Background

The mitochondrial proteome consists of at least 1000 different proteins. The genes encoding many of these proteins were initially encoded within the original respiring endosymbiont but have undergone intracellular transfer to the nucleus over evolutionary time, so that the proteins must be targeted back to the mitochondrion to perform their function. The number of retained mitochondrial protein-coding genes varies widely among eukaryotes, from 67 in the jakobid Reclinomonas americana [1] to only 3 in apicomplexans such as Plasmodium falciparum [2]. Genes retained in the mitochondrion encode proteins involved in fundamental mitochondrial processes such as electron transport, ATP synthesis, gene expression, and protein maturation/import. In Reclinomonas mitochondria, genes for the translational machinery comprise the largest single category, with 27 ribosomal protein genes [1].

In streptophytes (vascular plants, bryophytes, and charophycean green algae), the mitochondrial genome typically contains about 30 to 40 protein-coding genes of identified function. Approximately 20 of these genes are universally present, whereas the others (or a subset thereof) have been lost from various plant groups [3]. Genes encoding ribosomal proteins and subunits of the succinate dehydrogenase complex are most commonly absent [3], although loss or pseudogenization of other genes, such as cox2 [4, 5], nad7 [6, 7], atp8 [7], and cytochrome c biogenesis subunits [7, 8] has occurred as well. Typically, a gene is deleted from the plant mitochondrial genome only after successful transfer of a copy to the nucleus, although examples exist where loss is correlated with functional replacement of a "native" mitochondrial ribosomal protein by a nucleus-encoded plastid or cytosolic homolog [9, 10]. The timing of migration of mitochondrial ribosomal protein genes to the nucleus during eukaryotic evolution can be followed by comparative analysis [11, 12].

The mitochondrial genomes of seed plants are particularly large and recombinogenic. They contain many potential unknown open reading frames (ORFs) which have often been annotated as such in genomic sequencing projects when longer than 100 codons. However, most of these ORFs are not broadly conserved, which has brought into question their potential functionality. Moreover, it is not uncommon for plant mitochondrial DNA rearrangements to give rise to novel chimeric ORFs in specific lineages, and in certain instances such ORFs are correlated with mitochondrial dysfunction in the form of cytoplasmic male sterility [13]. On the other hand, a few ORFs have shown conservation among plants, and over recent years these have been upgraded to known mitochondrial genes. This list includes atp4 [14, 15], atp8 [15–17] and mttB (or tatC) [18, 19], which previously were denoted as orf25, orfB, and orfX, respectively. Within the three complete non-vascular plant mitochondrial genomes, there is another unidentified conserved ORF, named orf-bryo1 in the hornwort Megaceros aenigmaticus [7], orf187 in the moss Physcomitrella patens [20], and orf168 in the liverwort Marchantia polymorpha [21], suggesting that it may in fact code for a functional mitochondrial product in plants.

Results and Discussion

Mitochondrial orf-bryo1is conserved across streptophytes

To determine whether this bryophyte mitochondrial ORF might be more widespread among plants, blastp searches were performed using these three protein sequences to query the NCBI protein database. A homolog was found in the completely sequenced mitochondrial genomes of the angiosperms Nicotiana tabacum (orf159b) [22] and Vitis vinifera (orf159) [23] and, albeit with low sequence similarity, in the charophytes Chaetosphaeridium globosum (orf126) [8] and Chlorokybus atmophyticus (orf295) [24]. An unnamed predicted protein from cDNA analysis (XP_002332837) was also identified from Populus trichocarpa. Interestingly, the moss orf187 shows weak similarity to ribosomal protein L10 from several bacteria, including Rickettsia prowazekii and other members of the alpha-proteobacteria, the lineage from which mitochondria originated [25], as well as to mitochondrial L10 from the jakobid Reclinomonas americana, a protist that possesses the most "primitive" and gene-rich of all mitochondrial genomes [1]. These observations suggested that the moss orf187 (and its homologs) might encode mitochondrial L10 in plants. Indeed, annotated L10 domains can be found in the GenPept records for Physcomitrella orf187 (BAE93086) and Chlorokybus orf295 (ABO15139).

A variety of computational and experimental approaches were used to determine the distribution of mitochondrial rpl10-like sequences among streptophytes, and the results are summarized in Figure 1. To extend the database search, tblastn queries were conducted against the nucleotide nr and EST-others databases at GenBank. Indeed, homologous unannotated ORFs are present within the complete mitochondrial genomes of the charophyte Chara vulgaris, [26], the gymnosperm Cycas taitungensis [27], and the angiosperm Carica papaya (EU431224) as well as in partial mitochondrial genome entries for the angiosperms Solanum lycopersicum and Helianthus annuus. In addition, several truncated and/or frameshifted sequences were identified in the mitochondrial genomes of Brassica napus, Oryza sativa, and Bambusa oldhamii, suggestive of recent erosion of the rpl10-like gene. Searches of the EST-others database also revealed numerous homologs from a wide range of angiosperms as well as two gymnosperms, Picea glauca and Welwitschia mirabilis. Their high nucleotide similarity to counterparts identified in completely sequenced mitochondrial genomes of other seed plants suggests that these are in fact encoded in the mitochondrial genome, unless there has been extremely recent gene transfer to the nucleus. One exception is a divergent rpl10-like sequence from the fern Adiantum capillus-veneris (DK949045) that has an amino-terminal extension of 25 residues with a weak predicted mitochondrial targeting signal, and might therefore be nuclear-located.

Figure 1
figure 1

Distribution of mitochondrial rpl10 -like sequences in streptophytes. Functional genes, pseudogenes, and genes lost from the mitochondrion are shown as filled squares, open squares, and open circles, respectively. Genes with evidence for expression as determined by RNA editing status are marked with a plus symbol. The 'nuc?' note next to the Adiantum sequence indicates that it may be encoded in the nucleus. The origin of each sequence is given in parentheses using the following abbreviations: E - EST sequence from GenBank; G - genome sequence from GenBank; N - nucleotide sequence from GenBank; P - PCR product generated during this study; R - RT-PCR product generated during this study. Phylogenetic relationships are taken from the Angiosperm Phylogeny Website [55].

Full size image

To determine how widely this mitochondrial rpl10-like gene is represented in seed plants and to gain more insight into the prevalence and timing of apparent pseudogenization in certain lineages, a PCR survey was undertaken using primers designed from the angiosperm and gymnosperm sequences identified above. Sequencing revealed the presence of this gene in another 24 seed plants, of which 5 were pseudogenes (Figure 1). Overall, these results show that homologs to the orf-bryo1 gene can be found across virtually all major streptophyte lineages, although it should be noted that lycophytes are not represented in this data set and no homologous sequences were detected in the mitochondrial data recently presented for Isoetes engelmannii [28]. Notably, the rpl10-like gene appears to have been independently lost at least five times during angiosperm history: from the asterid Pentas, from the caryophyllid Beta, from the crucifers Arabidopsis and Brassica, from monocots, and from the conifer Podocarpus.

Angiosperm orf-bryo1homologs are transcribed, edited and likely encode a functional mitochondrial L10

At the DNA level, the mitochondrial rpl10-like gene appears to be functional in a very wide range of streptophytes, and the derived amino acid sequence alignments for selected species are shown in Figure 2. Amino acid conservation is higher in the amino-terminal region than at the carboxy-terminus, and the latter also shows variation in length, in keeping with features also common to L10 proteins in non-plants (see below). The initiation codons for Cycas and Megaceros are predicted to be generated by C-to-U RNA editing of ACG to AUG. Within the Megaceros coding sequence, three potential stop codons are presumably removed by U-to-C RNA editing prior to translation, as previously postulated for many Megaceros mitochondrial transcripts including orf-bryo1 [7]. To assess whether the coding sequences are under functional constraint, the ratio (ω) of non-synonymous (dN) to synonymous (dS) divergence was calculated for all pairwise sequence comparisons between 6 representative streptophytes (Table 1). In all 15 cases, ω was less than 1 consistent with purifying selection acting to maintain the protein sequences. The average over all tests was 0.39 with a high of 0.62 between Marchantia and Cycas and a low of 0.19 between Chara and Cycas.

Figure 2
figure 2

Alignment of L10 ribosomal proteins from plant mitochondria and eubacteria. Amino acids within a column are shaded if at least 75% are identical (black) or similar (gray). Columns in which RNA editing was observed in one or more sequences are marked with a red asterisk, and those positions are shaded in red. In the sequences translated from DNA, positions shown in lowercase and shaded in yellow were inferred to result from RNA editing by comparison to sequences from Physcomitrella, Marchantia, and Chara and from angiosperms with known editing data.

Full size image
Table 1 Pairwise ω (dN/dS) for plant rpl10 sequences
Full size table

We have also established that the mitochondrial rpl10-like gene is expressed and edited in angiosperms (Table 2). The cDNA sequences obtained from four angiosperm species (Aristolochia, Artemisia, Breynia and Ceropegia) all showed C-to-U RNA editing at between 5 and 8 sites, which verifies that they were derived from RNA template rather than contaminating mitochondrial DNA. In addition, 7 edit sites were identified for Citrus by comparison of its EST and gene sequences. Editing in all 5 plants predominantly alters the encoded amino acids, with each coding sequence having only one silent editing event. Furthermore, these non-synonymous editing events improve protein similarity of the angiosperm sequences to one another and to species that are known to have infrequent editing, such as Physcomitrella [29], or no editing, as for Marchantia [21] and Chara [26]. This pattern of editing is characteristic for functional plant mitochondrial genes but not necessarily for pseudogenes [30], and most unconserved ORFs are not edited at all [31–33]. The rpl10-like EST sequences provided further evidence of transcription, although in the absence of accompanying DNA sequence information the evidence is less certain. The EST sequences from Petunia, Theobroma, and Zinnia generally have T at confirmed edit positions and therefore likely derive from genuine RNA rather than mitochondrial DNA contamination, although it cannot be excluded that some of these T residues are already encoded in the genome. In contrast, homologs based on EST data from additional plants lack several expected edits and reflect either mitochondrial DNA contamination or partially edited transcripts (data not shown). In total, rpl10-like sequences from 8 distantly-related angiosperms provide strong evidence for appropriate expression at the RNA level (Figure 1).

Table 2 The effect of RNA editing on amino acid sequence
Full size table

In Figure 2, the amino acid alignment of plant and charophycean green algal mitochondrial orf-bryo1 homologs also includes the Reclinomonas americana mitochondrion-encoded L10 protein and homologs from the eubacteria Escherichia coli, Rickettsia prowazekii, and Thermotoga maritima. The L10 ribosomal protein is universally present in the ribosomes of eubacteria, archaea, and eukaryotes, and the crystal structure of L10-L7/L12 stalk has been determined [34]. It is worth noting that the amino-terminal domain is more highly conserved than the carboxy-terminal half. For example, the Rickettsia prowazekii and E. coli L10 proteins share only about 26% amino acid identity over their full length, whereas the beta-1 to alpha-5 region (of 85 amino acids) within the amino-terminal half shows ~35% identity. It is the amino-terminal domain of L10 (or more specifically, the alpha-1 to alpha-3 region) that binds directly to the large subunit ribosomal RNA, whereas the carboxy-terminal domain of L10 (and alpha-8 in particular) interacts with the L7/L12 stalk; together with L11, this complex plays a key role in recruiting translation factors to the ribosome and stimulating GTP hydrolysis [34, 35]. The flowering plant mitochondrial L10 proteins share about 23% amino acid identity with the Rickettsia L10 homolog over the amino-terminal beta-1 to alpha-5 region of 85 amino acids, compared to 27-28% identity seen between Rickettsia L10 and the comparable region of the Physcomitrella or Reclinomonas mitochondrial counterparts. Of particular note are several highly conserved blocks that are believed to be important for protein structure [34]. They contain Gly (and Pro) residues for beta-turns between beta1-alpha2 and alpha4-beta3 in L10 proteins of eubacteria and archaea. Interestingly, 7 of 8 positions of RNA editing lie within conserved blocks, consistent with their functional importance, a hallmark of RNA editing in plant mitochondria [36].

In bacterial, archaeal and eukaryotic cytosolic ribosomes, the amino terminal domain of the L10 protein is known to bind specifically to helices H42, H43, and H44 of the large subunit rRNA [34, 35], and in plant mitochondria, this helical region of the 26S rRNA has retained the correct structure for L10 binding and is very highly conserved among streptophytes (Figure 3). Indeed this stretch of 80 nt is identical in sequence among most seed plants and there has been only one nucleotide substitution relative to either the Physcomitrella or Marchantia homolog, that is, during a period of about 400 million years. Thus, it seems likely that a conventional L10 protein (or at least for the amino-terminal portion) will be present in plant mitochondrial ribosomes.

Figure 3
figure 3

Sequence and structure of the LSU rRNA region that binds to L10 ribosomal protein. Shown are helices H42, H43 and H44 of the LSU rRNA. The primary sequence shown is a consensus of this mitochondrial 26S rRNA region from Marchantia, Physcomitrella, and numerous seed plants, with differences shown in red. Positions that differ between plant mitochondria and bacteria (represented by E. coli) are shaded in gray. Yellow shading indicates compensatory changes in E. coli that maintain base pairing in stem regions. Nucleotide coordinates are shown for Triticum aestivum mitochondrial 26s rRNA [56] and in parentheses for E. coli 23s rRNA [34, 35].

Full size image

Status of mitochondrial L10 in grasses and crucifers

For the reasons discussed above, one might expect that all seed plants would possess a mitochondrial-type rpl10 gene either within the mitochondrion or alternatively within the nucleus since the simplest explanation for cases of gene loss from the mitochondrion (see Figure 1) is that successful gene transfer to the nucleus has occurred. Curiously, no mitochondrial-type L10 protein sequences were detected in tblastn searches of the completely sequenced nuclear genomes of Arabidopsis [37] or rice [38, 39]. However, both these genomes do contain duplicated copies of the chloroplast-derived rpl10 gene (data not shown). In land plants, the chloroplast rpl10 gene is located in the nucleus, and proteomic analysis of spinach chloroplast ribosomes has established its precise protein content [40]. The chloroplast L10 orthologs in Arabidopsis (NP_196855) and rice (NP_001049761) share about 70% amino acid identity (excluding the acquired N-terminal targeting extensions). In contrast, the second chloroplast-type L10-related copy shows only ~41% amino acid identity between the Arabidopsis (NP_187843) and rice (NP_001054498) counterparts, and these proteins are predicted to be localized in the mitochondrion based on targeting programs such as TargetP [41], PSort [42], and Predotar [43] Interestingly, the two Arabidopsis chloroplast-derived L10 paralogs are more closely related to each other (~58% identity) than are two rice ones (~46% identity), suggesting a more recent duplication event in the crucifer lineage. This would also be consistent with their independent recruitment as functional substitutes for the mitochondrial L10 protein at different times during angiosperm evolution, although it cannot be formally excluded that gene conversion events in the Arabidopsis lineage contribute to the higher sequence similarity.

Although the duplicated chloroplast-type L10-related gene is an attractive candidate to serve as a replacement in the mitochondrial ribosome for those plants which lack the "native" mitochondrial rpl10 gene, these proteins in Arabidopsis and rice lack a number of the expected conserved residues, ones that are observed in the plant mitochondrion-encoded genes. Alternative possibilities are that the chloroplast L10 might be dual targeted to both the plastid and the mitochondrion or that the cytosolic ribosomal protein L10 counterpart (called L10e or P0) has been recruited. It is perhaps even possible that plants such as rice and Brassica, which possess what appear to be remnant pseudogene fragments in the mitochondrion, actually have several short genes (mitochondrial or nuclear) that generate a discontinuous L10 protein structure, a phenomenon observed for the mitochondrial rpl2 gene in certain flowering plants [44]. Finally, it is worth noting that non-homologous proteins have been known to perform molecular mimicry in the evolution of the large ribosomal subunit among eubacteria and archaea [45].

Conclusion

In summary, these observations provide strong evidence that a functional rpl10 gene exists in the mitochondrion of many streptophytes. Despite the fact that there are now over 20 streptophytes with complete mitochondrial genome sequences, this gene has been missed until now due to the unlikely coincidence that most of the plant mitochondrial genomes that were first completely sequenced - the crucifers Arabidopsis thaliana [46] and Brassica napus [33]; the grasses Oryza sativa [32], Zea mays [47] and Triticum aestivum [48]; and the sugar beet Beta vulgaris [49] - are from lineages where this gene has been lost or pseudogenized. Only with the more recent sequence data from diverse streptophytes such as Cycas taitungensis [27], Physcomitrella patens [20] and Megaceros aenigmaticus [7] does the general pattern emerge that this gene is in fact widely present. Indeed, the bryophyte orf-bryo1 sequences were particularly informative in bridging the evolutionary distance between mitochondrial L10 gene homologs in seed plants and those of charophycean green algae/protists, which nicely illustrates the power of obtaining sequence information from diverse organisms in order to reconstruct events related to gene and genome history.

Methods

Total genomic DNAs and RNAs were isolated using the DNeasy and RNeasy Plant Mini Kits (QIAGEN) from leaf tissue available in the living collection of the Beadle Center Greenhouse (University of Nebraska). To prepare first-strand cDNA, RNAs were treated with DNase I (Fermentas) to remove contaminating DNA and then reverse transcribed using M-MuLV Reverse Transcriptase (Fermentas) and random hexamers (Fermentas) according to the manufacturer's instructions.

Sequences for rpl10 were amplified from DNA or cDNA by polymerase chain reaction using GoTaq DNA Polymerase (Promega) and forward primer F1 (5'-ATGCCATTCGGAAGAAGTMT) with reverse primer R159 (5'-TTAGGTGGTATYCCGAGATYGA) or R148 (5'-GGAACACACGAAASAAAGATATRAAC). Each reaction was run on a DNA Engine Dyad (Bio-Rad) for 35 cycles (30 sec at 94°C, 1 min at 48°C, 2 min at 72°C), with an initial step of 3 min at 94°C and a final step of 10 min at 72°C. Amplified products were sequenced on both strands at the High-Throughput Genomics Unit (University of Washington). Sequences generated in this study were deposited in GenBank under accession numbers GQ402491-GQ402514; additional sequences used in the comparative analysis were downloaded from GenBank (Table 3).

Table 3 Taxonomy and GenBank accession numbers for rpl10 sequences in this study
Full size table

Sequences were aligned using Muscle 3.7 [50] and manually adjusted in BioEdit 7.0.9 [51]. Edit sites were identified by comparison of DNA sequences with cDNA and/or EST sequences. To examine levels of functional constraint, poorly-aligned regions were first identified and removed using Gblocks 0.91b [52], then pairwise dN and dS were computed in MEGA 4.0.2 [53] using the Nei-Gojobori Model with a Jukes-Cantor correction for multiple hits.

Note added in proof

Another group has independently discovered the rpl10 gene in the mitochondrial genome of plants [54]. Similar to our study, Kubo and Arimura find that the mitochondrial gene is widely distributed among plants, is transcribed and RNA edited in multiple species, and has been lost from several lineages, including Arabidopsis and rice. These authors suggest, as we do, that a duplicated copy of the nucleus-encoded chloroplast rpl10 gene has functionally replaced the lost mitochondrial rpl10 gene independently in Arabidopsis and rice. For both species, they experimentally show that these putative mitochondrially-functioning L10 proteins have targeting signals that indeed induce localization to the mitochondrion, and also the chloroplast.