United by conflict: Convergent signatures of parental conflict in angiosperms and placental mammals

Abstract

Endosperm in angiosperms and placenta in eutherians are convergent innovations for efficient embryonic nutrient transfer. Despite advantages, this reproductive strategy incurs metabolic costs that maternal parents disproportionately shoulder, leading to potential inter-parental conflict over optimal offspring investment. Genomic imprinting-parent-of-origin-biased gene expression-is fundamental for endosperm and placenta development and has convergently evolved in angiosperms and mammals, in part, to resolve parental conflict. Here, we review the mechanisms of genomic imprinting in these taxa. Despite differences in the timing and spatial extent of imprinting, these taxa exhibit remarkable convergence in the molecular machinery and genes governing imprinting. We then assess the role of parental conflict in shaping evolution within angiosperms and eutherians using four criteria: 1) Do differences in the extent of sibling relatedness cause differences in the inferred strength of parental conflict? 2) Do reciprocal crosses between taxa with different inferred histories of parental conflict exhibit parent-of-origin growth effects? 3) Are these parent-of-origin growth effects caused by dosage-sensitive mechanisms and do these loci exhibit signals of positive selection? 4) Can normal development be restored by genomic perturbations that restore stoichiometric balance in the endosperm/placenta? Although we find evidence for all criteria in angiosperms and eutherians, suggesting that parental conflict may help shape their evolution, many questions remain. Additionally, myriad differences between the two taxa suggest that their respective biologies may shape how/when/where/to what extent parental conflict manifests. Lastly, we discuss outstanding questions, highlighting the power of comparative work in quantifying the role of parental conflict in evolution.

Keywords: endosperm; genomic imprinting; hybrid inviability; kinship theory; placenta.

Fig. 1.

Imprinting cycles in mammals and angiosperms. (a) Imprinting cycle in mammals: de novo imprints are established during gametogenesis in sperm and egg via DNA methyltransferases 3a/b (DNMT3a/b). This is followed by somatic maintenance, mostly via DNMT1, throughout development in the embryo and placenta except for the primordial germ cells that migrate to the developing gonads where imprints are erased allowing for de novo re-establishment in gametes. (b) Imprinting cycle in angiosperms: differential imprinting is established in late gametogenesis: in male and female gametophytes, the vegetative and central cells are hypomethylated via DNA glycosylase DME leading to an increased TE activity and the activation of MEGs in the central cell. siRNAs, produced from the increased TE activity, can then migrate from the vegetative/central cell to egg/sperm cells triggering a DNA methylation response. In the sperm cell, DNA methylation is established via RdDM and METHYLTRANSFERASE 1 (MET1). Additionally, the hypomethylation of the central cell can also trigger the deposition of histone H3 lysine-27 tri-methylation (H3K27me3) via Polycomb Repressive Complex 2 (PRC2). After fertilization, imprinting is maintained in the endosperm throughout seed development while the embryo itself shows minimal imprinting. Once development is complete, the imprinted endosperm is fully consumed in dicots, although it persists in monocots. In both cases, it does not contribute to the next generation. As imprinting is rarely detected in the mature embryos, imprints must be re-established in late gametogenesis each generation. Female and male alleles are represented in yellow and blue, respectively. White boxes represent genes with biallelic expression, while black boxes with lollipops represent imprinted genes. Gray dashed arrows denote enzymes or protein complexes recruited.

Fig. 2.

Regulation of imprinting in angiosperms and placental mammals. The four panels show the regulation of MEGs and PEGs in angiosperms (a and b) and placental mammals (c and d). (a) In both the female and male gametophyte, DME actively removes methylation from the central cell and vegetative cell, respectively. Paternal alleles of MEGs are then re-methylated via RdDM, and methylation status can be reinforced via METHYLTRANSFERASE1 (MET1), resulting in repressed expression. Conversely, maternal copies of MEGs are hypomethylated and expressed. (b) PEGs can be expressed through multiple mechanisms: 1) DNA methylation via MET1 or RdDM may prevent deposition of tri-methylate lysine 27 on histone H3 (H3K27me3) by FIS-PRC2, allowing for expression. 2) Paternal alleles of PEGs are neither methylated nor the target of H3K27me3 deposition, allowing expression. Or 3) H3K27me3 marks are removed, allowing expression. The maternal alleles of PEGs are silenced via FIS-PRC2 deposition. (c) MEGs expression in mammals is achieved by the suppression of the paternal allele either through DNA methylation via DNA methyltransferase 3a/b (DNMT3) at imprinting control regions (ICRs) or the binding of Polycomb repressive complex (PRC). (d) PEGs expression in mammals is achieved by the suppression of the maternal alleles through DNA methylation at ICRs via DNMT3a/b, PRC, or the histone methyltransferase G9a. In all cases, a blunted arrow indicates that expression is inhibited, while a pointed arrow denotes that expression occurs (indicated by the black waves).

Fig. 3.

Differences in the strength of parental conflict are caused by factors that affect the extent of genetic relatedness of siblings. Many demographic and life history traits can influence the extent to which offspring are related via their maternal- and paternal-inherited alleles (denoted by solid lines). While some of these factors, such as mating system, have much support in both mammals and plants, others, such as sex-biased dispersal or survival have received much less attention. We also note that the specific factors (denoted by dashed lines) that comprise “life history” or “demography” can also interact; for example, aspects of demography can influence mating system. Lastly, we denote some factors with a question mark as theoretical explorations of how these factors influence parental conflict are an open question.

Fig. 4.

Four genetic models of imprinting divergence that could contribute to parent-of-origin growth effects in hybrids. In all models, we show the evolution of 2 loci (denoted by circles vs squares), with maternal and paternal alleles (denoted by yellow and blue, respectively). Pointed arrows denote a positive role in gene expression, resulting in gene products (waves), while blunt-ended arrows denote a negative effect on gene expression. Gray shapes denote species-specific transcription factors that regulate imprinting. In all cases, species 1 denotes a taxon with a stronger history of parental conflict, while species 2 denotes a taxon with a weaker history of parental conflict. In all four panels, hybridization results in transgressive expression profiles and an imbalance of maternal:paternal expression at locus 1 and locus 2. (a) Loss of Imprinting (LOI) in hybrids could be caused by the origin of novel imprinted MEGs and PEGs. Species 1 evolves a new MEG and PEG (locus 1 and 2, respectively), while species 2 maintains the ancestral biallelic expression at both loci. The resultant hybrids exhibit either: loss of imprinting and excess maternal expression from locus 2 when species 1 serves as the dam, or loss of imprinting and excess paternal expression from locus 1 when species 2 serves as the dam. (b) Divergence in the extent of imprinted genes (IGs) expression. Here, species 1 and 2 differ in the extent of expression at a gene that is imprinted in both species. Both reciprocal hybrids do not exhibit a loss of imprinting, rather, they have an imbalance of maternal:paternal expression. (c) X-chromosome-dependent Imprinting. In both species, X-chromosomes mediate the expression of imprinted autosomal genes via species-specific transcription factors. X-autosomal incompatibilities are caused by allelic divergence between an imprinted autosomal gene and its X-chromosome regulator. In hybrids, normal imprinted expression of the incompatible X-chromosome does not properly regulate the imprinted autosomal locus, resulting in diminished autosomal expression. We note, however, that this relationship could manifest in many ways, including increased expression of the autosomally imprinted locus or loss of imprinting entirely. (d) Dosage imbalance caused by allelic divergence of non-imprinted regulatory genes. In both species, a non-imprinted autosomal locus mediates the expression of imprinted genes. In hybrids, both the incompatible and compatible regulatory alleles are expressed and compete, leading to a decreased expression of imprinted genes. We again note that we have depicted this scenario where incompatibility results in reduced gene expression, but incompatibility could also result in increased expression or loss of imprinting.