A Case of Gene Fragmentation in Plant Mitochondria Fixed by the Selection of a Compensatory Restorer of Fertility-Like PPR Gene

Abstract

The high mutational load of mitochondrial genomes combined with their uniparental inheritance and high polyploidy favors the maintenance of deleterious mutations within populations. How cells compose and adapt to the accumulation of disadvantageous mitochondrial alleles remains unclear. Most harmful changes are likely corrected by purifying selection, however, the intimate collaboration between mitochondria- and nuclear-encoded gene products offers theoretical potential for compensatory adaptive changes. In plants, cytoplasmic male sterilities are known examples of nucleo-mitochondrial coadaptation situations in which nuclear-encoded restorer of fertility (Rf) genes evolve to counteract the effect of mitochondria-encoded cytoplasmic male sterility (CMS) genes and restore fertility. Most cloned Rfs belong to a small monophyletic group, comprising 26 pentatricopeptide repeat genes in Arabidopsis, called Rf-like (RFL). In this analysis, we explored the functional diversity of RFL genes in Arabidopsis and found that the RFL8 gene is not related to CMS suppression but essential for plant embryo development. In vitro-rescued rfl8 plantlets are deficient in the production of the mitochondrial heme-lyase complex. A complete ensemble of molecular and genetic analyses allowed us to demonstrate that the RFL8 gene has been selected to permit the translation of the mitochondrial ccmFN2 gene encoding a heme-lyase complex subunit which derives from the split of the ccmFN gene, specifically in Brassicaceae plants. This study represents thus a clear case of nuclear compensation to a lineage-specific mitochondrial genomic rearrangement in plants and demonstrates that RFL genes can be selected in response to other mitochondrial deviancies than CMS suppression.

Keywords: Rf-like PPR proteins; c-type cytochrome maturation; mitochondria; mitochondrial translation; plant respiratory mutant.

Fig. 1.

The Arabidopsis rfl8 mutant is embryonic lethal but can be rescued by in vitro culture of immature embryos. (A) Schematic representation of the RFL8 PPR protein and its 13 P-type PPR repeats (shown as dark gray arrows). The position of the SALK_0015489 T-DNA insertion with respect to the PPR motif organization of RFL8 is indicated by an inverted black triangle. (B) Arabidopsis seeds in siliques of rfl8 heterozygous plants showing discolored seeds (red arrows) containing homozygous rfl8 mutant embryos. (C) Nomarski photograph of an rfl8 mutant embryo arrested at the bent cotyledon stage. (D) Wild-type Col-0 Arabidopsis plant grown for 5 weeks on embryo-rescue culture medium after germination. (E) rfl8 mutant plantlet grown for 22 weeks after germination on embryo-rescue culture medium after germination.

Fig. 2.

rfl8 plants contain dramatically reduced levels of complexes III, IV, and cytochrome c/c1 heme lyase. (A) Immunoblots of BN–PAGE gels probed with antibodies against the Rieske iron–sulfur (RISP, a subunit of complex III) and Cox2 (a subunit of complex IV) proteins. About 100 µg of crude mitochondrial extracts prepared from wild-type (Col-0), rfl8, and complemented rfl8 plants (Cpl) were used in the analysis. * shows bands corresponding to the indicated respiratory complexes. (B) Steady-state level analysis of various mitochondrial proteins in wild-type, rfl8, and functionally complemented rfl8 plants. About 12.5, 25, or 50 µg of proteins from crude mitochondrial preparations were loaded in each lane and probed with antibodies specific to the indicated mitochondrial proteins (right). PORIN was used as loading control to verify equal loading across samples. Molecular weight (MW) of detected proteins is indicated (left). (C) Same analysis as shown in (A) except that antibodies to CCMA and CCMH were used to detect the heme-handling and heme–lyase complexes, respectively.

Fig. 3.

Translation of the mitochondrial ccmF_N2 transcript is reduced in rfl8 plants. (A) Ribo-Seq analysis of mitochondrial mRNAs in rfl8 plants compared with the wild type. The bars depict log₂ ratios of ribosome footprint abundance for each mitochondrial mRNA normalized to both mRNA length and abundance in rfl8 plants relative to the wild type (Col-0). Values between genotypes were also normalized to the numbers of Ribo-Seq reads mapping to mitochondrial ORFs. The reported values are means of three independent biological replicates (error bars indicate SD). (B) Screenshots from the Integrated Genome Viewer software showing the distributions of ribosome footprints along the ccmF_N2 transcript in both wild-type (Col-0) and the rfl8 mutant. The distributions were normalized to the number of reads mapping to the mitochondrial genome. The red arrow points to a ribosome footprint peak near the proposed translation start site of ccmF_N2 that is visible in the wild type (Col-0) but undetectable in the rfl8 mutant.

Fig. 4.

The Brassica rapa RFL8 homolog can functionally complement the Arabidopsis rfl8 mutant. (A) Along with ccmF_N2, Brassicaceae plants all encode a very close homolog to the RFL8 protein. Phylogenetic relationship between the closest RFL8 homologs identified in a representative panel of Brassicaceae (orange) and non-Brassicaceae (black) angiosperm plants (see details below). The Arabidopsis PPR protein PPR336 was chosen as outgroup. (B) Brassicaceae closest RFL8 homologs are more closely related to RFL8 than to other Arabidopsis Rf-like (RFL) proteins. Phylogenetic relationship between Brassicaceae RFL8 homologs and all other Arabidopsis thaliana RFL proteins. The red arrow points to the Arabidopsis RFL8 protein and the RFL8 clade is shown in orange. Sequence alignments were done with T-coffee and the tree constructed with iTOL. (C) Sequence alignment of a part of ccmF_N2 5′ region from Arabidopsis thaliana and Brassica rapa. The proposed ccmF_N2 translational start codon (GTG) is boxed in red and the predicted RFL8 binding site is underlined in red. A multiple sequence alignment of 1 kb of DNA sequence upstream of the GTG codon from a representative panel of Brassicaceae plants is shown in supplementary figure S10, Supplementary Material online. (D) Comparative growth of a wild-type (Col-0) plant and a homozygous Arabidopsis rfl8 mutant expressing the B. rapa RFL8 homolog, 8 weeks of culture after sowing. (E) Prediction of the Arabidopsis and Brassica rapa RFL8 RNA binding sites. The amino acids at positions 5 and 35 of each RFL8 PPR repeat are shown from N- to C-terminus. The obtained amino acid combinations were used to calculate the probabilities of nucleotide recognition by each individual PPR repeat according to the PPR code (Barkan et al. 2012) and the most likely target sequence identified in the mitochondrial genome of Arabidopsis and B. rapa is indicated. This sequence is found in the 5′ region of ccmF_N2 (see panel C). Ath, Arabidopsis thaliana; Osa, Oryza sativa; Dca, Daucus carota; Csa, Cucumis sativus; Bvu, Beta vulgaris; Vvi, Vitis vinifera; Nta, Nicotiana tabacum; Gma, Glycine max; Adu, Arachis duranensis; Tha, Tarenaya hassleriana; Cru, Capsela rubella; Aly, Arabidopsis lyrata; Esa, Eutrema salsugineum; Rsa, Raphanus sativus; Bol, Brassica oleraceae; Bra, Brassica rapa.

Fig. 5.

The RFL8 protein associate with the ccmF_N2 mRNA in vivo. (A) RIP-Seq assay on mitochondrial extracts prepared from complemented rfl8::RFL8-3HA and wild-type plants. Coimmunoprecipitated RNA was used for cDNA synthesis and then analyzed by Illumina deep sequencing. Obtained reads were mapped to the Arabidopsis mitochondrial genome. The reported values are ratios of read counts per mitochondrial ORF between rfl8::RFL8-3HA and wild-type plants (RFL8-3HA/Col-0 ratios). They are means of two independent biological replicates performed on each genotype. Differential enrichment performed with edgeR identified ccmF_N2 with a false discovery rate control at 5%. (B) Similar RIP-Seq experiment as performed in (A) except that initial mitochondrial extracts were predigested with RNase-I prior to immunoprecipitation. Sequencing reads were mapped to the mitochondrial genome and screenshots from the Integrated Genome Viewer software showing read distributions along the ccmF_N2 coding sequence and 5′-UTR in wild type (Col-0) and complemented rfl8::RFL8HA plants are depicted. The diagram below the plots materializes the proposed position of the ccmF_N2 ORF (Rayapuram et al. 2008) and its 5′-UTR as well as that of the three RNA probes (a, b, and c) used in gel mobility shift assay used in panel (C). The location and sequence of the RFL8 binding site predicted by the PPR code are also indicated. (C) Gel mobility shift assays confirming the ability of RFL8 to associate with the RNA sequence predicted the PPR code. 0, 100, 200, or 400 nM of recombinant RFL8 were assayed in the shown lanes combined with gel-purified RNA probes (a, b, and c) indicated in panel (B). U, unbound probe; B, bound probe.