Seeds are products of successful fertilization involving the merging of maternal and paternal gametes. Interspecific pollination can often produce viable seeds but, on germination, these unwanted offspring are generally rejected by an epigenetic barrier, mediated by the endosperm.
Endosperm is a nutritional tissue of seeds in angiosperms, supporting embryonic and seedling growth through the transfer of nutrients stored in its cells. During angiosperm double fertilization, sperm fertilizes both the central cell and egg cell, resulting in development of the endosperm tissue and embryo, respectively. Although the embryo can accommodate a paternal genome from a relatively distant species, the endosperm often displays defects when the paternal genome comes from a different species1,2. Reporting in this issue of Nature Plants, Lafon-Placette et al. uncovered evidence implicating paternally expressed imprinted genes (PEGs) as the driver of this reproductive block3.
Classical genetic approaches have shown that roles for the maternally and paternally derived genomes are different in the endosperm2. In intraspecific cross, bi-parental contributions to the offspring are balanced, which ensures the normal developmental transitions of the endosperm. However, interploidy or interspecific crosses often yield one of two opposite abnormalities, precocious or delayed developmental transitions (Fig. 1). As these developmental progressions in the endosperm do not match those in the embryo, due to parental dosage imbalance, the seeds eventually abort. Researchers have proposed that the ‘intrinsic factors’ causing this imbalance could be the effective ploidy level, termed ‘endosperm balance numbers’ (EBNs), of the parental species2,4. However, the molecular basis of EBNs remains enigmatic. Working in the genus Capsella, Lafon-Placette et al. uncovered a correlation between EBNs and the number of paternally expressed imprinted genes (PEGs), which is likely to be under the epigenetic control of transposable elements (TEs)3. They conclude that a tug-of-war between PEG dosages from the parental species modulates endosperm developmental transitions.
Genomic imprinting refers to the unequal gene expression of maternally or paternally derived alleles, which can be seen predominantly in the endosperm and occasionally in the embryo in many angiosperms. These parent-of-origin-specific gene expression patterns controlled by epigenetic mechanisms5 are frequently used to explain the opposing phenotypes seen in endosperm resulting from reciprocal interploidy or interspecific crosses2,6, suggesting that the dosage of parent-of-origin specific imprinted genes might control EBNs. To test this hypothesis, Lafon-Placette et al. examined three diploid Capsella species (2n = 2x = 16), the obligate outbreeder Capsella grandiflora (Cg), and the selfers Capsella rubella (Cr) and Capsella orientalis (Co). At first, the authors recapitulated a classical genetic approach to determine EBNs in these species by reciprocal crosses in all possible combinations. The timing of cellularization indicates endosperm developmental transition (Fig. 1a). Precocious cellularization occurs when the maternal species has a larger EBN, while delayed cellularization happens when the paternal species has a larger EBN. In these inter-specific crosses, endosperms from a maternal Cg species always exhibited precocious cellularization, while those from a maternal Co species always displayed delayed cellularization. As a result, they concluded that Cg has the largest EBN, followed by Cr and then Co.
The molecular basis for endosperm developmental transition has been well documented in Arabidopsis thaliana6. The key player for the timing of cellularization (Fig. 1) is a type-I MADS-box transcription factor, AGL62. This protein likely forms a heterodimer with the paternally expressed PHERES1 (PHE1)7. Although a direct mechanism interfering the timing of cellularization in interspecific crosses remains elusive, a possible scenario is that paternally controlled MADS-box transcription factors and unidentified transcription networks counteract the activity of POLYCOMB REPRESSIVE COMPLEX 2 (PRC2) that catalyses genome-wide histone H3 lysine tri-methylation. It should also be noted that some components of PRC2 often display maternal-specific imprinted gene expression patterns in many angiosperm species8. So, mechanistically, imprinted genes probably underlie the timing of endosperm cellularization.
Lafon-Placette et al. next searched for imprinted genes in each of the three Capsella species. The numbers of PEGs well correlated to their EBNs, that is, Cg has the most PEGs and Co has the least. In addition, as imprinted genes are largely diverged in these species, species-specific PEGs also exhibit species-specific CHH methylation, likely targeted by RNA-directed DNA methylation. Furthermore, the authors showed that the gain of species-specific CHH methylation is related to TE insertions in each genome. As such, the species-specific insertions of TEs likely drive the divergence of CHH methylation and consequent divergence of PEGs between species, contributing to the endosperm-mediated reproductive barrier.
In fact, these results have been foreshadowed by a series of prior observations. First, in plants, imprinted genes are not highly conserved, even in close relatives9, and imprinted genes are frequently associated with TE insertions and repeated sequences — indicating a link between TE-driven imprinting and species divergence. Moreover, the copy number of TEs is thought to be associated with the mode of reproduction of the host; specifically, obligate sexual organisms tend to have a large number of TEs, while obligate asexual species tend to have a small number of TEs10. While self-incompatibility (SI) is maintained in C. grandiflora, the ancestor of C. rubella was estimated to have lost SI around 100 thousand years ago, and C. orientalis lost SI almost a million years ago.
There may be more complex mechanisms underlying these results, nevertheless this study shows a clear correlation between the copy numbers of TEs, and the numbers of PEGs and EBNs in the endosperm. It also generates many more questions that remain to be tested, such as why copy numbers of maternally expressed genes are not increased in the C. grandiflora? What is the molecular mechanism that determines EBN power in the female side? Do similar tendencies exist in other plant species? Lafon-Placette et al. have presented an interesting correlation and opened a new avenue towards understanding the role of imprinted genes and TEs in speciation.
Gehring, M., Choi, Y. & Fischer, R. L. Plant Cell 16, 203–213 (2004).
Kinoshita, T. Genes Genet. Syst. 82, 177–186 (2007).
Lafon-Placette, C. et al. Nat. Plants https://doi.org/10.1038/s41477-018-0161-6 (2018).
Johnston, S. A., den Nijs, T. P., Peloquin, S. J. & Hanneman, R. E. Jr. Theor. Appl. Genet. 57, 5–9 (1980).
Satyaki, P. R. & Gehring, M. Crit. Rev. Biochem. Mol. Biol. 1–13 (2017).
Lafon-Placette, C. & Kohler, C. Mol. Ecol. 25, 2620–2629 (2016).
Roszak, P. & Kohler, C. Proc. Natl Acad. Sci. USA 108, 20826–20831 (2011).
Tonosaki, K. & Kinoshita, T. Front Plant Sci. 6, 144 (2015).
Hatorangan, M. R., Laenen, B., Steige, K. A., Slotte, T. & Kohler, C. Plant Cell 28, 1815–1827 (2016).
Wright, S. & Finnegan, D. Curr. Biol. 11, 296–299 (2001).
The author declares no competing interests.