The genomes of cytoplasmic organelles (mitochondria and plastids) are maternally inherited in most eukaryotes, thus excluding organellar genomes from the benefits of sexual reproduction and recombination. The mechanisms underlying maternal inheritance are largely unknown. Here we demonstrate that two independently acting mechanisms ensure maternal inheritance of the plastid (chloroplast) genome. Conducting large-scale genetic screens for paternal plastid transmission, we discovered that mild chilling stress during male gametogenesis leads to increased entry of paternal plastids into sperm cells and strongly increased paternal plastid transmission. We further show that the inheritance of paternal plastid genomes is controlled by the activity of a genome-degrading exonuclease during pollen maturation. Our data reveal that (1) maternal inheritance breaks down under specific environmental conditions, (2) an organelle exclusion mechanism and a genome degradation mechanism act in concert to prevent paternal transmission of plastid genes and (3) plastid inheritance is determined by complex gene–environment interactions.

Pentatricopeptide repeat (PPR) proteins constitute the largest family of proteins in angiosperms, and most members are predicted to play roles in the maturation of organellar RNAs. Here we describe the novel mitochondrial editing factor 31 (MEF31), an E-PPR protein involved in editing at two close sites in the same transcript encoding subunit C of the twin-arginine translocation (tat) pathway. MEF31 is essential for editing at site tatC-581 and application of the recently proposed amino acid code for RNA recognition by PPR proteins supports the view that MEF31 directly targets this site by recognizing its cis sequence. In contrast, editing at site tatC-586 five nucleotides downstream is only partially affected in plants lacking MEF31, being restored to wild-type levels in complemented plants. Application of the amino acid code and analysis of individual RNA molecules for editing at sites 581 and 586 suggest that MEF31 does not directly target site tatC-586, and only indirectly influences editing at this site. It is likely that editing at site tatC-581 improves recognition of the site tatC-586 cis sequence by a second unknown PPR protein.