Genome shock in monkeyflower hybrids

In angiosperms, double fertilization during sexual reproduction produces seeds with an embryo and an endosperm. Similar to the placenta in mammals, the endosperm nourishes and supports the growth of the embryo. Hybrids resulting from viable interspecies crosses often exhibit increased overall fitness compared to their parents, a phenomenon known as heterosis or hybrid vigor. The underlying mechanisms remain elusive, but their investigation holds considerable interest for agriculture, where hybrids are widely used for their superior performance and are considered key to address increasing food demands of an ever-growing world population (Hochholdinger and Baldauf, 2018).

However, interspecies crosses often result in seed lethality characterized by endosperm developmental defects. The endosperm is a triploid tissue, with two copies of the maternal genome and a single copy of the paternal genome. This 2:1 ratio of maternal:paternal genomes is essential for regular endosperm development. The distinct endosperm phenotypes observed in interspecies crosses are recapitulated in interploidy crosses, in which the endosperm genome balance is changed, suggesting that interspecies crosses result in changes in parental genome dosage in the endosperm. Furthermore, the endosperm is a tissue where genomic imprinting occurs, an epigenetic mechanism that results in differential allelic expression based on the parent-of-origin. Interploidy and interspecies crosses affect imprinted gene expression in the endosperm, suggesting an epigenetic basis for endosperm-based hybridization barriers (Batista and Köhler, 2020).

In their new work, Kinser et al. (2021) investigate the impact of genome hybridization in Mimulus (monkeyflowers) endosperm. Mimulus offers a wide array of phenotypic, ecological, and genomic diversity and has emerged as a model system for studies of evolutionary functional genomics. In addition, endosperm development in Mimulus is considered to reflect the ancestral state (Flores-Vergara et al., 2020).

Kinser and colleagues first characterized seeds resulting from reciprocal crosses between closely related Mimulus guttatus (diploid) and Micrococcus luteus (allotetraploid with A and B subgenomes). The team found that seeds were reduced in size with late endosperm growth defects when M. guttatus was the seed parent. In the reciprocal cross, however, early endosperm development arrest resulted in embryo abortion and failure of the seed set (see Figure).

Figure.

Reciprocal crosses between closely related Mimulus species, diploid M. guttatus (2X) and allotetraploid M. luteus (4X), result in distinct seed phenotypes. Both crosses lead to reduction in seed size, but severe endosperm defects lead to embryo arrest and unviable seeds when M. guttatus was the seed parent (2X × 4X cross). Top, scanning electron micrographs of hybrid seeds. Bottom, histological sections of seeds at 11 days after pollination, showing the seed coat (orange), endosperm (blue), and embryo (green). Adapted from Kinser et al. (2021), Figure 2.

The team then performed RNA-seq of hybrid embryos and endosperm to identify imprinted genes. They found that similar to other angiosperms, imprinting essentially takes place in the endosperm. Both species had more paternally expressed than maternally expressed genes (PEGs and MEGs, respectively), with 37 PEGs and 16 MEGs in diploid M. guttatus and 270 PEGs and 6 MEGs in tetraploid M. luteus. Remarkably, irrespective of the orientation of the cross, imprinted expression overall was lost in hybrid endosperm and the M. luteus genome dominated the expression landscape. Kinser and colleagues were further able to extrapolate the contributions of M. luteus A and B subgenomes by following the expression of a subset of genes that could be unambiguously attributed to a subgenome and found that each subgenome contributed equally.

DNA methylation underlies genomic imprinting in the endosperm (Batista and Köhler, 2020). In plants, cytosine can be methylated in CG contexts, but also in non-CG contexts, CHG, and CHH, where H is any base but C. The team first assessed genomic DNA methylation patterns at genes and transposable elements (TEs) in M. luteus and M. guttatus endosperm. After normalization, patterns of methylation were relatively similar overall, with quantitative differences mostly in non-CG contexts. Kinser et al. then performed reciprocal crosses and assessed DNA methylation levels in hybrid endosperms. Strikingly, methylation patterns remained similar in CG but significantly changed in non-CG contexts. In the CHH context particularly, methylation levels increased in both M. luteus and M. guttatus genomes when M. guttatus was the seed parent (2X × 4X). However, the reciprocal cross (4X × 2X), which leads to complete seed abortion, resulted in a drastic loss of CHH methylation at both genes and TEs in M. luteus and M. guttatus genomes.

A recent study in Arabidopsis reported that 2X × 4X crosses resulted in loss of CHH methylation in the endosperm. Loss of RNA-dependent DNA methylation (RdDM) pathway, which in part controls CHH methylation, was further found to rescue 2X × 4X seed viability (Satyaki and Gehring, 2019). In their work, Kinser et al. identify the 4X × 2X cross as lethal in Mimulus, and found that this is accompanied by a loss of CHH methylation in the endosperm. Because the crosses were interspecies as well, it is difficult to compare these results with Arabidopsis. However, this clearly points to a role for CHH methylation disruption, and potentially RdDM, as an endosperm-based hybridization barrier mechanism in angiosperms. Thus, the work by Kinser et al. provides a framework to further understand the components of hybridization barriers and their evolution in angiosperms.