Mitochondrial genes leave home

Mitochondria contain a small yet essential set of genes retained from their endosymbiotic α-proteobacterial ancestor, even though most genes from the respiring invader were either transferred to the host genome or lost very early in eukaryotic evolution. It has now become increasingly clear that successful migration of mitochondrial genes to the nucleus is still going on (at least in lineages which have the standard genetic code in the mitochondrion) and stages in this evolutionary process are depicted in Fig. 1 (see also reviews by Brennicke et al., 1993; Adams & Palmer, 2003; Timmis et al., 2004). In plants, such movement and subsequent loss of the native mitochondrial copy occur remarkably often, albeit sporadically, among various lineages. Transfer rates in the most rampant cases are even estimated to rival synonymous nucleotide substitution rates (Adams et al., 2002). The work now reported by Keith Adam's group in this issue of New Phytologist (Choi et al., pp. 429–439) provides us with further insight into such mitochondrion-to-nucleus gene transfers. Evolutionary scenario for mitochondrion-to-nucleus gene transfer in plants. The gene of interest (black bar) escapes from the mitochondrion (e.g. during organelle lysis or division) and the cDNA (edited and spliced) form is integrated into the nuclear genome (short grey bar preceding black bar denotes acquisition of nuclear expression signals, although there is not necessarily an amino-terminal presequence as mitochondrial targeting signal). The transition stage ends when (i) the now-redundant mitochondrial gene degenerates into a pseudogene and is lost; (ii) the nuclear-located gene copy degenerates and is lost; or (iii) both functional copies are permanently retained because of selective advantage (e.g. specialized function or organ-specific expression). The broken grey arrow indicates that scenario (iii) has not as yet been documented in plants. Note that, although it is not shown in the schematic, each plant cell has many mitochondria, each of which contains many copies of the mitochondrial genome. mt, mitochondrion; cp, chloroplast; nuc, nucleus. Their approach was to search the recently sequenced poplar nuclear genome (which has a size about four times that of Arabidopsis) as well as EST databanks, looking for functional copies of the set of 40 protein-coding genes that are collectively found among the mitochondrial genomes of flowering plants. Choi et al. uncovered eight such actively expressed genes; two encode succinate dehydrogenase subunits (sdh3, sdh4) and six specify ribosomal proteins (rps2, rps10, rps11, rps14, rps19 and 3′rpl2). Transfer times were deduced to range from very early in eudicot evolution (in the case of three ribosomal protein genes) to quite recently, based on their discovery of very similar sdh4 copies in both the mitochondrion and nucleus in poplar. '… evidence points to a constant bombardment of the nucleus by mitochondrial DNA/RNA (perhaps escaping during organelle division or lysis)' When one considers the challenges faced by a mitochondrial gene migrating to the nucleus, they seem daunting if not insurmountable. It must acquire appropriate regulatory elements to be transcribed, the RNA properly processed and then correctly translated in the cytoplasm. The protein must have suitable signals to be routed back to (and imported into) the mitochondrion. All the while, it must fend off inactivating mutations if it is to dethrone the native mitochondrial-located copy. Moreover, in plants, the relocated gene must be a cDNA copy which has the requisite RNA editing codon 'corrections' and group II introns removed. Given all this, the frequency of successful transfers in plants seems astonishing; however, in its favour, evidence points to a constant bombardment of the nucleus by mitochondrial DNA/RNA (perhaps escaping during organelle division or lysis) (reviewed by Timmis et al., 2004; Leister, 2005). Evidence also points to a period of accelerated evolution (exceeding the known higher rate of nucleotide substitution in plant nuclear vs mitochondrial genes) during adaptation to its new milieu. How might proper expression signals be acquired by the itinerant gene? The simplest scenario would seem to be its lucky integration into a transcriptionally active nuclear location which can provide ready-made expression/protein-targeting signals, either via direct fusion or through intron-mediated exon shuffling. Such signals might be derived from duplicate copies of other pre-existing nuclear genes which encode mitochondrial proteins. Indeed, the poplar sdh3 gene was identified by Choi et al. to possess hsp70-derived (mitochondrial heat shock protein) amino-terminal sequences. Certain other transferred genes in poplar (namely rps14 and rps2), however, lack coding extensions, so signals are presumably intrinsic to the native proteins. Indeed, it appears that about 25% of the bacterial-origin mitochondrial-type ribosomal proteins encoded by Arabidopsis and rice nuclear genes lack acquired amino-terminal extensions (Bonen & Calixte, 2006). Another clever strategy for expression would be to exploit alternative splicing by hitchhiking within the intron of a pre-existing gene, and in fact this is the case for rps14 (located within an sdh2 intron) in cereals (reviewed in Adams & Palmer, 2003). Adaptation to improve expression in the new environment is also expected to include the acquisition of spliceosomal introns and a shift in codon usage patterns. A further intriguing finding by Choi et al. is the apparently independent recruitment of the same amino-terminal RNA binding (RRM) domain by rps19 following separate transfers in the Arabidopsis and Populus lineages, and a parallel situation whereby sdh3 independently acquired hsp70-derived expression signals in these two lineages. At first glance, it seems rather like lightning striking twice, but Choi et al. present strong arguments in its favour. Consequently, one wonders if there might turn out to be a relatively small set of 'usual suspects' (such as hsp70) that predominate as partners, either because they confer a strong advantage or because of their relative availability. Choi et al. note that several recently transferred genes are near other mitochondrial-type genes, including ones encoding PPR-type pentatricopeptide proteins (which comprise a huge multigene family implicated in organelle RNA metabolism) as well as hsp70. The clustering of genes encoding certain mitochondrial proteins has also been observed in the Arabidopsis nuclear genome (Elo et al., 2003). One provocative possibility is that external stressors from time to time trigger localized nuclear genomic amplifications which create 'safe havens' for migrant mitochondrial genes that are assaulting the nucleus (after their release from stress-induced broken organelles and conversion to cDNA by reverse transcriptase encoded by nuclear retrotransposons or mitochondrial group II introns). This would result in the punctuated and variegated pattern of mitochondrion-to-nucleus gene flow seen among plant lineages. No doubt further insight will emerge as additional nuclear genomic data become available. In such analyses, however, there may be confounding factors that obscure tracing the history (in addition to the inevitable attrition of sequence similarity over time). For example, post-transfer shuffling of fused upstream coding sequences (as well as gene duplication) could occur, as has been documented for rps11 in rice (Kadowaki et al., 1996). Also gene conversion between recruited and donor copies of targeting presequences (in independent lineages after an earlier common mitochondrial gene transfer) would generate high sequence similarity, which might be falsely interpreted as reflecting separate recent events. As seen in the present study, the 5′coding extensions acquired by transferred genes can be much longer than the 15–50 amino acid length expected of typical mitochondrial targeting signals, and this raises the possibility that multifunctional proteins are being created. Likewise, because the 5′-truncated second copy of sdh3 in poplar was observed by Choi et al. to be expressed, it might be recruited for a new (nonmitochondrial) function. Incidentally, the RRM domain fused to rps19 was initially speculated to substitute for rps13, which is absent from both the nuclear and mitochondrial genomes in Arabidopsis (Sanchez et al., 1996); however, it was later established that a duplicated, divergent copy of chloroplast rps13 performs that function (reviewed in Adams & Palmer, 2003), so the role of the RRM domain remains unknown. It is also worth noting that ribosomal protein sequences can show considerable plasticity. For example, the 3′ half of rpl2 was relocated to the nucleus early in eudicot evolution, whereas the 5′ part has been retained in the mitochondrion in many lineages, so that two (smaller) proteins presumably reconstitute the structure needed in the ribosome (Adams et al., 2001). A particularly interesting step in the evolutionary process of intracellular gene transfer is the intermediate stage when there are functional copies in both compartments (before the loss of one or the other), and Choi et al. have caught a rare example of such a gene-in-transition. They observed active sdh4 genes in both the nucleus and mitochondrion in Populus (and in the sister genus Salix). Both copies showed RNA-level expression in leaf, petiole, root, stem and flower tissue. The two SDH4 predicted proteins differ at seven amino acid sites (out of a total of 109) and it will be of interest to see if both forms are present in the respiratory chain complex. The authors caution that the presence of transcribed and edited mitochondrial sequences might be deceptive in that a gene-in-decline may be just 'taken along for the ride' if physically near an active gene (and the poplar cox3-sdh4 linkage is very close), and thus protected from erosion caused by the recombinogenic nature of plant mitochondrial genomes. The only other examples to date are for cox2 in legumes (Adams et al., 1999) and rpl5 in wheat (Sandoval et al., 2004). In evolutionary terms, the transition stage formally ends when one copy 'wins out' and the other degenerates into a pseudogene and is lost. Indeed, there are now many well-documented cases of recent successful transfers to the nucleus in plants. On the other hand, cases of a mitochondrial copy 'winning out' are more difficult to detect (the outcome being indistinguishable from the ancestral state) and have only rarely been reported, one example being rps19 in rice (Fallahi et al., 2005). It will be interesting to learn how long a transition stage might persist, and indeed whether there might be a selective advantage to retaining both copies, as exemplified by the two distinctive atp9 genes in filamentous fungi where the normally 'silent' mitochondrial copy is expressed in germinating conidia (Bittner-Eddy et al., 1994). Plants provide a unique opportunity to examine mitochondrion-to-nucleus gene transfers that have occurred very recently during evolution, and to assess the biological impact of such genetic fluidity and cross-talk. The application of such knowledge in biotechnological endeavours can be envisioned (cf. Manfredi et al., 2002), and there is no shortage of intriguing evolutionary questions to ponder. Sequence data now emerging from plant nuclear genome projects, as seen here for poplar, should greatly assist in this quest.

Récupéré en direct depuis OpenAlex et désinversé. Les résumés ne sont pas conservés dans cette base de données : les index inversés représentent 8,6 Go des 9,3 Go de texte de la base, et le serveur dispose de 13 Go libres.