Gene transfer: anything goes in plant mitochondria

Parasitic plants and their hosts have proven remarkably adept at exchanging fragments of mitochondrial DNA. Two recent studies provide important mechanistic insights into the pattern, process and consequences of horizontal gene transfer, demonstrating that genes can be transferred in large chunks and that gene conversion between foreign and native genes leads to intragenic mosaicism. A model involving duplicative horizontal gene transfer and differential gene conversion is proposed as a hitherto unrecognized source of genetic diversity. See research article: http://www.biomedcentral.com/1741-7007/8/150

. The parasitic plant Cuscuta wrapping around one of its many possible hosts, Arabidopsis. Image kindly provided by Dr Collin Purrington, Swarthmore College. mitochondria possess active DNA uptake systems and are capable of fusion; chloroplasts do not and are not (see [3] and references therein). Thus, given direct physical contact between host and parasite tissue, ample oppor tunity for mtDNA uptake and exchange would seem to exist. Yet despite extensive phylogenetic evidence support ing the notion that plant mitochondrial HGT is rampant, numerous mechanistic uncertainties remain. These include the question of whether DNA or RNA is the donor molecule and whether a virus or some other vector mediates the transfer.

Gene transfer, gene conversion, and intragenic mosaicism
Mower et al. [4] sought to elucidate the process of plant toplant HGT by using PCR to survey mitochondrial genes in species of the parasitic plant Cuscuta ( Figure 1) and one of its many hosts, the dicot weed Plantago. After initially casting a wide net to capture ten protein and RNA genes from both Cuscuta gronovii and Plantago coronopus mtDNA, they sequenced three genes, atp1, atp6 and matR, from a range of host and parasite relatives and showed that each gene appears to have been transferred recently (within the last few million years) from the mitochondrial genome of Cuscuta to that of Plantago. In and of itself this is no longer surprising, but the authors demonstrate that (i) the three genes appear to have been transferred together in the context of a relatively large fragment of DNA (and not RNA, which can be inferred due to the presence of unedited cytidine residues at sites known to undergo RNA editing); (ii) both native and 'foreign' homologs (xenologs), the latter in the form of pseudogenes, coexist in several species; and (iii) multiple gene conversion events have occurred between coresident loci.
Hao et al. [5] have gone even further. They uncovered a 'gorgeous mosaic' of multiple mitochondrial genes and gene fragments in various plant hostparasite lineages, including a striking chloroplasttomitochondrion transfer involving a region of atp1. In total, approximately one third of the HGT events investigated (5 of 17 genes) appear to have involved at least some gene conversionthe nonreciprocal exchange of DNA between homolo gous sequences -suggesting that this process plays an important role in the integration of foreign genetic material. The true significance of HGTassociated gene conversion may in fact be underestimated because of differential gene loss, lack of phylogenetic resolution and insufficiently sensitive detection methods. The authors propose a model -duplicative HGT and differential gene conversion (DHDC) -in which intra and inter organellar gene transfer and recombination are creative forces in the generation of mitochondrial genetic diversity.
By duplicative HGT, Hao et al. mean the transfer and integration of a foreign set of genes, a single gene, or a gene fragment from donor to recipient mtDNA that does not instantly replace its endogenous counterpart. Herein lies the key to the model, as the coexistence, however transient, of foreign and native loci within the same subcellular compartment allows gene conversion to occur. Gene conversion is a wellunderstood process (for example, as a generator of allelic diversity during meiosis), and in the context of DHDC, gene conversion, occurring in either a continuous or discontinuous manner, gives rise to patchwork recombinants. Such heterogeneity is not phenotypically 'silent': the recombinant atp1 and matR genes uncovered by Hao et al. yield proteins with different amino acid sequences [5].

DH-DC in plant mtDNA: impact and implications
The results of Mower et al. [4] and Hao et al. [5] are both troubling and satisfying. Troubling because recombi na tion between resident and foreign gene copies, no matter how transient the latter, has the potential to 'wreak havoc' on the results of even the most cautious phylogeneticist intent on inferring organismal phylogenies from plant mitochondrial genes [5]. In addition, mutation rate estimates for plant mtDNA -painstakingly obtained and long considered to be exceptionally low [3,11] -should, in the face of the DHDC model, now be considered over estimates: a certain fraction of singlenucleotide differ ences observed between sequences will be due to gene conversion, not point mutation. Ultimately, though, satis faction comes from a deeper understanding and appre cia tion of the true complexity of plant mtDNA evolution.
More generally, Hao et al. [5] raise the intriguing possibility that DHDC could be a driver of gene evolu tion in prokaryotes, one that has thus far gone un detected. And what of nuclear genomes? In concluding their study of PlantagoCuscuta HGT, Mower et al. [4] state that '…unravelling this history will probably require sequencing multiple mitochondrial and nuclear genomes from Plantago. ' Given the pace at which sequencing technologies are evolving, it is hard to imagine this not happening in the near future, not just within plants but also for microbes and multicellular organisms across the full breadth of eukaryotic diversity. With the exception of specific lineages such as fungi, whose nuclear genomes are being sequenced at both shallow and deep evolu tionary divergences, the field of comparative genomics is still in 'gap filling' mode. The Hao et al. [5] and Mower et al. [4] studies underscore the fact that when it comes to HGT the devil is in the details: only in the context of meticulous comparisons of both closely and distantly related genomes is a deep understanding of the pattern, process and full scope of eukaryotic HGT likely to emerge.