When different genetic compartments exist in close proximity to one another there is the potential for intercompartmental DNA transfer. Land plants and eukaryotic algae exemplify this point. Within each of their cells there are three distinct genomes: nuclear, mitochondrial, and plastid, which are made up of nucDNA, mtDNA, and ptDNA, respectively. The sequencing of these different DNAs has uncovered staggering amounts of intracellular gene transfer (Timmis et al., 2004; Kleine et al., 2009). For example, the nuclear genomes of many seed plants, including rice and corn, contain thousands of nucleotides of recent mitochondrial and plastid origin, most of which are nonfunctional (Hazkani-Covo et al., 2010; Smith et al., 2011). Similarly, seed plant mitochondrial genomes have accumulated large amounts of DNA from the neighboring plastid and nuclear genomes (Wang et al., 2007; Smith, 2011). More than 5% of the cumber (Cucumis sativus) and zucchini (Cucurbita pepo) mitochondrial genomes, for instance, are represented by plastid- and nuclear-derived sequences (Alverson et al., 2010, 2011).
Given the prevalence of intracellular DNA migration into the nuclear and mitochondrial genomes of seed plants, one might expect their plastid genomes to also harbor large amounts of horizontally acquired DNA. But the opposite is true. Seed plant plastid genomes, and ptDNAs as a whole, are surprisingly devoid of either mitochondrial- or nuclear-derived DNA (Richardson & Palmer, 2007; Smith, 2011). This observation has led to the assumption that the intracellular transfer of mtDNA or nucDNA to plastids does not occur – a theory backed by the observed lack of an efficient DNA uptake system within plastids (Bock, 2010), but see Cerutti & Jagendorf (1995). Moreover, plastids, unlike mitochondria, do not regularly undergo fusion event, thus, limiting the potential for foreign DNA to sneak in. Recent data from the organelle genomes of carrots and milkweeds, however, are changing this view, proving that mitochondrion-to-chloroplast genetic transfer does occur and that plastids are not as impenetrable to foreign DNA as previously thought.
Clues from the carrot chloroplast
Three years after it was sequenced, the plastid genome of carrot, Daucus carota, was unexpectedly found (during a cursory blast analysis) to have a c. 1.5 kb region of potential mitochondrial origin (Goremykin et al., 2009). This region, which is dubbed the D. carota Mitochondrial-like Plastid (DcMP) sequence and located in the rps12-trnV intergenic spacer, is similar to conserved coding and noncoding segments of seed plant mtDNA, including cox1, suggesting it arose through mtDNA transfer to the plastid (Fig. 1). Support for this hypothesis came from the sequencing of the carrot mitochondrial genome and the discovery that it too contains a complete, but fragmented, DcMP sequence (Iorizzo et al., 2012a). Further analyses revealed that diverse members of the carrot family (Apiaceae) contain a large piece of DcMP within their mtDNA, but that only Daucus species and their close relative cumin (Cuminum cyminum) have a plastid DcMP. Together, these observations were taken as proof that a mitochondrial-located DcMP was present in the ancestor of the Apiaceae and subsequently spread to the plastid genome in the common ancestor of Daucus and Cuminum (Iorizzo et al., 2012b) (Fig. 1).
Complicating the story is that DcMP, apart from having a short cox1-like segment, shows little similarity to other sequences in GenBank, making it difficult to trace its origin within the Apiaceae, and leaving open the possibility that it began in the nuclear compartment and was then transferred independently to the mitochondrial and plastid genomes (Iorizzo et al., 2012a). Moreover, within the carrot mitochondrial genome, DcMP is fragmented and scrambled into three pieces, two of which are 80 kb apart from one another (Fig. 1). Assuming that DcMP migrated from the mtDNA to the ptDNA as an intact element and in a single event, its fragmentation within the mitochondrial genome must have occurred post-migration, most likely through mtDNA recombination (Iorizzo et al., 2012a), which frequently occurs in seed plants (Alverson et al., 2011). DcMP has some characteristics resembling those of non-LTR retrotransposons, including an open reading frame for a putative reverse transcriptase (Iorizzo et al., 2012b), and it is speculated that DcMP retrotransposed from the mitochondrial to the plastid genome – a scenario that could explain the short direct repeats bookending its ptDNA sequence. Some non-LTR elements target tRNA-coding regions, which may be why DcMP is located upstream of trnV within the plastid genome. There is also reason to believe that DcMP is acting as a surrogate promoter for trnV, having replaced the original promoter upon insertion (Iorizzo et al., 2012b). If true, it would mean that a mitochondrial-derived sequence is playing a functional role within a plastid genome.
Milkweeds: mitochondrial proof in the ptDNA pudding
A separate example of a DNA transfer from a mitochondrial to a plastid genome has come from the common milkweed, Asclepias syriaca, which belongs to a well-studied and large family of flowering plants called the Apocynaceae. Similar to the situation for D. carota, the A. syriaca plastid genome sequence was published and available in GenBank for over a year (Straub et al., 2011) before it was discovered to have a 2.4 kb mtDNA-like insert, located in the rps2-rpoC2 intergenic spacer (Ku et al., 2013) (Fig. 1). But unlike the case for carrot, the putative transferred element contains an intact exon for a mitochondrial ribosomal protein (rpl2), leaving little doubt about its mtDNA origin (Straub et al., 2013). Assembly of the A. syriaca mitochondrial genome confirmed that the rpl2-containing segment of the plastid genome does indeed match to milkweed mtDNA, specifically to the second exon of rpl2 and its flanking noncoding sequence, as well as to a piece of mtDNA from a completely different region (Straub et al., 2013). The sequence identity between the mitochondrial and plastid versions of rpl2 is high (c. 92%), but the plastid copy has a nonsense mutation, resulting in a premature stop codon and implying that it is a pseudogene, which is consistent with its observed low levels of transcription relative to other plastid-located genes (Straub et al., 2013). Remarkably, the sequence bordering the rpl2 insert in the A. syriaca plastid genome, including a portion of rpoC2, is present in the milkweed mitochondrial genome immediately adjacent to the genuine rpl2, meaning that the fragment of mtDNA that migrated to the plastid (rpl2) sits beside a section of ptDNA that migrated to the mitochondrion (rpoC2), and vice versa (Fig. 1).
Using PCR and next-generation sequencing, Straub et al. (2013) uncovered an rpl2 insert in the ptDNAs from other members of the Apocynaceae, and ultimately traced its origin back to a single mitochondrion-to-plastid DNA transfer event in the common ancestor of the Asclepiadeae and Eustegia (Fig. 1). They also found within various apocynacean species a mitochondrial-located rpoC2, which, as it turns out, is more broadly distributed within the group than the rpl2 insert, suggesting that the transfer of rpoC2 from the plastid to the mitochondrion predates the mtDNA-to-ptDNA rpl2 migration. In fact, the close proximity of the mitochondrial rpl2 to a ptDNA-derived sequence might have facilitated its eventual integration into the plastid genome via repair of a double-stranded break by homologous recombination between mtDNA and ptDNA (Straub et al., 2013).
More examples on the way?
In light of these new data from carrots and milkweeds, it is safe to say that mitochondrion-to-plastid DNA transfer does occur (but the reverse process occurs much more frequently). It is still not known in either of these cases, however, how the mtDNA entered the plastid or how it integrated into the plastid genome, be it by retrotransposition, homologous recombination, or some other process. The complete plastid genome sequences of D. carota and A. syriaca were both available for some time before the proposed mtDNA insert was discovered, hinting that there are likely other published ptDNA sequences with undiscovered mitochondrial-derived regions.
Of the six possible directions of intracellular genetic transfer within a plastid-containing species only one – nuclear-to-plastid DNA migration – remains to be described. But if the study of organelle genetics has taught us anything, it is to expect the unexpected. Next-generation sequencing techniques have greatly increased the rate at which plastid, mitochondrial, and nuclear genomes are becoming available. Soon we will be inundated with unprecedented amounts of organelle genomic data, some of which will most certainly contain more examples of foreign DNA within plastid genomes.
D.R.S. is supported by a Discovery Grant from the Natural Sciences and Engineering Research Council (NSERC) of Canada.