primary transcriptional responses. Steady-state transcript measurements conflate direct DED1 targets with indirect feedback effects, complicating target identification. Additionally, imprinting patterns and DED1 functions may not be conserved across angiosperms, limiting generalizability beyond cereals.