Nurse cell­–derived small RNAs define paternal epigenetic inheritance in Arabidopsis

INTRODUCTION

Epigenetic modifications to chromatin, such as the methylation of DNA cytosines, carry regulatory information across the generations in plants and animals. Epigenetic marks are inherited through germlines, which can undergo epigenetic reprogramming that affects germline function. Mammalian germlines undergo genome-wide demethylation and remethylation, which resets epigenetic marks and restores cellular pluripotency. Germlines of flowering plants differentiate later than those of mammals and undergo distinct DNA methylation reprogramming. Global methylation level changes have been observed in plant germ cells, as well as local de novo methylation that regulates germline gene expression. However, the reprogramming mechanism remains unclear, which limits the understanding of tissue-specific and transgenerational epigenetic inheritance in plants.

RATIONALE

De novo methylation of genes in the Arabidopsis male germline is catalyzed by the small RNA-directed DNA methylation pathway (RdDM). RdDM uses 24-nucleotide (nt) small interfering RNAs (siRNAs) transcribed from a source locus, typically a transposon, to guide methyltransferases to methylate the cognate region. RdDM is ubiquitous among cells and normally methylates transposons. We investigated how RdDM targets genes in the germline, how such reprogramming is limited to the germline, and how the hundreds of target genes are selected.

RESULTS

We show that in Arabidopsis, germline-specific methylation at hundreds of genes is established by 24-nt siRNAs transcribed from hypermethylated transposons with imperfect sequence homology. CRISPR deletion of siRNA source transposons abolishes DNA methylation at corresponding genes in male meiocytes, demonstrating the ability of these siRNAs to target methylation in trans with mismatches. Furthermore, we found that meiocytes are quiescent in 24-nt siRNA biogenesis, and the gene targeting siRNAs are produced in tapetal nurse cells and transported into meiocytes. The siRNA profile of tapetal cells resembles that of meiocytes, suggesting bulk siRNA movement through the cytoplasmic channels (known as plasmodesmata) that connect these cells during early meiosis. Although gene-targeting siRNAs accumulate to levels in the tapetum comparable with those in meiocytes, the target genes are not methylated in the tapetum, showing that meiocytes can use imperfectly matching siRNAs to target DNA methylation.

We also show that the biogenesis of gene-targeting siRNAs in the tapetum is driven by a putative chromatin remodeler, CLASSY3 (CLSY3). CLSY3 is absent in vegetative tissues but enriched in female and male organs. In male organs, CLSY3 is specifically expressed in tapetal cells. The clsy3 mutation eliminates de novo methylation at genes in male meiocytes as well as sperm, revealing the role of tapetal siRNAs in driving methylation reprogramming throughout the male germline. This conclusion was reinforced with experiments with a genetic mosaic system that restricted siRNA biogenesis to the tapetum. In the mosaic lines, tapetal siRNAs restore methylation in the male germline at genes specifically methylated in the germline and at transposons targeted by RdDM in all tissues. This shows that tapetal siRNAs shape the full spectrum of RNA-directed DNA methylation throughout the male germline, from meiocytes to sperm.

As tapetal siRNAs induce methylation at transposons in the male germline, we investigated whether these siRNAs serve to silence germline transposons in addition to their gene regulatory function. We found that tapetal siRNAs methylate and prevent the expression of a Gypsy retrotransposon (GP1) that is specifically active in reproductive cells, demonstrating that tapetal siRNAs ensure germline genome integrity by suppressing transposons.

CONCLUSION

Our work reveals the role of tapetum-derived 24-nt siRNAs in determining the paternal methylome. Tapetal siRNAs drive de novo methylation at genes in the germline and safeguard genome integrity by silencing transposons. The ability of meiocytes to use mismatched 24-nt siRNAs to induce methylation allows RdDM to target sequences similar to those of transposons that produce siRNAs in the tapetum. This should facilitate recognition and silencing of new transposons, indicating that meiosis is a key stage for genome surveillance and allows RdDM to specifically regulate germline gene expression. The broad targeting competence of tapetal siRNAs resembles that of metazoan gonad-expressed Piwi-interacting RNAs (piRNAs), suggesting convergent evolution of specialized small RNA pathways to control transposons and genes in plant and animal germlines.